Comparative evaluation of biogenic hydroxyapatite derived from fish bones and synthetic hydroxyapatite for the removal of flucloxacillin sodium Authors: Ali Boukra, Omar Boukra, Noureddine Akartasse, Souhayla Latifi, Harunor Rashid, Shehdeh Jodeh, Belkheir Hammouti, Hatem A. Abuelizz, Miloud El Karbane, Khalil Azzaoui, Sanaâ Saoiabi Published: 2026 ABSTRACT In this work, a comparative study of the adsorption of flucloxacillin sodium (FLCX-Na) from aqueous solution onto synthetic hydroxyapatite (Syn-CaHAp) and biogenic hydroxyapatite from fish bones (B-CaHAp) is reported. The adsorbents were prepared and characterized by X-ray diffraction (XRD), Fourier transform infrared spectroscopy (FTIR), thermogravimetric/differential thermal analysis (TGA/DTA), scanning electron microscopy coupled with energy-dispersive (SEM–EDS) and Transmission electron microscopy (TEM). The characterization results revealed that B-CaHAp possesses calcium-deficient carbonate-substituted apatite structure with heterogeneous morphology and higher density of accessible adsorption sites. Batch adsorption experiments indicated that the adsorbent dosage, contact time, initial concentration, and pH of the solution had a significant effect on the adsorption performance. The maximum experimental adsorption capacities of Syn-CaHAp and B-CaHAp were 15 and 26 mg·g− 1 with removal efficiencies of 62 and 77%, respectively, indicating the excellent adsorption performance of the biogenic material. The kinetic analysis showed that the adsorption process followed the pseudo-second-order model for both adsorbents. The equilibrium data were better fitted to the nonlinear Langmuir isotherm indicating that the monolayer adsorption was predominant under the studied conditions. The combined experimental and theoretical analyses imply that the adsorption of FLCX-Na occurs via specific interactions involving electrostatic attraction, hydrogen bonding and coordination between the functional groups of FLCX-Na and the calcium active sites of hydroxyapatite. Response surface methodology (RSM) indicated that the optimum conditions were approximately 0.10 g of adsorbent dosage, 25 mg·L− 1 of initial concentration, pH 6.5 and 160 min of contact time, yielding 83% maximum removal efficiency and approximately 20.8 mg·g− 1 of adsorption capacity. Additionally, B-CaHAp displayed high adsorption efficiency with rising NaCl concentrations, indicating its applicability in saline aqueous environments and its good tolerance to ionic strength. Furthermore, the regeneration experiments demonstrated that the adsorbent could retain ~ 92% of the initial adsorption capacity after five adsorption-desorption cycles with a slight decrease in removal efficiency from 83.0 to 76.1%, showing the excellent operational stability and reusability of the adsorbent. Finally, these results showed the great potential of B-CaHAp as an efficient, sustainable and reusable adsorbent for the removal of pharmaceutical contaminants from aqueous media. Introduction Water resources pollution is one of the most alarming environmental problems of the XXI century. The growth of the population, the rapid urbanization, the industrial expansion and the large scale consumption of medicines lead to a significant increase of micro-pollutants in urban and hospital waste [1, 2]. Unlike other drugs, antibiotics have a specific target: bacteria, especially pathogens, and do not damage human cells or tissues [3]. They can be classified according to their chemical structure or mechanism of action. Around 257 different antibiotics are clinically used and monitored globally for their impact on public health and microbial resistance, as listed in the World Health Organization AWaRe classification system [4, 5]. Among them, FLCX-Na is of increasing interest as an isoxazolyl penicillin commonly used for the treatment of infections caused by penicillinase-producing Staphylococcus aureus [6]. Its unmanaged release into the environment raises concerns: it has high water solubility, amphoteric properties that vary with pH (pKa₁ ≈ 2.76) [7], high stability in neutral environments that prolongs its persistence, and a substantial risk of spreading bacterial resistance. In response to this issue, various techniques for purifying polluted water have been studied, ranging from physicochemical processes (such as advanced oxidation, coagulation-flocculation, ion exchange, adsorption, membrane and electrochemical methods) to biological methods (such as bioremediation, phytoremediation, and bioadsorbents) [8, 9]. However, many of these techniques have limitations in terms of cost, selectivity, or the production of by-products. Adsorption is therefore seen as a preferred method for water purification due to its simplicity, efficiency, and potential for large-scale application [10]. In this sense, several adsorbent materials have been studied such as activated carbon, zeolites, processed clays and treated biomass. But calcium phosphates, especially hydroxyapatite (Ca10(PO4)6(OH)2), are of increasing interest. The latter is characterized by high biocompatibility and non-toxicity, chemical stability, neutrality and the possibility of ion substitution (carbonate, magnesium, sodium, strontium, etc.) which can significantly affect its porosity, specific surface area, surface charge and the ability to interact with various pollutants through adsorption, ion exchange and complexation processes [11, 12]. Among them, sorption methods are highly efficient and economical for removing antibiotics, making them the method of choice in many applications [13, 14]. Their simplicity of implementation, their effectiveness even at low concentrations, and the possibility of using natural or bio-based adsorbents explain the choice of adsorption as the main approach in this study [15–17]. In addition, a large number of studies have examined the potential benefits of CaHAp in the field of water treatment, mainly with regard to the removal of heavy metals (Pb²⁺, Cd²⁺, Zn²⁺), organic dyes, antibiotics, and other pharmaceutical substances [18]. Recent studies have focused on producing hydroxyapatite nanoparticles (CaHAp) that combines superior quality and biocompatibility with cost-effective and environmentally friendly methods [19, 20]. Double decomposition [21, 22] and the use of biogenic sources [19, 23], are among the most commonly used techniques, accounting for nearly 30% of research published since 1999 [20]. The success is attributed to their large advantages such as low cost, minimization of by-products, easy control of particle size and accessibility [20, 23]. Biogenic forms of CaHAp nanoparticle derived from animal or fish waste have been of particular interest due to their low cost, sustainability and added value to residues of the agri-food industry. CaHAp nanoparticle was synthesized from several types of fish including salmon, carp, Japanese anchovy, sardine, tilefish and tuna. These biogenic forms often contain trace elements (Na+, Mg2+, Sr2+, etc.) that are naturally incorporated into the apatite network. Such ion substitutions might affect the surface chemistry and functional properties of the material, especially for adsorption-related applications. According to the literature, the crystallinity of biogenic hydroxyapatites can fluctuate depending on the biological origin and heat treatment conditions [10, 24, 25]. Similarly, synthetic hydroxyapatite nanoparticles allow for precise control of chemical composition, crystallinity, and phase purity through rigorously controlled synthesis methods. However, high production costs and the lack of natural ionic substitutes may pose obstacles to its widespread use in certain environmental fields [26, 27]. Comparative research between synthetic CaHAp and biogenic CaHAp remains limited, especially in terms of its effectiveness in adsorbing antibiotics such as flucloxacillin sodium. It is crucial to understand the structural, morphological, and compositional differences between these two substances in order to optimize their application in the decontamination process [18]. Several mechanisms are responsible for the adsorption of the hydrophilic β-lactam antibiotic on CaHAp: pH-related electrostatic interactions;Creation of hydrogen bonds with the amide and carbonyl groups of FLCX-Na;Potential complexation between Ca²⁺ ions on the surface and carbonyl/β-lactam groups;Van der Waals interactions and surface adsorption. pH-related electrostatic interactions; pH-related electrostatic interactions; Creation of hydrogen bonds with the amide and carbonyl groups of FLCX-Na; Creation of hydrogen bonds with the amide and carbonyl groups of FLCX-Na; Potential complexation between Ca²⁺ ions on the surface and carbonyl/β-lactam groups; Potential complexation between Ca²⁺ ions on the surface and carbonyl/β-lactam groups; Van der Waals interactions and surface adsorption. Van der Waals interactions and surface adsorption. In addition to traditional parametric studies, statistical optimization strategies offer effective ways to investigate the combined effects of operational parameters and identify optimal conditions for adsorption processes. A mathematical model was developed for this purpose using an experimental design methodology based on a centered composite design, implemented using JMP Pro software [28, 29]. With these approaches, response surface methodology (RSM) enables the development of reliable predictive models with a limited number of experiments. It is particularly well-suited for complex systems where multiple parameters simultaneously influence adsorption performance. Thus, given the superior performance observed for biogenic hydroxyapatite (B-CaHAp), this approach was applied to this material to identify optimal operating conditions and to better understand the interactions between process variables [30, 31]. Beyond adsorption efficiency and process optimization, the practical applicability of an adsorbent depends on its stability during repeated use and its ability to maintain high performance in the presence of dissolved inorganic salts, which are frequently found in actual wastewater. It also depends on its operational stability and its ability to be regenerated after multiple uses. The regeneration and reuse of adsorbents have become an important criterion for evaluating their long-term applicability in wastewater treatment, from both an environmental and economic perspective. Thus, regeneration studies provide valuable information regarding the stability of the adsorption mechanism and the possibility of reusing the material during successive operating cycles. In this study, our objective is to: (i)Develop and characterize two distinct hydroxyapatites nanoparticles, Syn-CaHAp and B-CaHAp;(ii)Analyse their physicochemical characteristics using complementary techniques (X-ray diffraction (XRD), Fourier transform infrared spectroscopy (FTIR), Thermogravimetric analysis and differential thermal analysis (TGA/DTA), scanning electron microscopy coupled with energy-dispersive X-ray spectroscopy (SEM/EDS) and Transmission electron microscopy (TEM);(iii)Assess their adsorption capacity for FLCX-Na under conditions that mimic polluted effluents to establish the optimal parameters for their use in the depollution process.