Characterization and modeling of kidney bean shell biochar as adsorbent for caffeine removal from aquatic environments

2.1 Materials

Caffeine (C₈H₁₀FN₄O₂, 99% purity), potassium dihydrogen phosphate (KH₂PO₄, 99.99% purity), sodium hydroxide (NaOH, 98% purity), hydrochloric acid (HCl, 37% purity), orthophosphoric acid (H₃PO₄, ≥85% purity), ethanol (C₂H₅OH), and acetonitrile (CH₃CN, 99.9% purity) were purchased from Sigma-Aldrich and used without further purification.

A 100 mL caffeine stock solution with a concentration of 500 mg/L was initially prepared using deionized water as the solvent. Standard solutions with concentrations of 5, 10, 15, 20, 25, and 30 mg/L were prepared from the stock solution and used for calibration in the development of the high-performance liquid chromatography (HPLC) analysis method. For the experimental studies, 50 mL solutions with an initial caffeine concentration of 50 mg/L were prepared. pH adjustments were made using analytical-grade NaOH and HCl solutions (Merck). The bean husk used in this study was collected from a local garden in Zonguldak, Turkey.

2.2 Adsorbent preparation

Kidney bean shells were initially washed with tap water and subsequently with distilled water to remove dust and other impurities. Following the washing process, they were dried in a dark and enclosed room for a week, then ground and sieved (particle size range <1.00 mm). The ground kidney bean shells underwent pyrolysis at 550°C for 30 min, with a heating rate of 10°C/min under an N₂ gas flow rate of 0.5 L/min. A mixture was prepared using the biochar obtained from pyrolysis at a mass ratio of 3:1 (KOH/biochar), and this mixture was allowed to stand for 24 h. After the 24-h standing period, the sample was stirred, dried at room temperature, and then at 105°C. The completely dried samples were then subjected to carbonization in a furnace at 800°C for 1 h, with a heating rate of 10°C/min under an N₂ flow rate of 1 L/min. The nitrogen gas flow was continued until the temperature dropped to 25°C. After thorough cleaning with distilled water three times, the activated carbon was immersed in 1 M HCl and subsequently washed once more with deionized water and ethanol to remove chloride ions before being allowed to dry naturally. It was then dried in an oven at 105°C and stored for characterization and adsorption experiments.

2.3 Adsorbent characterization

The synthesized adsorbent was characterized using a range of analytical techniques. The functional groups present on the surface of the biochar were identified using Fourier transform infrared (FTIR) spectroscopy with a Bruker ALPHA instrument.

The specific surface area (SSA) of the activated carbon was measured using the Brunauer–Emmett–Teller (BET) method with a Micromeritics Gemini 2360 analyzer. For this analysis, 1 g of the sample was used, pretreated under a nitrogen atmosphere at 250°C, and the dry mass was determined using an analytical balance. High-purity nitrogen served as the adsorbate gas during the measurements.

The surface morphology and elemental composition of the activated carbon were examined by scanning electron microscopy (SEM) using a LEO 1430 VP model SEM. Additionally, FTIR analysis (Bruker ALPHA model) was employed again to further confirm the functional groups in the structure of the activated carbon, with spectra recorded in the range of 400–4,000 cm⁻¹. Additionally, the point of zero charge (pHpzc) of the adsorbent surface was determined by adjusting the pH of 50 mL of 0.1 M NaCl solution between 2 and 12 and then adding 50 mg of adsorbent. The mixture was continuously shaken for 24 h. The final pH value of the solution was measured, and pHpzc was reached when pH_initial − pH_final = 0.

2.4 Adsorption studies

To investigate the effects of adsorption parameters (pH, initial concentration, adsorbent dosage, contact time, and temperature) and to determine the optimum adsorption conditions, experiments were conducted in a batch mode with three replicates, where one variable was changed while keeping the other parameters constant. All experiments were performed using Erlenmeyer flasks, each containing 50 mL of caffeine solution and a water bath shaker (WSB-30, WITEG, Germany) operating at 180 rpm.