(iv)Apply statistical optimization using response surface methodology (RSM) to determine optimal operating conditions and analyze the interactions between the main process variables. The practical applicability of the optimized B-CaHAp was evaluated by studying the effect of ionic strength on its adsorption performance and its regeneration–reusability in successive adsorption–desorption cycles. (i)Develop and characterize two distinct hydroxyapatites nanoparticles, Syn-CaHAp and B-CaHAp; Develop and characterize two distinct hydroxyapatites nanoparticles, Syn-CaHAp and B-CaHAp; (ii)Analyse their physicochemical characteristics using complementary techniques (X-ray diffraction (XRD), Fourier transform infrared spectroscopy (FTIR), Thermogravimetric analysis and differential thermal analysis (TGA/DTA), scanning electron microscopy coupled with energy-dispersive X-ray spectroscopy (SEM/EDS) and Transmission electron microscopy (TEM); Analyse their physicochemical characteristics using complementary techniques (X-ray diffraction (XRD), Fourier transform infrared spectroscopy (FTIR), Thermogravimetric analysis and differential thermal analysis (TGA/DTA), scanning electron microscopy coupled with energy-dispersive X-ray spectroscopy (SEM/EDS) and Transmission electron microscopy (TEM); (iii)Assess their adsorption capacity for FLCX-Na under conditions that mimic polluted effluents to establish the optimal parameters for their use in the depollution process. Assess their adsorption capacity for FLCX-Na under conditions that mimic polluted effluents to establish the optimal parameters for their use in the depollution process. (iv)Apply statistical optimization using response surface methodology (RSM) to determine optimal operating conditions and analyze the interactions between the main process variables. The practical applicability of the optimized B-CaHAp was evaluated by studying the effect of ionic strength on its adsorption performance and its regeneration–reusability in successive adsorption–desorption cycles. Apply statistical optimization using response surface methodology (RSM) to determine optimal operating conditions and analyze the interactions between the main process variables. The practical applicability of the optimized B-CaHAp was evaluated by studying the effect of ionic strength on its adsorption performance and its regeneration–reusability in successive adsorption–desorption cycles. Although hydroxyapatite has been the subject of extensive research as an adsorbent for various water pollutants, studies directly comparing synthetic and biogenic hydroxyapatites for the removal of flucloxacillin sodium (FLCX-Na) remain rare. In this context, this study is the first to propose a systematic comparison between hydroxyapatites derived from fish bones and chemically synthesized hydroxyapatites for the adsorption of FLCX-Na under the same experimental conditions. In addition to evaluating their adsorption performance, this study is distinguished by its in-depth characterization of the materials, equilibrium and adsorption kinetic analyses, and optimization using response surface methodology (RSM) to determine the relationship between their physicochemical properties and their adsorption behavior. It also includes an adsorption efficiency test of the material in environments containing increasing concentrations of NaCl to evaluate its adsorption for this antibiotic in almost realistic scenarios. The study further includes an assessment of the regeneration and reusability of the optimized biogenic hydroxyapatite. This allows for a more comprehensive evaluation of its potential for the sustainable removal of antibiotics. Overall, this research demonstrates the potential of fishbone-derived hydroxyapatite as a sustainable, efficient, and reusable adsorbent for antibiotic-contaminated waters and explores the relationship between its physicochemical properties, adsorption behavior, optimization, and operational stability through the use of complementary characterization techniques and regeneration studies. Experimental Materials used Syn-CaHAp was developed using a double decomposition method. We used ammonium hydrogen phosphate (NH₄)₂HPO₄ (99%), calcium nitrate tetrahydrate (Ca(NO₃)₂·4 H₂O), and ammonium hydroxide (NH₄OH), all supplied by Sigma-Aldrich and with an analytical purity greater than 99%. All syntheses were performed using distilled water [32]. In addition, B-CaHAp was prepared from sardine bone waste obtained from the fish market of Place Yaakoub L’Mensour (Rabat, Morocco). The fish bone residues were obtained as waste products from the commercial processing of fish for human consumption. No live fish were purchased, handled, or euthanized for this study. Distilled water and acetone (96%) were used for the cleaning, degreasing, and pretreatment phases, without the use of other chemicals. Preparation of CaHAp Preparation of Syn-CaHAp nanoparticles A double decomposition procedure (wet precipitation) was used to create Syn-CaHAp. A stoichiometric Ca/P ratio of 1.67 was maintained by using calcium nitrate tetrahydrate (Ca(NO₃)₂·4 H₂O, 10.28 g) and ammonium hydrogen phosphate (NH₄)₂HPO₄, 3.45 g) as sources of calcium and phosphorus, respectively. The calcium solution was added gradually to the phosphate solution while stirring constantly with a tricolumn. Although calcium phosphate precipitation can occur within a pH range of approximately 4.5 to 6, these conditions generally favor the formation of less crystalline intermediate calcium phosphate phases. To stabilize and maintain an alkaline environment, a solution of ammonium hydroxide (NH₄OH, 25%) was added gradually until the pH reached 10. NH₄OH was used in this synthesis to regulate the pH required for the nucleation and crystal growth of hydroxyapatite, as well as to provide hydroxide ions (OH⁻), which are essential for the formation of hydroxyapatite and promote the deprotonation of phosphates. NH₄OH is not present in the final hydroxyapatite, although OH⁻ ions participate in the precipitation reaction and are incorporated into the apatite structure [24]. Under alkaline conditions, hydroxyapatite precipitates, and any calcium hydroxide (Ca (OH)₂) that may temporarily form during pH adjustment is rapidly consumed by reaction with phosphate species. A soluble byproduct, ammonium nitrate (NH₄NO₃), is formed and remains dissolved in the reaction medium. The reaction mixture was magnetically stirred at 85 °C. The instantaneous appearance of a milky-white deposit indicated the formation of hydroxyapatite nanoparticles. The chemical reaction can be expressed as follows [33]: 10 Ca(NO₃)₂·4 H₂O + 6 (NH₄)₂HPO₄ + 8 NH₄OH → Ca₁₀(PO₄)₆(OH)₂ + 20 NH₄NO₃ + 6 H₂O. The precipitate obtained was filtered, then carefully washed with distilled water to remove traces of (NH4NO3), before being dried at 80 °C for one day in an oven. Once cooled, the dried product was ground into powder to ensure a uniform particle size (Fig. 1) [21, 34]. Fig. 1Schematic representation of CaHAp synthesis using the double decomposition method [32] Schematic representation of CaHAp synthesis using the double decomposition method [32] Preparation of B-CaHAp B-CaHAp was synthesized from sardine bone waste sourced from the fish market at Place Yaakoub L’Mensour (Rabat, Morocco), which served as the main ingredient. The bones were carefully cleaned to remove soft tissue, fat, and other organic impurities. They were then boiled in distilled water for three hours to remove all organic matter. After boiling, the samples were rinsed under running water until no visible traces remained. The cleaned bones were then dried in the sun for three days, followed by placement in an oven at 80 °C for 24 h to remove any residual moisture. The dried material was ground into a fine powder and then degreased with 96% acetone at a solid-to-liquid ratio of 1:50 (w/v). For two hours, the mixture was stirred at 60 °C until all remaining hydrophobic lipids were removed. Once the treatment was complete, the material was filtered and thoroughly washed with distilled water. The pretreated material was then placed in a muffle furnace and heated to 900 °C over 5 h, at a rate of approximately 10 °C·min⁻¹ to avoid thermal shock. In order to guarantee the total elimination of any remaining organic matter (collagen, lipids, and proteins) and to facilitate the conversion of the biogenic mineral into a highly crystalline and phase-pure hydroxyapatite, the calcination conditions (900 °C for 5 h) were chosen. According to earlier research on hydroxyapatite generated from fish bones, this heat treatment is necessary to obtain a stable adsorbent with repeatable physicochemical properties appropriate for adsorption applications, despite the relatively large energy input required [35]. Once cooled to room temperature, the calcined product was ground into powder to achieve a uniform particles consistency, allowing for in-depth characterization and adsorption testing (Fig. 2) [36, 37]. Fig. 2Illustrative diagram of the protocol for extracting and preparing hydroxyapatite nanoparticles from fish bones Illustrative diagram of the protocol for extracting and preparing hydroxyapatite nanoparticles from fish bones Technical characterization XRD measurements The crystal structure of the composite materials was analyzed using the XRD method. The Panalytical X’Pert Pro device (Malvern Panalytical GmbH, Kassel, Germany) was used to capture the XRD data, using a Cu Kα radiation source at 40 kV and 30 mA, with a scanning speed of 2°/min. The average crystallite size of Syn-CaHAp and B-CaHAp was calculated using the Scherrer equation (Eq. 1) on the isolated (002) diffraction peak as follows:1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:D=\frac{K\:\lambda\:}{\beta\:\:cos\theta\:}\:\:$$\end{document} where D is the crystallite size, K is the shape factor (0.9), λ is the wavelength of Cu Kα radiation (1.5406 Å), β is the full width at half maximum (FWHM) in radians, and θ is the Bragg angle. FTIR measurements FTIR was used to study the evolution of CaHAp. An FT/IR-4700 spectrometer (JASCO International LTD., Tokyo, Japan) was used to obtain the spectra. Each spectrum contained an average of 32 scans at a resolution of 4 cm-1, with a spectral range from 400 to 4000 cm-1. A background check was performed before each analysis. In order to compare the chemical compositions of Syn-CaHAp and B-CaHAp, FTIR analysis was used to determine the hydroxyapatite samples’ distinctive functional groups and assess the presence of phosphate, hydroxyl, and carbonate groups. The discussion is focused on the spectral region with the characteristic absorption bands important for hydroxyapatite identification and comparison between the investigated samples. Thermogravimetric analysis Standard thermogravimetric analyzers, the TA Instruments TGA Q500 and Q50, offer a temperature range of 20 to 900 °C and a heating rate of 10 °C/min. The differential scanning calorimetry (DSC) curves were recorded on a DSC Q2000 V24.4 Build 116. Before being fed to the DSC cell, ten-milligram samples were placed inside aluminum DSC pans and sealed. Thermogravimetric analysis was used to determine the mass-loss events related to moisture removal, residual volatile species, and structural changes that occur during heating, as well as to assess the thermal stability of the synthesized materials. Morphological studies We used a Thermo Scientific™ Quattro ESEM (ThermoFisher Scientific, Paisley, UK) equipped with an energy-dispersive spectrometer (EDS), operating at an acceleration voltage of 15 kV, to obtain the SEM images. Before imaging, the films were coated with a thin layer of gold by cathodic sputtering to optimize their conductivity to the electron beam. The FEI Company transmission electron microscope (TEM) is a high-resolution field-emission instrument operating at an acceleration voltage of 200 kV, providing a spatial resolution of 0.12 nm in imaging mode and equipped with TEM, HR-TEM, HAADF-STEM, and HR-STEM modes, along with single-tilt and double-tilt specimen holders (± 35°/±30°), a 4k × 4k pixel Ceta™ 16 M camera, and an EDX Super-X detector with an energy resolution of 136 eV. The surface morphology and particle aggregation of the hydroxyapatite samples were studied by SEM and the elemental composition and Ca/P ratio were analyzed by EDS analysis. TEM observations were carried out to examine morphology, crystal size, and aggregation state at higher resolution. Point of zero charge (PZC) determination Using the pH derivation method, the point of zero charge (pHpzc) of Syn-CaHAp and B-CaHAp samples has been established. Using diluted solutions of HCl (0.1 M) or NaOH (0.1 M), solutions of NaCl (0,01 M, 50 mL) were adjusted to starting pH values between 2 and 11. To achieve the acid-base balance, a fixed amount of adsorbent (50 mg) was then added to each solution, and the suspensions were then kept under agitation for 24 h at room temperature. The final pH was measured. The point at which the ΔpH (pH final – pH initial) versus pH initial curve crosses the x-axis (ΔpH = 0) was known as the PZC [32]. Adsorption experiments FLCX-Na was chosen as the model pollutant (Fig. 3) to assess the effectiveness of synthetic and biogenic hydroxyapatites (CaHAp) as adsorbents. Fig. 3Chemical structure of flucloxacillin sodium Chemical structure of flucloxacillin sodium All adsorption experiments were carried out in batch mode using 100 mL of FLCX-Na solution in Erlenmeyer flasks under constant magnetic stirring at 25 ± 1 °C and 150 rpm. One of the operating parameters: adsorbent dosage (W), solution pH, contact time (t), and initial FLCX-Na concentration (C₀) was changed while the others remained constant to assess their impact. Adsorbent dosage of 0.20 g, initial FLCX-Na concentration of 20 mg·L⁻¹, contact time of 180 min, and solution pH of 6 were the reference experimental conditions, unless otherwise noted. Except for the parameter being studied, these reference conditions were used in all parametric, kinetic, and equilibrium adsorption investigations. Either diluted hydrochloric acid (HCl, 0.1 M) or sodium hydroxide (NaOH, 0.1 M) solutions were added to the solutions to change their pH. The residual concentration of FLCX-Na was determined by a UV-Vis spectrophotometer. The suspension was filtered through filter paper at the end of each adsorption experiment to completely separate the adsorbent from the solution. The absorbance of the clear filtrate was determined at the maximum absorption wavelength of FLCX-Na (λmax = 274 nm). Before the adsorption experiments, a calibration curve was obtained for standard FLCX-Na solutions of known concentrations. Absorbance values obtained for each filtrate were converted into the respective residual concentration (Ce) using the calibration curve [32]. The obtained values of Ce were then used to calculate the adsorption capacity (qe) and the removal efficiency (using Eqs. (2) and (3)) [38]:2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\%}\mathrm{R}\mathrm{e}\mathrm{m}\mathrm{o}\mathrm{v}\mathrm{a}\mathrm{l}=\frac{{C}_{0}-{C}_{e}}{{C}_{0}}\bullet\:100$$\end{document}3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{q}_{e}=\frac{{C}_{0}-{C}_{e}}{W}\:V$$\end{document} Where C0 represents the initial concentration of the pollutant (mg·L⁻¹), while Ce indicates the equilibrium concentration of this pollutant (mg·L⁻¹). qe indicates the equilibrium capacity per unit mass of adsorbent (mg.g⁻¹), W represents the mass of adsorbent in grams (g), and V denotes the volume of the solution in liters (L). For consistency, the adsorbent dosage effect was assessed as removal efficiency, since this parameter is most frequently used to find the optimum amount of adsorbent. The effects of pH, contact time and initial concentration were discussed as adsorption capacity (qₑ), which is more adequately reflecting the adsorption performance under different operating conditions. Kinetics of the adsorption process The adsorption kinetics of FLCX-Na on hydroxyapatite were studied using different common kinetic models, including pseudo-first-order, pseudo-second-order, and intra-particle diffusion models. The linearized versions of these equations are shown below [38]:4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\mathrm{ln}({q}_{e\:}-\:{q}_{t})=\mathrm{ln}{q}_{e}-{\it{K}}_{1}\it{t}$$\end{document}5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{\it{t}}{{\it{q}}_{\it{t}}}=\frac{1}{{\it{K}}_{2}{q}_{e}^{2}}+\frac{\it{t}}{{\it{q}}_{\it{e}}}$$\end{document}6\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Qt = K_{id}t^{1/2} + C$$\end{document}7\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\rm{ln}(1-F) = -\it{K}_{fd}^*\it{t}$$\end{document} Where qe symbolizes the equilibrium adsorption capacity expressed in mg.g⁻¹., while qt refers to the adsorption capacity at time t (min). The rate constant associated with the pseudo-first-order reaction is symbolized by K1(min⁻¹), while that of the pseudo-second-order reaction is symbolized by K2 (g.mg⁻¹.min⁻¹). To evaluate the three distinct phases, indicatingpreventing the rate, the intra-particle diffusion model was applied to the experimental data, Eq. (6) [39]. The thickness of the boundary layer can be defined using C (mg.g⁻¹), where Kid is the intraparticle diffusion rate constant, expressed in \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\rm{mg.g}}^{-1}. {\rm{min}}^{-1/2}$$\end{document}. In Eq. (7), Kfd represents the diffusion rate of the film in \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$({\rm{mg.g}}^{-1} .{\rm{min}}^{-1/2})$$\end{document}, while the equilibrium fraction obtained is symbolized by F (where F = qt/qe). Adsorption isotherms The adsorption of FLCX-Na on hydroxyapatite (CaHAp) was modeled using two classical models: the Langmuir model [40] and the Freundlich model [41]. The Langmuir model equation, which assumes monolayer adsorption on homogeneous sites, is formulated as follows (Eq. (8)):8\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{{C}_{e}}{{q}_{e}}=\frac{1}{{q}_{\rm{max}}}{C}_{e}+\frac{1}{{q}_{\rm{max}}{K}_{L}}$$\end{document} Ce This represents the equilibrium concentration of FLCX-Na expressed in parts per million (mg·L⁻¹), qe indicates the capacity removed per unit mass of CaHAp at equilibrium (in mg.g⁻¹), qmax symbolizes the maximum monolayer adsorption capacity of the adsorbent (indicated in mg.g⁻¹) and KL represents the Langmuir constant, expressed in liters per milligram (L.mg⁻¹). Equation (9) presents the dimensionless constant separation factor used by the Langmuir isotherm model to determine whether adsorption will be beneficial [38]:9\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{R}_{L\:}=\frac{1}{1+\:{K}_{L\:}{C}_{0}}$$\end{document} Where C0 represents the initial concentration of FLCX-Na and KL represents the Langmuir constants. The dimensionless factor RL was used to assess the propensity of the adsorption process. An RL greater than one indicates unfavorable adsorption, an RL equal to one indicates linear adsorption, while values between zero and one indicate favorable adsorption. Equations (10) and (11) illustrate the Freundlich model, which deals with adsorption on heterogeneous surfaces [38]:10\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\mathrm{ln}({q}_{e\:})=\mathrm{ln}{k}_{F}+\frac{1}{n}\mathrm{ln}{C}_{e}$$\end{document}11\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$q_{e}= K_{F} C_{e}^{1/n}$$\end{document} Where KF symbolizes the adsorption capacity, while 1/n illustrates the intensity of adsorption, values of 0 < 1/n < 1 indicate favorable adsorption, with lower values illustrating more intense interactions between the adsorbent and the adsorbate. To further study the adsorption mechanism, the equilibrium data were analyzed by the Dubinin–Radushkevich (D–R) isotherm model. The model was applied to estimate the theoretical adsorption capacity (qm), D-R constant (β), and the mean free adsorption energy (E) using Eqs. (12)–(14) [42]:12\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\mathrm{l}\mathrm{n}\:{q}_{e}\text{}=\mathrm{l}\mathrm{n}\:{\mathrm{q}}_{\mathrm{m}}\text{}-{\upbeta\:}{{\upepsilon\:}}^{2}$$\end{document}13\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\upepsilon\:}=\mathrm{R}\mathrm{T}\:\mathrm{l}\mathrm{n}(1+\frac{1}{{\it{C}}_{\mathrm{e}}}\text{})$$\end{document}14\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\it{E}=\frac{1}{\sqrt{2{\upbeta\:}\text{}}\text{}}$$\end{document} where qm (mg·g-1) is the theoretical adsorption capacity, β (mol2·kJ-2) is the D-R isotherm constant related to the adsorption energy, ε (kJ·mol-1) is the Polanyi potential, E (kJ·mol⁻¹) is the mean free adsorption energy, R is the universal gas constant (8.314 × 10− 3 kJ·mol-1·K-1), and T is the absolute temperature (298 K). Statistical optimization using response surface methodology (RSM) The optimization of the adsorption conditions for FLCX-Na was conducted using response surface methodology (RSM) based on a centered composite design (CCD) in JMP Pro software. Based on preliminary tests, four independent variables were selected [43]: adsorbent dose (X1), contact time (X2), initial concentration (X3), and solution pH (X4). Each factor was studied at five coded levels (axial and central points) to ensure a good estimate of the response within the experimental range (Table 1). Table 1Coded and actual levels of the independent variables used in the CCDFactorsCoded variables (Xi)- α-10+ 1+ αStep ΔxiAdsorbent dosage (g)X10,0520,10,20,30,3480,1Contact time (min)X210112016020021940Initial concentration (mg·L⁻¹)X35,1710203034,4810pHX45,526788,481 Table 1Coded and actual levels of the independent variables used in the CCDFactorsCoded variables (Xi)- α-10+ 1+ αStep ΔxiAdsorbent dosage (g)X10,0520,10,20,30,3480,1Contact time (min)X210112016020021940Initial concentration (mg·L⁻¹)X35,1710203034,4810pHX45,526788,481 Coded and actual levels of the independent variables used in the CCD The sequential approach consists of three parts: a factorial design (Nf = 2k); a star design (N0 = 2k); and center points (N0). For our modeling, the centered composite design used has four factors (K = 4) and two center points (N0 = 2). The desired polynomial model is expressed as follows: Y= a0 + a1 × 1 + a2 × 2 + a3 × 3 + a4 × 4+ a11 × 12 + a22 × 22+ a33 × 32 + a44 × 42 + a12 × 1 × 2 + a13 × 1 × 3 + a14 × 1 × 4 + a23 × 2 × 3 + a24 × 2 × 4 + a34 × 3 × 4. where Y is the predicted response (qₑ), a₀ is the intercept, ai are the linear coefficients, aii are the quadratic coefficients, and aiⱼ are the interaction coefficients. We used analysis of variance (ANOVA) to assess whether the model was statistically significant and adequate. Response surface plots were used to evaluate the effects of the operating variables and their interactions on the adsorption response, and the optimum operating conditions were determined by maximizing the adsorption capacity [44]. The optimisation study focused on B-CaHAp because the comparative study indicated its superior adsorption performance. Effect of ionic strength on FLCX-Na adsorption The effect of ionic strength on the adsorption of flucloxacillin sodium (FLCX-Na) by B-CaHAp was studied under adsorption conditions optimized by response surface methodology. Sodium chloride (NaCl) was used as the background electrolyte at concentrations of 0, 10, 50, and 100 mM [45]. After reaching adsorption equilibrium, the residual concentration of FLCX-Na was determined by the analytical method described in Sect. 2.4, and the removal efficiency and adsorption capacity were calculated using Eqs. (2) and (3). Regeneration and reusability study The regeneration and reuse of biogenic hydroxyapatite (B-CaHAp) were investigated over five successive adsorption-desorption cycles under optimal operating conditions determined by response surface methodology (RSM). Briefly, 0.10 g of B-CaHAp was added to 100 mL of a 25 mg L⁻¹ FLCX-Na solution at pH 6.5. The mixture was stirred at 