Adsorption studies were carried out using different adsorbent amounts (0.020, 0.040, 0.060, 0.080, and 0.100 g), different pH levels (4, 7, 11), and different temperatures (295, 305, and 315 K), while keeping the initial caffeine solution concentration constant at 50 mg/L.

The reusability of the synthesized biochar was evaluated through four successive adsorption–desorption cycles. For this, the same amount of adsorbent was used in each cycle. After each adsorption run, the spent adsorbent was soaked in a 20% ethanol solution, treated with ultrasound for 1 h, rinsed with deionized water, filtered, and then oven-dried for reuse.

The caffeine concentrations in the solutions before and after the adsorption process were measured using an HPLC system (Agilent 1260 Series, USA) equipped with a UV detector. Caffeine detection was carried out at a wavelength of 275 nm. A C18 chromatographic column (Waters Symmetry®, 150 × 4.6 mm, 5 μm) was employed for the separation process. The mobile phase consisted of a dihydrogen phosphate buffer and acetonitrile mixture (87.50: 12.50, v/v), adjusted to pH 2.2, with a flow rate of 1 mL/min.

The equilibrium adsorption capacity (q _e) and the percentage of caffeine removal were determined using equations (1) and (2), respectively:

Here, C _o represents the initial caffeine concentration (mg/L), C _e represents the equilibrium caffeine concentration (mg/L), V represents the volume of the solution (L), and m represents the mass of the adsorbent (g).

2.5 Adsorption isotherms

Adsorption isotherms are mathematical models that illustrate how adsorbate molecules are distributed between the liquid phase and the surface of the adsorbent. These models are essential for understanding the adsorption mechanism, as they provide insight into the interactions between the adsorbent and adsorbate at a constant temperature [49]. Numerous isotherm models have been proposed in the literature to describe adsorption behavior. Among the most widely applied are the Langmuir, Freundlich, Temkin, and Dubinin–Radushkevich isotherms [50,51,52,53]. The Langmuir isotherm is based on the assumption that adsorption occurs in a monolayer on a homogeneous surface, with uniform adsorption energy across the surface. As the initial concentration of the adsorbate increases, adsorption also increases until a saturation point is reached. The linear form of the Langmuir isotherm equation is given as follows [54,55]:

where K _L (L/g) is the Langmuir constant related to adsorption energy, q _max (mg/g) is the maximum adsorption capacity of the adsorbent, C _e (mg/L) is the adsorbate concentration in the solution after equilibrium, and q _e (mg/g) is the amount of the adsorbate adsorbed per unit mass of the adsorbent.

In the Langmuir isotherm, the spontaneity of the adsorption process is evaluated using the R _L (separation factor) parameter:

where C ₀ (mg/L) represents the initial concentration of the adsorbate, and K _L represents the Langmuir constant. An R _L value in the range 0 < R _L < 1 indicates favorable (spontaneous) adsorption, R _L > 1 suggests an unfavorable process, R _L = 1 corresponds to linear adsorption, and R _L = 0 signifies irreversible adsorption. Thus, the R _L parameter plays a critical role in interpreting the thermodynamic feasibility and behavior of the adsorption process.

The Freundlich isotherm describes multilayer adsorption on heterogeneous surfaces. The adsorbate concentration on the adsorption surface changes proportionally to the increase of the adsorbate concentration in the solution. The linear form of the Freundlich isotherm equation is expressed as follows [56]:

Here, q _e (mg/g) is the amount of the adsorbed substance at equilibrium, K _F is the experimental constant indicating adsorption capacity, n is the experimental constant determining heterogeneity, and C _e (mg/L) is the adsorbate concentration remaining in the solution after equilibrium.

The 1/n value varies between 0 and 1. As 1/n approaches zero, the surface becomes more heterogeneous. The Temkin isotherm model assumes that adsorption is homogeneously distributed on the adsorbent surface and that adsorption energies decrease linearly and not exponentially, as in the Freundlich model. The linear form of the Temkin isotherm equation is as follows [57]:

Here, B = RT/b _T (J/mol) is the Temkin isotherm constant, A _T (L/g) is the equilibrium binding constant, R (8.314 J/mol K) is the gas constant, and T (K) is the absolute temperature.

The Dubinin–Radushkevich isotherm model describes an adsorption process that initially follows a volume-filling mechanism. It is applicable to both homogeneous and heterogeneous surfaces. The linear equation of the model is expressed as follows [58]:

where q e represents the equilibrium adsorption capacity (mol/g), q _m denotes the maximum adsorption capacity (mol/g), K _DR is the activity coefficient associated with the adsorption energy (mol²/J²), and Ɛ refers to the Polanyi potential.

2.6 Adsorption kinetics

Adsorption kinetics is examined using models that are used to understand the adsorption mechanism, mass transfer, and rate-controlling steps [59]. In this study, four different kinetic models were applied: pseudo-first-order (PFO), pseudo-second-order (PSO), Elovich, and intraparticle diffusion (IPD) model.

The PFO kinetic model, developed by Lagergren, explains the adsorption mechanism depending on the adsorbent capacity. The linear form of the model is expressed as follows [60]:

Here, q _e (mg/g) is the amount of the adsorbed substance at equilibrium, q _t (mg/g)is the adsorption amount at time t, k ₁ (min⁻¹) is the rate constant of the PFO kinetic model, and t (min) is the time.

This model is generally in good agreement with the experimental data for the first 30 min of adsorption but is not in good agreement with the experimental data for all ranges of contact time [60].

The PSO kinetic model derived by Ho and McKay is related to the adsorption capacity and is a valid model for the whole adsorption process [60]. Its linear form is given as follows:

Here, k ₂ (g/mg min) is the rate constant of the PSO kinetic model, q _e (mg/g) is the equilibrium adsorption capacity, q _t(mg/g) is the adsorption amount at time t, and t (min) is the time.

This model shows that adsorption is chemically controlled and adsorbate–adsorbent interactions are important.

The IPD model describes the transport of adsorbate from the solution to the adsorbent surface and its diffusion within the particle. Its linear form is expressed as follows [61]:

Here, q _t (mg/g) is the adsorption amount at time t, K _p (mg/g min^1/2) is the intra-particle diffusion model constant, t (min) is the time, and C (mg/g) is a constant referring to the boundary layer thickness.

The larger the value of C, which represents the boundary layer thickness, the more limited the adsorption is by surface control. Small C values indicate that diffusion is the main determinant of the process.

2.7 Adsorption thermodynamics

The thermodynamics of the adsorption process is used to understand the energy changes that occur in an adsorption process and explain the adsorption mechanism, whether it is spontaneous or not, and its response to temperature changes. In this context, thermodynamic parameters such as enthalpy change (ΔH ⁰), entropy change (ΔS ⁰), and Gibbs free energy change (ΔG ⁰) are determined. These parameters help to explain the adsorption mechanism and the nature of the interactions between adsorbent and adsorbate. The thermodynamic parameters are determined using equations (13) and (14):

where K _c is the equilibrium constant, expressed in q _e/C _e; R is the universal gas constant (=8.314 J/mol/K); and T is the temperature, expressed in Kelvin [51].

3.1 Biochar characterization

3.1.1 SEM

SEM is a widely employed technique for surface morphology analysis. Figure 1a and b illustrates the SEM images of kidney bean biochar (BFB), revealing a homogeneous distribution of pores with varying diameters across the biochar surface. These pores are likely formed due to the release of volatile compounds, such as H₂, CO, and CH₄, during pyrolysis [52]. The observed variations in pore volume can be attributed to differences in the size and proportion of volatile components within the material’s structure. Additionally, energy-dispersive X-ray spectroscopy (EDS) mapping was conducted to determine the chemical composition of BFB, as shown in Figure 1c. The results indicate that carbon and oxygen constitute the primary framework of the prepared material.

Figure 1

Surface morphology of kidney bean biochar: (a) 2 µm, (b) 20 µm, and (c) EDX analysis.