25 ± 2 °C at 150 rpm for 160 min. The adsorbent was separated by filtration after adsorption, and the residual concentration of FLCX-Na was determined by UV-visible spectrophotometry. The removal efficiency (%) was calculated using Eq. (2) respectively. The spent B-CaHAp was regenerated by magnetic stirring (150 rpm) with 100 mL of a 0.01 M NaOH solution for 2 h at room temperature. The regenerated adsorbent was filtered, washed several times with distilled water to remove traces of NaOH, and dried overnight at 80 °C for reuse under the same adsorption conditions. The adsorption-desorption process was repeated five times. The removal efficiency (%) after each cycle was determined to evaluate the regeneration efficiency and operational stability of the adsorbent. Similar regeneration procedures, using dilute alkaline solutions, are commonly employed for hydroxyapatite-based adsorbents since they promote desorption while preserving the apatite structure [46]. Results and discussion X-ray diffraction Fig. 4XRD diffractograms of Syn-CaHAp and B-CaHAp XRD diffractograms of Syn-CaHAp and B-CaHAp The diffractograms of Syn-CaHAp and B-CaHAp show all the characteristic reflections of the hexagonal apatite phase associated with the P6₃/m space group, including the crystallographic planes (002), (211), (300), (202), and (310) (Fig. 4). All these peaks match well with JCPDS reference card No. 09-0432, indicating that a well-defined calcium phosphate hydroxyapatite was obtained [47]. The Syn-CaHAp sample prepared by wet precipitation at moderate temperature (85 °C) shows relatively broad and low-intensity diffraction peaks. This indicates moderate crystallinity and smaller size of crystallite. This is a common behavior for hydroxyapatites obtained by rapid aqueous processes in which nucleation dominates over crystal growth. This prevents formation of a very ordered crystal lattice [48]. However, B-CaHAp shows much sharper and more resolved diffraction peaks, which are also more intense, indicating higher crystallinity and better structural organization. The improvement is mainly due to the rigorous heat treatment it has undergone during its preparation, especially calcination at 900 °C. This process encourages crystal growth and recrystallization, and eliminates any remaining amorphous material. B-CaHAp contains various ions capable of substituting for one another, such as Na⁺, Mg²⁺, K⁺, Si⁴⁺, and Sr²⁺. These ions are often considered to distort the crystal lattice [18, 49]. However, heat treatment appears to compensate for this effect, resulting in an overall more crystalline structure. Similar results. have been documented for high-temperature-calcined biogenic apatites derived from bone or fish waste [36, 37]. To further support the qualitative interpretation of the XRD patterns, the mean crystallite size of both hydroxyapatite samples was calculated by the Scherrer equation applied to the isolated (002) diffraction peak. The crystallite size calculated was approximately 20.8 nm for Syn-CaHAp and 38.4 nm for B-CaHAp. The improved crystallite size achieved for B-CaHAp quantitatively corroborates the reduction in peak broadening and the increased crystalline organization of the material post-calcination, which is in line with the qualitative findings from the XRD patterns. Furthermore, the absence of additional peaks associated with secondary phases such as β-tricalcium phosphate (β-TCP), CaO, or other calcium phosphates attests to the high phase purity of the two materials examined. These observations highlight the crucial impact of the synthesis technique and heat treatment on the crystallinity of hydroxyapatite, which explains why B-CaHAp has greater crystallinity than Syn-CaHAp. The structural trends highlighted by XRD analysis will be corroborated in the following section by a morphological study using scanning electron microscopy (SEM). FT-IR spectra FTIR spectra confirmed the apatite phase formation in both materials. They also show that Syn-CaHAp and B-CaHAp have different structures (Fig. 5). The v(OH⁻) stretching band observed around 3570 cm⁻¹ for Syn-CaHAp is sharp and intense, indicating the presence of a significant number of structural hydroxyl groups. Also visible are the broad bands between 3000 and 3300 cm − 1 and the δ(H2O) deformation band centered around 1637 cm − 1. This points to the presence of water adsorbed between the molecules. This is a usual result of aqueous synthesis processes. This result is in agreement with the observations reported for hydroxyapatites prepared by the wet methods [10]. On the other hand, the bands related to water are significantly diminished or almost absent in the spectrum of B-CaHAp, which is explained by the calcination at high temperature, which allows the elimination of surface and interstitial water and partial dehydroxylation of the apatite structure [36, 37]. Fig. 5Comparative FTIR spectra of Syn-CaHAp and B-CaHAp Comparative FTIR spectra of Syn-CaHAp and B-CaHAp The characteristic bands of PO₄³⁻ phosphate groups at 1018 cm⁻¹, 1093 cm⁻¹, and 965 cm⁻¹ are present in both spectra, checking the apatite structure formation. These phosphate bands observed in the FTIR spectrum of Syn-CaHAp appear sharper than those of B-CaHAp. However, this sharpness should not be interpreted solely as evidence of increased crystallinity. Variations in hydration, carbonate substitution, and the immediate chemical environment around the phosphate groups can also influence the sharpness of the FTIR bands. Consequently, variations in local structural organization are likely reflected in the sharper phosphate bands observed in Syn-CaHAp [10]. Finally, the bands associated with carbonate groups (CO₃²⁻) observed around 1424–1450 cm⁻¹, 1329 cm⁻¹ and 874 cm⁻¹, are significantly more intense in B-CaHAp, confirming a type B carbonate substitution (CO₃²⁻ ↔ PO₄³⁻), characteristic of apatites of biological origin. The weak carbonate bands in Syn-CaHAp are probably due to atmospheric CO₂ dissolved during the wet precipitation process. In alkaline conditions some of these carbonate ions are replaced in the apatite lattice by B-type substitution (CO₃²⁻ replacing PO₄³⁻) [50]. These FTIR results are fully consistent with XRD analyses, showing that, despite higher carbonation and local heterogeneity of the lattice, B-CaHAp retains a high crystalline organization, while Syn-CaHAp has a more homogeneous but less recrystallized structure. Thermal analysis The TGA curves (Fig. 6, a) show that Syn-CaHAp loses mass slowly but significantly over the whole temperature range studied, reaching about 15–18% at 800–900 °C. The elimination of physically adsorbed water below 200 °C, loosely bound water between 200 and 400 °C, and the gradual breakdown of structurally integrated carbonate species between 400 and 700 °C are all responsible for the mass loss. FTIR analysis indicates that these carbonate species are mainly associated with B-type carbonate substitution (CO₃²⁻ substituting PO₄³⁻) in the apatite lattice. Such thermal behaviour is often reported for hydroxyapatite precipitated in wet conditions [51–53]. B-CaHAp, on the other hand, loses only about 4–5% of its mass, demonstrating its exceptional thermal stability. The success of earlier high-temperature calcination is confirmed by the modest loss seen below 200 °C, which corresponds to the removal of a small amount of adsorbed water. Between 200 and 400 °C the mass remains nearly constant therefore there is no significant amount of weakly bound water. Fig. 6Comparative thermal analyses of Syn-CaHAp and B-CaHAp: (a) thermogravimetric (TGA) curves and (b) heat flow profiles Comparative thermal analyses of Syn-CaHAp and B-CaHAp: (a) thermogravimetric (TGA) curves and (b) heat flow profiles Between 400 and 700 °C, B-CaHAp exhibits a slight mass loss. Although the majority of the carbonate species should have been eliminated by the previous calcination at 900 °C, the FTIR analysis still reveals faint carbonate bands, suggesting that some structurally integrated carbonate is still present in the material. Consequently, rather than the breakdown of the initial carbonate content, this heat event is thought to be connected to the evolution of these remaining carbonate species. This behavior is in line with earlier findings that, depending on the structural context of the carbonate ions in the apatite lattice, the carbonate content may not entirely disappear following calcination [52, 54, 55]. A pseudo-thermal plateau in this region suggests a stabilization phase rather than actual mass gain. Both materials achieve increased heat stability above 700 °C. Nevertheless, there is a small residual mass loss in B-CaHAp, which is explained by the late removal of carbonate groups that are tightly bonded to the apatite structure. This behavior is consistent with the higher thermal stability of B-type substituted carbonate species that have been reported for calcined biogenic hydroxyapatites [56]. Syn-CaHAp, on the other hand, exhibits earlier stabilisation, with most thermally unstable species removed at lower temperatures. Even with this late loss, B-CaHAp overall mass loss is still much lower than Syn-CaHAp, demonstrating its higher overall thermal stability [10]. These findings are supported by the heat flow graphs (Fig. 6, b). Water desorption and decarbonation are linked to a number of endothermic events for Syn-CaHAp, which are followed by exothermic transitions ascribed to structural rearrangements and high-temperature recrystallization processes. The B-CaHAp sample, which was previously calcined and thermally stabilised, shows a more continuous thermal history that is characterised by relaxation and structural reorganisation processes [53]. In conclusion, despite a higher degree of carbonation, B-CaHAp shows better thermal stability than Syn-CaHAp, which is explained by its densified structure and harsh thermal pretreatment. These findings are in complete agreement with the XRD and FTIR investigations, indicating that the origin and preparation technique have a significant impact on the hydroxyapatites thermal behaviour. SEM–EDS analysis SEM micrographs reveal distinct morphologies for Syn-CaHAp and B-CaHAp (Fig. 7). Syn-CaHAp consists of finely agglomerated particles forming compact clusters with a relatively smooth surface. This morphology, characteristic of materials obtained by aqueous precipitation, this agglomeration of particles, can limit the accessibility of adsorption sites [57]. Fig. 7Morphological and elemental characterization (SEM–EDS) of hydroxyapatites: (a) Syn-CaHAp and (b) B-CaHAp Morphological and elemental characterization (SEM–EDS) of hydroxyapatites: (a) Syn-CaHAp and (b) B-CaHAp In contrast, B-CaHAp exhibits more individualized grains with sharper contours and a more open microstructure inherited from the bone matrix and stabilized by the calcination. Despite higher crystallinity as evidenced by XRD, this organization promotes the exposure of well-defined crystalline surfaces as well as localized surface defects resulting from calcium non-stoichiometry and ion substitutions [57]. EDS analysis confirms, for both materials, the majority presence of the constituent elements of hydroxyapatite (Ca, P, and O), with no detection of secondary phases (Table 2). Table 2Elemental composition obtained by EDS analysis for Syn-CaHAp and B-CaHApElementsMass % (Syn-CaHAp)Atom % (Syn-CaHAp)Mass % (B-CaHAp)Atom % (B-CaHAp)O48.2267.9343.7863.64P17.8512.9921.9316.46Ca33.9319.0834.2919.90Ca/P-1.47-1.21 Table 2Elemental composition obtained by EDS analysis for Syn-CaHAp and B-CaHApElementsMass % (Syn-CaHAp)Atom % (Syn-CaHAp)Mass % (B-CaHAp)Atom % (B-CaHAp)O48.2267.9343.7863.64P17.8512.9921.9316.46Ca33.9319.0834.2919.90Ca/P-1.47-1.21 Elemental composition obtained by EDS analysis for Syn-CaHAp and B-CaHAp The Ca/P atomic ratios are lower than the ideal stoichiometric value (1.67), indicating calcium-deficient hydroxyapatites. Syn-CaHAp has a Ca/P ratio of approximately 1.47, reflecting moderate non-stoichiometry, while B-CaHAp has a lower ratio (≈ 1.21), typical of biogenic apatites and associated with the presence of calcium vacancies and carbonate substitutions [58]. Consequently, SEM-EDS analysis suggests that B-CaHAp has a more accessible contact surface and a higher density of active surface sites than Syn-CaHAp, which suggests better adsorption performance with respect to flucloxacillin sodium, in line with the structural and chemical differences highlighted by complementary techniques [57]. TEM analysis of hydroxyapatites nanoparticles The TEM images reveal distinct morphologies and states of aggregation between the two samples. In the case of Syn-CaHAp (a), the nanoparticles appear as elongated crystals, characterized by acicular or rod-like shapes, with a relatively uniform distribution. Their dimensions range from a few tens to about a hundred nanometers in length, with a tendency toward loose aggregation that leads to the formation of irregular clusters, though without excessive compaction. In contrast, B-CaHAp (b) has a distinct morphology consisting of smaller, less elongated nanoparticles that combine to form aggregates with a more ordered distribution and a very regular geometric organization. The 200-nm scale indicates that these aggregates are typically between 100 and 200 nm in size. Nevertheless, this. arrangement is compatible with the SEM data, which showed a more open and diverse microstructure rather than a compact morphology. These differences in shape and aggregation state may be caused by the distinct synthesis pathways and carbonate substitution in B-CaHAp, which can influence nucleation and crystal growth during material production. Overall, the TEM data demonstrate that the aggregation and particle morphology of the two hydroxyapatites are very different. While B-CaHAp is made up of smaller nanoparticles with a relatively more organized aggregation pattern, Syn-CaHAp is made up of loosely aggregated elongated nanocrystals. These observations are in agreement with the SEM analysis showing the differences in the surface morphology of the particle agglomerates. Hence, SEM and TEM results give complementary structural information at different observations. (Fig. 8). Fig. 8TEM analysis of hydroxyapatites nanoparticles: (a) Syn-CaHAp and (b) B-CaHAp TEM analysis of hydroxyapatites nanoparticles: (a) Syn-CaHAp and (b) B-CaHAp Point of zero charge (PZC) analysis The pH derivation method (ΔpH = f(pH_initial)) has been used to determine the point of charge null (pHpzc) as shown in Fig. 9). Syn-CaHAp is a hydroxyapatite that is close to stoechiometry and whose surface is mostly controlled by the Ca–OH groups. It has a pHpzc of about 6.5. According on the pH of the environment, these amphotères control protonation and deprotonation and determine surface charge [59]. Fig. 9Determination of the zero charge point (pHpzc) of hydroxyapatites: Syn-CaHAp and B-CaHAp Determination of the zero charge point (pHpzc) of hydroxyapatites: Syn-CaHAp and B-CaHAp In line with its calcium-deficient nature (Ca/P = 1.21) and the type B carbonate substitution (CO₃²⁻ ↔ PO₄³⁻) detected by FTIR, B-CaHAp exhibits a lower pHpzc (≈ 5,8). These substitutions, along with a partial deshydroxylation during calcination, reduce the density of the Ca–OH groups and make the surface somewhat more acidic while maintaining the appropriate structure [60]. From an applied perspective, the surface becomes negatively charged when the solution’s pH rises over pHpzc, which encourages the electrostatic attraction of cationic or partially protonated species; conversely, when pH falls below pHpzc, the positive surface may inhibit these interactions [61]. As a result, the two materials’ different pHpzc values directly reflect their structural changes and provide a crucial parameter for interpreting how they interact with ionisable pollutants in aqueous solution. Flucloxacillin sodium adsorption on synthetic and biogenic hydroxyapatite Effect of adsorbent mass To determine the ideal quantity of hydroxyapatite needed to enhance adsorption efficiency, the impact of the adsorbent mass on the removal of the adsorbate was investigated (Fig. 10). Increasing the mass of the adsorbent enhances the percentage of removal for both compounds under study, indicating an increase in the number of active sites on the solid’s surface. The clearance percentage for Syn-CaHAp rises steadily, peaking at roughly 62% at a mass of 0.2 g. Beyond this mass, no significant improvement is observed. This trend can be attributed to the saturation of accessible active sites and the agglomeration of particles with high solid load, a phenomenon that likely reduces the available adsorption sites due to agglomeration of particles, as frequently reported for synthetic hydroxyapatites prepared by aqueous precipitation [62]. Fig. 10Influence of Syn-CaHAp and B-CaHAp mass on the adsorption of FLCX-Na in an aqueous medium. Volume of solution = 100 mL; Concentration = 20 mg·L⁻¹; Time = 3 h; Stirring speed = 150 rpm; Temperature = 25 °C Influence of Syn-CaHAp and B-CaHAp mass on the adsorption of FLCX-Na in an aqueous medium. Volume of solution = 100 mL; Concentration = 20 mg·L⁻¹; Time = 3 h; Stirring speed = 150 rpm; Temperature = 25 °C Conversely, B-CaHAp achieves a maximum elimination percentage of approximately 77% at a lower mass of 0.1 g, indicating superior adsorption efficiency. This performance can be linked to its more open microstructure, its non-stoichiometric calcium content, and the presence of natural ion substitutions (CO₃²⁻, Mg²⁺, Na⁺), which contribute to an increased density of active sites and promote multiple interactions with FLCX-Na [63]. For both adsorbents, the absence of a significant increase in the percentage of removal beyond the optimal masses is mainly due to the saturation of active sites and agglomeration phenomena. These results highlight that B-CaHAp achieves high removal efficiency with a reduced mass of adsorbent compared to Syn-CaHAp, thus emphasizing its strong potential for sustainable applications in the treatment of antibiotic-contaminated water. Effect of pH The adsorption of FLCX-Na is highly dependent on pH for both materials (Fig. 11). For Syn-CaHAp, the adsorption capacity increases gradually from an acidic environment to a pH close to neutrality (≈ 6), then decreases in an alkaline environment. This change can be explained by the moderate protonation of surface groups (≡ Ca–OH₂⁺), which promotes electrostatic interactions and coordination between surface Ca²⁺ ions and the carboxylate functions of the antibiotic [64]. However, the more stoichiometric structure, which is less rich in defects, limits the number of accessible reactive sites. In contrast, B-CaHAp has higher adsorption capacities across the entire pH range, with a maximum in the slightly acidic range (≈ 5.5–6). The observed enhanced performance is attributed to its calcium-deficient nature (low Ca/P ratio) and B-type carbonate substitution that alter the local chemical environment of the apatite lattice and promote complexation and ion-exchange interactions with FLCX-Na [65, 66]. In an alkaline environment (pH > 7), the deprotonation of apatite leads to the formation of ≡ Ca–O⁻ sites, while the increase in OH⁻ ion concentration causes competition with the pollutant for coordination at the surface calcium centers. The electrostatic repulsion between the negatively charged surface and the anionic species of FLCX-Na, combined with this competition, causes a net decrease in adsorption [64, 66]. Fig. 11Influence of pH on FLCX-Na adsorption onto Syn-CaHAp and B-CaHAp. Volume of solution = 100 mL; Adsorbent mass = 0.2 g; Concentration = 20 mg·L⁻¹; Time = 3 h; Stirring speed = 150 rpm; Temperature = 25 °C Influence of pH on FLCX-Na adsorption onto Syn-CaHAp and B-CaHAp. Volume of solution = 100 mL; Adsorbent mass = 0.2 g; Concentration = 20 mg·L⁻¹; Time = 3 h; Stirring speed = 150 rpm; Temperature = 25 °C Effect of initial antibiotic concentration Increasing the initial concentration of FLCX-Na leads to an increase in adsorption capacity for both hydroxyapatites, reflecting a progressive interaction between the antibiotic molecules and the active surface sites (Fig. 12) [67]. For Syn-CaHAp, maximum performance is achieved at an optimal initial concentration of 50 mg·L⁻¹, corresponding to a maximum experimental capacity of 15 mg·g⁻¹. In contrast, B-CaHAp exhibits optimal performance at a higher initial concentration of 70 mg·L⁻¹, with a maximum capacity of 26 mg·g⁻¹, reflecting increased affinity and better utilization of available adsorption sites [10]. At initial concentrations above the optimal values, the consistency of the relative adsorption capacity values observed for both materials is attributed to the saturation of the available active sites [67]. Comparison of the two materials thus highlights the superior ability of B-CaHAp to treat solutions containing higher initial concentrations of our pollutant, confirming its suitability for applications involving the treatment of water contaminated with antibiotics [68]. Fig. 12Influence of the initial concentration of FLCX-Na on its adsorption by Syn-CaHAp and B-CaHAp in an aqueous medium. Volume of solution = 100 mL; Adsorbent mass = 0.2 g; Time = 3 h; Stirring speed = 150 rpm; Temperature = 25 °C Influence of the initial concentration of FLCX-Na on its adsorption by Syn-CaHAp and B-CaHAp in an aqueous medium. Volume of solution = 100 mL; Adsorbent mass = 0.2 g; Time = 3 h; Stirring speed = 150 rpm; Temperature = 25 °C Effect of contact time The effect of contact time on the adsorption of FLCX-Na was studied to better understand the dynamics of interaction between the antibiotic and hydroxyapatite surfaces (Fig. 13). For both materials, the adsorption capacity increases rapidly during the initial contact time, reflecting effective interaction with the initially available active surface sites [67]. The adsorption capacity of Syn-CaHAp rises steadily until it reaches a maximum capacity of 6.17 mg·g⁻¹, which corresponds to an ideal contact period of 100 min. After this point, the adsorption capacity essentially stays constant, signifying the creation of an equilibrium between the adsorbent’s surface and the aqueous phase [67]. For B-CaHAp, adsorption occurs more rapidly, with maximum performance being achieved at a shorter contact time of 60 min, with a maximum capacity of 7.93 mg·g⁻¹. After this optimal time, no significant variation in adsorption capacity is observed, indicating stabilization of the system [68]. This phase of stability observed after the optimal contact times for both materials suggests that the majority of accessible active sites are occupied and that exchanges between the solution and the solid surface reach a dynamic equilibrium. The faster kinetics and higher capacity observed for B-CaHAp confirm its suitability for applications involving the treatment of water contaminated with antibiotics, particularly in processes requiring reduced contact times [68]. Fig. 13Influence of contact time on the adsorption of FLCX-Na on Syn-CaHAp and B-CaHAp in an aqueous medium. Volume of solution = 100 mL; Adsorbent mass = 0.2 g; Initial concentration = 20 mg·L⁻¹; Stirring speed = 150 rpm; Temperature = 25 °C Influence of contact time on the adsorption of FLCX-Na on Syn-CaHAp and B-CaHAp in an aqueous medium. Volume of solution = 100 mL; Adsorbent mass = 0.2 g; Initial concentration = 20 mg·L⁻¹; Stirring speed = 150 rpm; Temperature = 25 °C Kinetic studies A kinetic