3.1.2 FTIR spectroscopy

To determine the structural composition of BFB, FTIR analysis was performed, and the resulting spectrum is illustrated in Figure 2. The observed peak is shown at around 3,662 cm⁻¹, which is related to the vibrations of the –OH stretching vibration in alcohol or phenol and the vibrations of the H₂O molecule. The observed peaks at about 2,977–2,891 cm⁻¹ can be attributed to the C–H stretching of aliphatic groups. The peaks observed at 1,600–1,550 cm⁻¹ are attributed to aromatic C═C ring stretching indication of benzene-like rings . Finally, the peak observed at 1,061 cm⁻¹ can be attributed to symmetric C–O–C stretching for cellulose, hemicelluloses, and lignin. C–O stretching in acids, alcohols, phenols, and esters appeared between 1,000 and 1,200 cm⁻¹ [62].

Figure 2

FT-IR spectrum of BFB.

3.1.3 Analysis of structural characteristics

The structural quality of an adsorbent is critical when determining its maximum sorption capacity and the related process [63]. If the main mechanism of adsorption is pore filling, an adsorbent with a higher SSA and total pore volume usually exhibits better sorption ability of pharmaceuticals [64,65]. The structural properties (SSA and pore volume) of BFB were calculated using nitrogen adsorption–desorption isotherms at 77 K (Figure 3).

Figure 3

Adsorption–desorption isotherms of BFB.

As illustrated in Figure 3, the adsorption–desorption isotherms of BFB correspond to Type IV, characterized by hysteresis loops due to capillary condensation, which is indicative of a mesoporous structure [64]. The associated structural parameters are summarized in Table 1. Based on these properties, the SSA of BFB was determined to be 506.61 m²/g.

Table 1

Textural properties of BFB

Adsorbent	BET surface area (m2/g)	BJH adsorption cumulative pore volume (cm3/g)	Average pore diameter (nm)
Biochar derived from kidney beans (BFB)	506.61	0.1890	2.86

BFB exhibited a pore volume of 0.1890 cm³/g and an average pore diameter of 2.86 nm, confirming its microporous structure. These results demonstrate that BFB’s enhanced adsorption performance stems from physical adsorption facilitated by micropores, enabling uniform distribution of caffeine molecules on BFB’s surface. Increased interaction probability between caffeine and active sites due to optimal pore accessibility. Given caffeine’s molecular dimensions, the 2.86 nm pore diameter of BFB is ideally suited for efficient caffeine uptake. This alignment promotes pore-filling mechanisms within BFB’s primary pore channels, maximizing adsorption capacity. Notably, BFB’s elevated porosity and pore size distribution indicate a positive correlation with caffeine uptake capacity, underscoring porosity as a critical determinant of BFB’s adsorption efficacy.

3.2 Adsorption studies

3.2.2 Effect of solution pH on adsorption performance

The pH of the aqueous solution significantly influenced the adsorption efficiency of caffeine onto the biochar-based adsorbent (BFB) by modulating the surface charge of the adsorbent and the ionization state of caffeine molecules. As illustrated in Figure 4c, the adsorption efficiency followed a pH-dependent trend, with optimal performance observed at neutral pH.

Figure 4

(a) Determination of pHzpc using the pH shift method, (b) effect of adsorbent dose on caffeine removal, (c) effect of solution pH, and (d) effect of contact time.

At pH 4 (acidic conditions), the removal efficiency reached 95.86%, attributable to the protonation of caffeine (pH < pK _a) and electrostatic attraction between the positively charged nitrogen in caffeine and the adsorbent surface, which remains near-neutral at this pH (below the pHPZC of 7.80, Figure 4a) [59,60]. The highest efficiency (98.10%) was achieved at pH 7 (neutral), where the balance between caffeine’s protonation state and the adsorbent’s slightly negative surface charge (approaching pHPZC) minimized repulsive forces while maximizing adsorption affinity [61].

Under alkaline conditions (pH 10), the efficiency declined to 93.22%, likely due to the deprotonation of caffeine and the increasingly negative surface charge of BFB (pH > pHPZC, Figure 4a), resulting in weaker electrostatic interactions or potential repulsion [62,63]. This trend aligns with the molecular structure of caffeine, where the high dipole moment enhances its sensitivity to polarity changes in the medium [69].

The selection of pH 7 as the optimal condition not only maximizes adsorption but also aligns with the typical pH range of industrial wastewater, ensuring practical applicability without requiring extensive pH adjustment. These findings underscore the dominance of electrostatic interactions in the adsorption mechanism, as corroborated by the pHPZC analysis (Figure 4a) and prior studies on polar organic compounds [70,71,72,73].