study of the adsorption of our pollutant on hydroxyapatites was conducted to analyze the speed of the process and the mechanism controlling the adsorbent-adsorbate interaction. The experimental data were fitted to pseudo-first-order (PFO) and pseudo-second-order (PSO) kinetic models, as illustrated by the corresponding curves (Fig. 14) [67]. In order comparison, the two models were studied and the one giving the highest correlation coefficients (R2) and the best agreement between the calculated and experimental qe values was considered the most appropriate for describing the adsorption kinetics. For Syn-CaHAp, the kinetic curves show a gradual increase in the amount adsorbed over time, followed by the attainment of equilibrium. The data fit, which demonstrates that the pseudo-second-order model more accurately describes the experimental behavior than the pseudo-first-order model, is supported by the higher correlation coefficients displayed in the summary table (Table 3) and the improved correlation between the calculated and experimental values of qe [69]. Similar to this, B-CaHAp exhibits rapid adsorption during the initial contact phase before stabilizing at equilibrium. Furthermore, the pseudo-second-order model better explained the kinetic data, suggesting that it is the most suitable model to depict the adsorption kinetics. However, the nature of the adsorption mechanism cannot be conclusively demonstrated by the kinetic model alone. Therefore, for a complete understanding of the adsorption process, complementing equilibrium and mechanistic studies are required. Fig. 14Kinetic modeling of (a) pseudo-first order and (b) pseudo-second order adsorption of FLCX-Na by Syn-CaHAp and B-CaHAp Kinetic modeling of (a) pseudo-first order and (b) pseudo-second order adsorption of FLCX-Na by Syn-CaHAp and B-CaHAp Comparison of the kinetic parameters obtained for the two materials reveals higher rate constants and equilibrium capacity values closer to the experimental data for B-CaHAp, reflecting faster adsorption kinetics and more effective interaction with FLCX-Na. These results confirm the superiority of the biogenic material in terms of process speed, a key criterion [69]. Table 3Pseudo-first-order and pseudo-second-order kinetic parameters for the adsorption of FLCX-Na on Syn-CaHAp and B-CaHApSyn-CaHApB-CaHAp q e, exp (mg.g − 1 ) 6.177.93 Pseudo first order R 2 0.95060.6893 q e,1 (mg.g − 1 ) 2.622.71 k 1 (min − 1 ) 0.030.03 Pseudo seconde order R 2 0.99970.9987 q e,2 (mg.g − 1 ) 6.448.24 K 2 (g.mg⁻¹.min⁻¹) 0.070.14 Table 3Pseudo-first-order and pseudo-second-order kinetic parameters for the adsorption of FLCX-Na on Syn-CaHAp and B-CaHApSyn-CaHApB-CaHAp q e, exp (mg.g − 1 ) 6.177.93 Pseudo first order R 2 0.95060.6893 q e,1 (mg.g − 1 ) 2.622.71 k 1 (min − 1 ) 0.030.03 Pseudo seconde order R 2 0.99970.9987 q e,2 (mg.g − 1 ) 6.448.24 K 2 (g.mg⁻¹.min⁻¹) 0.070.14 Pseudo-first-order and pseudo-second-order kinetic parameters for the adsorption of FLCX-Na on Syn-CaHAp and B-CaHAp The phases in charge of FLCX-Na adsorption on Syn-CaHAp and B-CaHAp were identified using the intraparticulate diffusion (IPD) model. For all materials, the qt curves versus \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\rm{t}}^{1/2}$$\end{document} show three distinct phases, indicating a multi-step adsorption process (Fig. 15). Rapid adsorption, mostly regulated by external mass transfer, is the hallmark of phase I. The initial phase of adsorption is faster with B-CaHAp due to its greater affinity for FLCX-Na compared to Syn-CaHAp. Fig. 15Intraparticle diffusion (IPD) curves for the adsorption of FLCX-Na on Syn-CaHAp and B-CaHAp Intraparticle diffusion (IPD) curves for the adsorption of FLCX-Na on Syn-CaHAp and B-CaHAp Intraparticulate diffusion is associated with Phase II. Because B-CaHAp has a higher adsorption capacity than Syn-CaHAp, this process is more noticeable, indicating a more effective transfer of FLCX-Na molecules to the material’s interior pores. The high R² values observed during this step indicate that the IPD model accurately represents this phase and that internal diffusion plays an important role in managing the adsorption process (Table 4) [70]. Phase III is associated with reaching equilibrium. The very gentle slopes observed for both adsorbents reflect the gradual saturation of active sites, with a higher final capacity for B-CaHAp. Low rate constants (K₃) suggest limited diffusion, with the process being essentially regulated by [71]. The fact that the lines do not pass through the origin in all phases confirms that diffusion within the particles is not the only process involved. Adsorption is the result of a combination of external mass transfer, diffusion within the particles, and final equilibrium [72]. Table 4Intraparticle diffusion (IPD) model parameters for the adsorption of FLCX-Na on Syn-CaHAp and B-CaHApAdsorbentsPhase IPhase IIPhase IIIK1 (mg.g 1min− 1/2)C (mg.g− 1) R 2 K2 (mg.g −1min− 1/2)C (mg.g− 1) R 2 K3 (mg.g −1min− 1/2)C (mg.g− 1) R 2 Syn-CaHAp1.0570.9270.9650.3493.2360.9930.0195.9960.864B-CaHAp1.994-1.5000.9830.4094.7910.9960.0197.6540.388 Table 4Intraparticle diffusion (IPD) model parameters for the adsorption of FLCX-Na on Syn-CaHAp and B-CaHApAdsorbentsPhase IPhase IIPhase IIIK1 (mg.g 1min− 1/2)C (mg.g− 1) R 2 K2 (mg.g −1min− 1/2)C (mg.g− 1) R 2 K3 (mg.g −1min− 1/2)C (mg.g− 1) R 2 Syn-CaHAp1.0570.9270.9650.3493.2360.9930.0195.9960.864B-CaHAp1.994-1.5000.9830.4094.7910.9960.0197.6540.388 Intraparticle diffusion (IPD) model parameters for the adsorption of FLCX-Na on Syn-CaHAp and B-CaHAp Adsorption isotherms In order to avoid the statistical distortions associated with linearized transformations and to provide a more accurate description of the adsorption behavior, the equilibrium adsorption data were fitted using the nonlinear forms of the Langmuir and Freundlich isotherm models (Fig. 16). The fitting parameters are summarized in Table 5 [67]. Fig. 16Non-Linear representations (a) of the Langmuir model (b) of the Freundlich model for the adsorption of FLCX-Na on Syn-CaHAp and B-CaHAp in an aqueous medium Non-Linear representations (a) of the Langmuir model (b) of the Freundlich model for the adsorption of FLCX-Na on Syn-CaHAp and B-CaHAp in an aqueous medium The nonlinear Langmuir model provided a better fit for the equilibrium data of Syn-CaHAp with a greater coefficient of determination (R2 = 0.9501) than the nonlinear Freundlich model (R2 = 0.8793). The Langmuir model showed a theoretical monolayer adsorption capacity of 23.47 ± 2.75 mg·g⁻¹, and the Langmuir affinity constant (KL = 0.0236 ± 0.0066 L·mg⁻¹) indicated a relatively strong affinity between flucloxacillin sodium molecules and the synthetic hydroxyapatite surface. The theoretical adsorption capacity was greater than the highest experimental adsorption capacity; this disparity is expected as the Langmuir parameter is the theoretical monolayer capacity estimated by the nonlinear regression, and complete saturation of the adsorption sites was not reached in the observed concentration range. Likewise, the adsorption equilibrium on B-CaHAp was best fitted to the nonlinear Langmuir model (R2 = 0.9665) rather than the nonlinear Freundlich model (R2 = 0.9335). The biogenic material exhibited superior adsorption potential, as evidenced by its estimated monolayer adsorption capacity of 49.38 ± 7.88 mg·g − 1, which was substantially higher than that of Syn-CaHAp. The calcium-deficient structure, carbonate substitution, and surface heterogeneity of B-CaHAp, which increase the number of accessible adsorption sites, are plausible explanations for this capacity increase. Because the maximum monolayer saturation was not reached under the experimental conditions, as was seen for Syn-CaHAp, the theoretical Langmuir capacity was still greater than the maximum adsorption capacity found experimentally. The Langmuir model was shown to be the best model to explain the equilibrium adsorption data for both adsorbents, according to nonlinear regression analysis. This result suggests that monolayer formation at energetically comparable active sites is the dominant adsorption mode within the studied concentration range. Compared to linearized isotherm equations, nonlinear regression allows for a more precise estimation of the parameters and, consequently, a more rigorous basis for model selection, as it preserves the initial error structure of the experimental data. It is important to emphasize that the predominance of the Langmuir isotherm does not contradict the previously established pseudo-second-order kinetic model. The Langmuir model is used to represent the equilibrium distribution of the adsorbate at the solid-solution interface, while the kinetic model describes the rate and mechanism of adsorption. Thus, these two models are complementary to each other and confirm the fact that the adsorption of flucloxacillin sodium onto Syn-CaHAp and B-CaHAp is controlled by specific interactions resulting in the equilibrium monolayer adsorption [73]. Table 5Langmuir and Freundlich isotherm parameters for the adsorption of FLCX-Na on Syn-CaHAp and B-CaHApSyn-CaHApB-CaHAp q e, exp (mg.g − 1 ) 1526Langmuirqm (mg.g− 1)23.47 ± 2.7549.38 ± 7.88KL (L.mg⁻¹).0.0236 ± 0.00660.0117 ± 0.0034R20.95010.9665FreundlichKF1.644 ± 0.6491.400 ± 0.536n1.987 ± 0.3701.561 ± 0.217R20.87930.9335 Table 5Langmuir and Freundlich isotherm parameters for the adsorption of FLCX-Na on Syn-CaHAp and B-CaHApSyn-CaHApB-CaHAp q e, exp (mg.g − 1 ) 1526Langmuirqm (mg.g− 1)23.47 ± 2.7549.38 ± 7.88KL (L.mg⁻¹).0.0236 ± 0.00660.0117 ± 0.0034R20.95010.9665FreundlichKF1.644 ± 0.6491.400 ± 0.536n1.987 ± 0.3701.561 ± 0.217R20.87930.9335 Langmuir and Freundlich isotherm parameters for the adsorption of FLCX-Na on Syn-CaHAp and B-CaHAp To further investigate the adsorption mechanism, equilibrium data were analyzed using the Dubinin-Radushkevich (D-R) isotherm model (Fig. 17), the parameters of which are summarized in Table 6. Fig. 17Plots showing the Dubinin–Radushkevich (D–R) isotherm for FLCX-Na adsorption onto Syn-CaHAp and B-CaHAp at 25 °C Plots showing the Dubinin–Radushkevich (D–R) isotherm for FLCX-Na adsorption onto Syn-CaHAp and B-CaHAp at 25 °C Compared to the Langmuir model, the D-R model described the adsorption equilibrium less satisfactorily and had comparatively low coefficients of determination (R² = 0.8238 for Syn-CaHAp and 0.7430 for B-CaHAp). Additionally, the experimental equilibrium adsorption capacities were much higher than the theoretical adsorption capacities determined by the D-R model (12.16 and 17.11 mg·g⁻¹ for Syn-CaHAp and B-CaHAp, respectively). This suggests that the D-R model is not appropriate for explaining the adsorption behavior of FLCX-Na on the two adsorbents. The Dubinin-Radushkevich (D-R) model yielded mean adsorption free energy (E) values of 0.209 and 0.203 kJ·mol⁻¹ for Syn-CaHAp and B-CaHAp, respectively. These values are generally associated with weak adsorption interactions, but their interpretation should be cautious due to the poor fit of the D-R model. Therefore, the adsorption mechanism cannot be definitively determined from the D–R isotherm only. The D–R model gives useful information about apparent adsorption energy. However, the interpretation should be supported by other experimental evidence. In the present study, the combination of nonlinear Langmuir isotherm, pseudo second order kinetic model and pH dependent. behavior of adsorption deliver a more comprehensive understanding of adsorption process. The non-linear Langmuir model suggests that the equilibrium adsorption is primarily governed by monolayer coverage on available active sites. While the kinetic results demonstrate the rate behavior of adsorption. The complementary analyses thus give a more reliable interpretation of the entire adsorption mechanism than any single model considered alone. In the next section, the adsorption mechanism at the molecular scale is discussed. Table 6Dubinin–Radushkevich (D–R) isotherm parameters for FLCX-Na adsorption on Syn-CaHAp and B-CaHApSyn-CaHApB-CaHApqm (mg·g− 1)12.1617.11 β (mol 2 ·kJ − 2 ) 11.4712.12 E (kJ·mol⁻¹) 0.2090.203 R² 0.82380.7430 Table 6Dubinin–Radushkevich (D–R) isotherm parameters for FLCX-Na adsorption on Syn-CaHAp and B-CaHApSyn-CaHApB-CaHApqm (mg·g− 1)12.1617.11 β (mol 2 ·kJ − 2 ) 11.4712.12 E (kJ·mol⁻¹) 0.2090.203 R² 0.82380.7430 Dubinin–Radushkevich (D–R) isotherm parameters for FLCX-Na adsorption on Syn-CaHAp and B-CaHAp Proposed adsorption mechanism Based on experimental and theoretical results, a plausible chemisorption mechanism for the adsorption of FLCX-Na onto hydroxyapatite is proposed, involving the interactions between the oxygen-containing functional groups of FLCX − and calcium active sites present on the hydroxyapatite surface (Fig. 18). In aqueous solution, flucloxacillin is largely deprotonated above pH ≈ 4 to negatively charged carboxylate species that can interact with surface Ca2+ centers [74]. The good agreement between the experimental data and the pseudo-second-order kinetic model implies that chemical interactions controrepresents the diffusion ratel the adsorption process [75]. At the molecular level, the proposed adsorption mechanism involves adsorption surface complexation between carboxylate groups of FLCX⁻ and exposed calcium sites such as ≡ Ca–OH or coordinatively unsaturated Ca²⁺ centers to form stable calcium–carboxylate complexes according to the following simplified reaction:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\equiv {\rm{Ca-OH + RCOO}}^{-} \longrightarrow \equiv {\rm{Ca-OOC-R + OH}}^{-}.$$\end{document} where RCOO- is the deprotonated flucloxacillin molecule. These coordination interactions are in line with previously reported density functional theory (DFT) studies describing the formation of stable O–Ca coordination bonds during antibiotic adsorption to hydroxyapatite surfaces [76]. However, the amide or β-lactam functional groups of flucloxacillin can also form hydrogen bonds with hydroxyl or phosphate groups on the surface of hydroxyapatite. Also, electrostatic interactions play a role in the adsorption mechanism, particularly at slightly acidic to near-neutral conditions. The maximum adsorption was found in the pH range 5.5–6.5, where the favorable electrostatic attraction and calcium complexation coexist. Conversely, in alkaline conditions, adsorption decreased owing to electrostatic repulsion and the competition of OH- ions with active calcium sites [77]. Furthermore, the nonlinear Langmuir isotherm provided the best fit for both synthetic and biogenic hydroxyapatites, indicating that FLCX-Na adsorption primarily occurred through monolayer coverage of the available active sites under the investigated experimental conditions. The calcium-deficient nature of biogenic hydroxyapatite, B-type carbonate substitution, and unique structural and morphological features identified by XRD, FTIR, SEM, TEM, and EDS analyses, which provide a better adsorption environment for FLCX-Na, may all contribute to its increased adsorption capacity [78]. Fig. 18Proposed adsorption mechanism of FLCX-Na onto hydroxyapatite Proposed adsorption mechanism of FLCX-Na onto hydroxyapatite Statistical optimization and modeling The central composite design (CCD) matrix and the corresponding experimental adsorption capacities (qe) are summarized in Table 7. Twenty-six experiments were conducted to investigate the effect of adsorbent dosage (W), contact time (t), initial FLCX-Na concentration (C0), and solution pH, individually and in combination, on the adsorption performance. The experimental adsorption capacities ranged from 1.518 to 29.532 mg·g⁻¹, demonstrating the varying impact of the studied operating variables on the selected experimental domain. The duplicate center-point experiments (Runs 4 and 20) were carried out under the same conditions (W = 0.20 g, t = 160 min, C0 = 20 mg·L− 1, and pH = 7) and yielded adsorption capacities of 7.711 and 7.741 mg·g− 1 with a relative deviation of only 0.4%. This good agreement indicates the high reproducibility of the experimental design and suggests that the observed variations in the adsorption capacity are mainly caused by the investigated operating variables and not by the experimental uncertainty. The adsorbent dosage was observed to have the highest negative effect on the adsorption capacity in the comparative study of the experimental runs. The increase of W from 0.052 g (Run 11) to 0.348 g (Run 7) decreased qe from 29.532 to 4.832 mg·g⁻¹, indicating a decrease in adsorption capacity per unit mass of adsorbent with an increase in adsorbent dosage at a constant amount of adsorbate. In contrast, the qe increased from 1.518 to 13.710 mg·g− 1 with the increase of the initial FLCX-Na concentration from 5.17 mg·L− 1 (Run 16) to 34.83 mg·L− 1 (Run 24) due to the higher mass transfer driving force. Conversely, only minor changes were observed for pH and contact time in the CCD ranges examined. For example, adsorption capacity at pH 5.52 (Run 19, 8.579 mg·g− 1) and pH 8.48 (Run 5, 7.253 mg·g− 1) was relatively close. Similarly, increasing the contact time from 101 min (Run 15, 7.632 mg·g− 1) to 219 min (Run 13, 7.774 mg·g− 1) caused a slight change in the adsorption capacity. Table 7Design of experiments: influence of operating parameters on the qₑ responseRunConfigurationW (g)t (min)C0 (mg.L− 1)pHqₑ (mg.g− 1) 1 ++++ 0,32003088,179 2 +−−− 0,31201062,692 3 +−++ 0,31203088,112 4 0000 0,21602077,711 5 000 A 0,2160208,487,253 6 −−−− 0,11201067,489 7 A000 0,3481602074,832 8 −+−+ 0,12001087,283 9 ++−− 0,32001062,73 10 −+++ 0,120030819,113 11 a000 0,05216020729,532 12 ++−+ 0,32001082,217 13 0A00 0,22192077,774 14 −−−+ 0,11201087,24 15 0a00 0,21012077,632 16 00a0 0,21605,1771,518 17 +++− 0,32003068,692 18 −+−− 0,12001067,65 19 000a 0,2160205,528,579 20 0000 0,21602077,741 21 +−+− 0,31203068,611 22 −−++ 0,112030819,033 23 −−+− 0,112030624,13 24 00A0 0,216034,83713,71 25 −++− 0,120030623,042 26 +−−+ 0,31201082,131 Table 7Design of experiments: influence of operating parameters on the qₑ responseRunConfigurationW (g)t (min)C0 (mg.L− 1)pHqₑ (mg.g− 1) 1 ++++ 0,32003088,179 2 +−−− 0,31201062,692 3 +−++ 0,31203088,112 4 0000 0,21602077,711 5 000 A 0,2160208,487,253 6 −−−− 0,11201067,489 7 A000 0,3481602074,832 8 −+−+ 0,12001087,283 9 ++−− 0,32001062,73 10 −+++ 0,120030819,113 11 a000 0,05216020729,532 12 ++−+ 0,32001082,217 13 0A00 0,22192077,774 14 −−−+ 0,11201087,24 15 0a00 0,21012077,632 16 00a0 0,21605,1771,518 17 +++− 0,32003068,692 18 −+−− 0,12001067,65 19 000a 0,2160205,528,579 20 0000 0,21602077,741 21 +−+− 0,31203068,611 22 −−++ 0,112030819,033 23 −−+− 0,112030624,13 24 00A0 0,216034,83713,71 25 −++− 0,120030623,042 26 +−−+ 0,31201082,131 Design of experiments: influence of operating parameters on the qₑ response We performed an analysis of variance (ANOVA) to assess the statistical significance and model fit. Table 8 presents the results. The experimental Fisher’s F-value (F(exp) = 7.50) is higher than the theoretical one (F0.01 (14,11) = 4.29), which means that the regression model is statistically significant. The F-value is the ratio of the model variance to the residual variance (the overall significance of the regression model). The p-value is the probability of obtaining the corresponding statistical test. Moreover, the very low p-value (p = 0.0009) confirms the adequacy of the proposed quadratic model for the adsorption process. Table 8Analysis of variance using JMPSourceDegrees of freedomSum of squaresMean squareF-valueModel141198,193585,58537,4968Residuals11125,578411,4162 Prob. > F Total251323,77190,0009* Table 8Analysis of variance using JMPSourceDegrees of freedomSum of squaresMean squareF-valueModel141198,193585,58537,4968Residuals11125,578411,4162 Prob. > F Total251323,77190,0009* Analysis of variance using JMP Furthermore, the close agreement between the experimental and predicted values (Fig. 19) attests to the reliability of the developed model. This model exhibited a high coefficient of determination (R² = 0.91) and a low RMSE value (3.38), confirming its strong predictive power. Fig. 19Correlation between the actual adsorbed amount and the predicted adsorbed amount of FLCX-Na Correlation between the actual adsorbed amount and the predicted adsorbed amount of FLCX-Na According to Table 9, the regression coefficients reveal that the mass of the adsorbent (X₁), the initial concentration (X₃), the quadratic term (X₁²), and their interaction (X₁X₃) are the most significant factors. Consequently, a simplified model was developed: Y = 9,403886–5,350889.X1 + 5,0928889.X3 + 7,2188952.(X1)2 − 1,9895.(X1 × 3). This model highlights the predominant influence of the amount of adsorbent and the initial concentration in controlling the adsorption process. Table 9Estimated regression coefficients, standard errors, t-values, and p-values of the quadratic model for FLCX-Na adsorptionTermEstimateStandard Errort-valuesp-value Constante 9,4038861,3393947,02<,0001X1-5,3508890,796388-6,72<,0001X35,09288890,7963886,39<,0001X1* X17,21889522,1113413,420,0057X1* X3-1,98950,844697-2,360,0381X3* X3-2,3491052,111341-1,110,2896X2* X2-2,2601052,111341-1,070,3073X4* X4-2,0471052,111341-0,970,3531X4-0,7252220,796388-0,910,3820X3* X4-0,521750,844697-0,620,5494X1* X40,472250,8446970,560,5873X2* X3-0,074250,844697-0,090,9315X2* X40,067750,8446970,080,9375X1* X20,067250,8446970,080,9380X2-0,0216670,796388-0,030,9788 Table 9Estimated regression coefficients, standard errors, t-values, and p-values of the quadratic model for FLCX-Na adsorptionTermEstimateStandard Errort-valuesp-value Constante 9,4038861,3393947,02<,0001X1-5,3508890,796388-6,72<,0001X35,09288890,7963886,39<,0001X1* X17,21889522,1113413,420,0057X1* X3-1,98950,844697-2,360,0381X3* X3-2,3491052,111341-1,110,2896X2* X2-2,2601052,111341-1,070,3073X4* X4-2,0471052,111341-0,970,3531X4-0,7252220,796388-0,910,3820X3* X4-0,521750,844697-0,620,5494X1* X40,472250,8446970,560,5873X2* X3-0,074250,844697-0,090,9315X2* X40,067750,8446970,080,9375X1* X20,067250,8446970,080,9380X2-0,0216670,796388-0,030,9788 Estimated regression coefficients, standard errors, t-values, and p-values of the quadratic model for FLCX-Na adsorption Effects of variables and interactions The prediction figures and graphical analyses (Figs. 20, 21 and 22) illustrate the impact of operating parameters on the adsorption capacity of FLCX-Na. The increase in adsorption capacity (qₑ) with initial concentration indicates an increase in the driving force for mass transfer. Conversely, an increase in the amount of adsorbent decreases qₑ due to a dilution effect on the active sites. Fig. 20Prediction profiles illustrating the effects of operating variables on the adsorption capacity (qₑ) of FLCX-Na Prediction profiles illustrating the effects of operating variables on the adsorption capacity (qₑ) of FLCX-Na Fig. 21Three-dimensional response surface plots and corresponding contour plots illustrating the combined effects of operating variables on the adsorption capacity (qₑ) of FLCX-Na: (a) pH and contact time \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$(W = 0.2 \, g, C_{o} = 20 \, mg \cdot L^{-1})$$\end{document}. (b) pH and adsorbent dosage \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$(C_{o} = 20 \, mg \cdot L^{-1}, t = 160 \min)$$\end{document}. (c) pH and initial concentration (W = 0.2 g, t = 160 min). (d) initial concentration and adsorbent dosage (t = 160 min, pH = 7) Three-dimensional response surface plots and corresponding contour plots illustrating the combined effects of operating variables on the adsorption capacity (qₑ) of FLCX-Na: (a) pH and contact time \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$(W = 0.2 \, g, C_{o} = 20 \, mg \cdot L^{-1})$$\end{document}. (b) pH and adsorbent dosage \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$(C_{o} = 20 \, mg \cdot L^{-1}, t = 160 \min)$$\end{document}. (c) pH and initial concentration (W = 0.2 g, t = 160 min). (d) initial concentration and adsorbent dosage (t = 160 min, pH = 7) The ideal pH is around 6.5, which creates conditions conducive to interactions between the hydroxyapatite surface and the pollutant ions. The effect of contact time is moderate, with equilibrium generally being reached after about 160 min. The ideal pH is around 6.5, which creates conditions conducive to interactions between the hydroxyapatite surface and the pollutant ions. The effect of contact time is moderate, with equilibrium generally being reached after about 160 min. The non-significance of pH and time in the quadratic model (p > 0.35) is explained by the fact that the CCD experimental domain, centred around the previously identified optimal values (pH 6-6.5; 160 min), corresponded to a plateau region of the response surface, where local sensitivity of adsorption to these factors becomes statistically negligible. The interaction plots (Fig. 22) indicate that the effect of adsorbent mass depends on the initial concentration. The Pareto chart (Fig. 23) and the aggregated response surfaces (Fig. 21) confirm this significant interaction, illustrating the interrelated nature of the adsorption phenomena. Fig. 22Interaction plots showing the combined effects of operating variables on the adsorption capacity (qₑ) of FLCX-Na Interaction plots showing the combined effects of operating variables on the adsorption capacity (qₑ) of FLCX-Na Fig. 23Pareto chart illustrating the relative significance of the main factors and their interactions affecting the adsorption capacity of FLCX-Na Pareto chart illustrating the relative significance of the main factors and their interactions affecting the adsorption capacity of FLCX-Na Response surface analysis and optimal conditions The three-dimensional response surfaces shown in Fig. 20 illustrate the combined effects of the process factors and determine the optimal adsorption range. According to this study, optimal performance is achieved under intermediate conditions, specifically a slightly acidic pH (≈ 6.5) and a contact time of approximately 160 min. High qₑ values are favored by a moderate amount of adsorbent, and the reaction is significantly improved by an increase in the initial concentration due to an increased driving force. It is clear from the way these elements interact that process control is essential. The model identified the following ideal conditions: pH ≈ 6.5, adsorbent mass ≈ 0.1 g, initial concentration ≈ 25 mg·L⁻¹, and contact time ∼ 160 min. A maximum removal efficiency of almost 83% is attained under these circumstances, demonstrating the effectiveness of the response surface methodology approach, corresponding to an adsorption capacity of 20.8 mg·g⁻¹, The RSM model predicts a slightly better removal efficiency (83%) than the experimental value obtained from conventional parametric investigations (77%), which may be explained by optimising the combined effects of the operating variables. The RSM methodology considers interactions between factors, which makes it possible to identify more advantageous operating circumstances and enhance adsorption performance, in contrast to the method of changing one element at a time. Effect of ionic strength on the adsorption performance of B-CaHAp The effect of ionic strength on the adsorption of FLCX-Na by B-CaHAp was studied using NaCl as a background electrolyte under the optimized operating conditions. As shown in Fig. 24, the NaCl concentration increased only slightly, reducing the adsorption performance. The removal efficiency decreased from 83.0% in the absence of NaCl to 81.5%, 79.3%, and 76.8% for concentrations of 10, 50, and 100 mM, respectively, with adsorption capacities of 20.8, 20.4, 19.8, and 19.2 mg g⁻¹. Fig. 24Effect of ionic strength (NaCl concentration) on the removal efficiency of FLCX-Na by B-CaHAp at the optimized adsorption condition Effect of ionic strength (NaCl concentration) on the removal efficiency of FLCX-Na by B-CaHAp at the optimized adsorption condition The slight decrease in adsorption performance can be attributed to the increase in ionic strength, which partially weakens the electrostatic attraction between the FLCX-Na molecules and the B-CaHAp surface. However, the small decrease means that the electrostatic interactions are not the only controlling factors of the adsorption. Meanwhile, the surface complexation between Ca 2+ active sites and the deprotonated carboxylate groups of FLCX-Na and hydrogen bonding between phosphate and hydroxyl groups also played an important role in the adsorption process. Therefore, the adsorption efficiency was still higher than 75% at the maximum NaCl concentration, which suggested a good tolerance of B-CaHAp to the dissolved inorganic salts and supported the potential application of B-CaHAp for the treatment of saline pharmaceutical wastewaters [45]. Regeneration and reusability of B-CaHAp The regeneration and reusability of B-CaHAp were studied over five consecutive adsorption-desorption cycles under optimal operating conditions determined by the RSM methodology. Figure 25 shows that the adsorbent maintained high removal efficiency during the regeneration study, although this efficiency decreased slightly progressively with the number of cycles. Removal efficiency decreased from 83% in the first cycle to 76.1% after the fifth cycle, corresponding to a reduction in. adsorption capacity from 20.8 mg·g⁻¹ to 19 mg·g⁻¹. representing an overall loss of adsorption performance of less than 7%. Fig. 25Regeneration and reusability of B-CaHAp for FLCX-Na adsorption over five consecutive adsorption–desorption cycles under the Optimum operating conditions determined by RSM Regeneration and reusability of B-CaHAp for FLCX-Na adsorption over five consecutive adsorption–desorption cycles under the Optimum operating conditions determined by RSM The slight decrease in adsorption efficiency after several reuses can reasonably be attributed to the incomplete desorption of flucloxacillin sodium molecules, which are strongly bound to a limited number of active adsorption sites. Regeneration with a dilute NaOH solution effectively removed most of the adsorbed antibiotic, but a small fraction of molecules remained bound due to relatively strong interactions involving surface calcium ions, phosphate groups, and hydroxyl groups. As a result, with each adsorption-desorption cycle, the quantity of available adsorption sites progressively decreased. Hydroxyapatite-based adsorbents have been shown to exhibit similar behavior, which is typically ascribed to an intrinsic impact of numerous regenerations rather than structural degeneration. It is noteworthy that the modest drop in adsorption effectiveness indicates that B-CaHAp’s structure was mainly maintained during the regeneration process. This observation is fully consistent with the previously published physicochemical characterization. The apatite network appears to be highly stable under the mild alkaline regeneration conditions used, based on the excellent crystallinity found by XRD and the good thermal stability shown by TGA following calcination at 900 °C. Additionally, FTIR research revealed that the adsorption-causing phosphate and hydroxyl functional groups are structurally integrated within the apatite network, which explains why the material was able to maintain the majority of its adsorption capacity after multiple reuses. The adsorption mechanism suggested in this study is further supported by the outstanding regeneration behavior. The interaction between FLCX and B-CaHAp is strong enough to guarantee effective adsorption while remaining partially reversible in a slightly alkaline environment, as evidenced by the regeneration treatment’s ability to restore the majority of the adsorption performance without appreciable loss of efficiency. For real-world wastewater treatment applications, reversible adsorption would be beneficial since it permits the adsorbent to be reused again, lowering material consumption and operational expenses. Overall, the regeneration study demonstrated that B-CaHAp exhibits excellent operational stability and can be reused for at least five consecutive adsorption-desorption cycles with high removal efficiency. Conclusion In this study, a comparative evaluation of Syn-CaHAp and B-CaHAp for the removal of FLCX-Na from aqueous solutions was conducted. The experimental results demonstrated that B-CaHAp exhibited a better adsorption performance than Syn-CaHAp, with the maximum adsorption capacities of 26 mg.g− 1 and 15 mg.g⁻¹ for Syn-CaHAp, respectively. The physicochemical characterization revealed distinct structural features of the biogenic material, particularly carbonate substitution and calcium deficiency, which may account for its improved adsorption behavior. The adsorption was greatly affected by the operating parameters like pH, adsorbent dose, initial concentration, and contact time. Adsorption was favorable at a slightly acidic pH (~ 6–6.5) and was mediated by electrostatic attraction and calcium complexation of FLCX-Na. The Langmuir isotherm model provided the best fit for the equilibrium data, and the pseudo-second-order kinetic model proved to be in better agreement with the experimental results, indicating that the adsorption process takes place on heterogeneous surfaces and is mainly controlled by chemisorption. Response surface methodology (RSM) was successfully applied to optimize the adsorption process, considering the interactions between the operating variables. The optimized conditions (pH ≈ 6.5, adsorbent dosage ≈ 0.1 g, initial concentration ≈ 25 mg·L− 1, and contact time ≈ 160 min) achieved an increase in the maximum removal efficiency from 77% (26 mg·g⁻¹) under conventional experimental conditions to 83%, corresponding to an adsorption capacity of approximately 20.8 mg·g⁻¹, while the developed model exhibited excellent predictive capability (R2 = 0.91). In addition to the excellent adsorption performance, B-CaHAp maintained high removal efficiency with the increase of ionic strength, indicating that the adsorption process was only slightly influenced by the presence of dissolved inorganic salts. This finding emphasizes its potential applicability for the treatment of pharmaceutical-contaminated waters under more realistic saline conditions. In addition, the regeneration experiment showed that B-CaHAp still had high adsorption efficiency after five successive adsorption–desorption cycles, and the removal efficiency slightly decreased from 83% to 76.1%, which indicated good operational stability and reusability. The present results further highlight the potential of fish bone derived hydroxyapatite as an efficient, low-cost, and reusable adsorbent for the removal of antibiotic contaminants from aqueous media. This approach of turning fishery waste into a value-added functional material for water remediation is completely consistent with the principles of sustainable development and circular economy. Future work will concentrate on the evaluation of the performance of regenerated B-CaHAp in real wastewater matrices including multiple pharmaceutical compounds and competing ions, its long-term stability over a larger number of adsorption–desorption cycles and the validation of its applicability in continuous-flow or pilot-scale treatment systems.