Figure 4c confirms that neutral pH provides the highest caffeine removal efficiency, while deviations to extreme pH values reduce performance. This behavior highlights the critical role of pH control in water treatment processes targeting caffeine removal.

3.3 Modeling of adsorption kinetics

In this study, the caffeine adsorption kinetics from aqueous solutions onto the bean shell-derived adsorbent were investigated using four distinct models: PFO, PSO, Elovich, and IPD. The suitability of these models was compared to elucidate the dynamics of the adsorption mechanism. The fitted curves are presented in Figure 5, and the corresponding results are summarized in Table 2.

Figure 5

Adsorption kinetics modeling of caffeine adsorption onto BFB.

The PFO kinetic model demonstrated a poor fit to the experimental data, evidenced by a significant discrepancy between the experimental qₑ value and the calculated qₑ (15.24 mg/g), along with a relatively low correlation coefficient (R² = 0.988). This discrepancy suggests the limitations of the model and highlights potential physical adsorption processes.

Conversely, the PSO kinetic model emerged as the most suitable, exhibiting a high correlation coefficient (R² = 0.999) and a close agreement between the experimental equilibrium capacity (38.33 mg/g) and the calculated value (39.68 mg/g). This finding implies that the adsorption process is primarily governed by chemical interactions, such as ionic bonding or electron transfer.

The Elovich model, while accounting for heterogeneous surface interactions (R ² = 0.910), showed a comparatively lower fit than the other models. Nevertheless, the parameters α (0.55 mg/g min) and β (4.93 g/mg) provided insights into the rapid initial adsorption rate and the effectiveness of surface coverage.

The IPD model revealed a multi-step adsorption mechanism. The high kₚ value (7.98 mg/g min^½) in the initial phase (0–15 min) indicated that surface adsorption was dominant. The moderate kₚ value (1.78 mg/g min^½) in the second phase (15–60 min) and the low kₚ value (0.26 mg/g min^½) in the third phase (60–150 min) suggested that IPD became the limiting factor. The non-zero C parameter further supported the significant role of boundary layer resistance.

Overall, the superior performance of the PSO model confirmed that caffeine adsorption was primarily controlled by chemical interactions, while the IPD model elucidated the multi-step nature of the process. These findings underscore the effectiveness of bean shell-based adsorbents in caffeine removal and highlight the importance of kinetic models for process optimization. Future studies should focus on investigating the kinetic behavior under various conditions and with surface-modified adsorbents. The agreement of the experimental data with the PSO kinetic model suggests that the rate-limiting step in the biosorption processes on BFB is the chemical interaction (which may involve ion exchange and/or electron sharing) between the caffeine molecule from the aqueous solution and the superficial functional groups of the biosorbent [74]. On the other hand, when the values of the rate constants are compared, it was found that caffeine molecules have higher affinity to the functional groups of the biosorbent, and the biosorption efficiency will be the highest.

3.4 Modeling of adsorption isotherms

In this study, the adsorption of caffeine from aqueous solutions onto bean shell-derived adsorbent was analyzed using four distinct isotherm models: Langmuir, Freundlich, Temkin, and Dubinin–Radushkevich. The suitability of these models was compared to elucidate the adsorption mechanism and evaluate the adsorbent’s performance. The non-linear isotherm model fits are illustrated in Figure 6, and the findings are presented in Table 3.

Figure 6

Adsorption isotherm modeling of olaparib adsorption onto BFB.

The Langmuir isotherm model was determined to be the most suitable, exhibiting a high correlation coefficient (R² = 0.999) and physically meaningful parameters. The maximum adsorption capacity (q _max = 40.32 mg/g) predicted by the model was consistent with the experimental data, indicating that adsorption occurred as a monolayer on a homogeneous surface. Furthermore, the dimensionless separation factor (R _L) value, ranging between 0 and 1, supported the feasibility of the adsorption process.

The Freundlich model, representing heterogeneous surface interactions, revealed a high K _F value (28.43 (mg/g)(L/mg)^(1/n)), suggesting a strong affinity of the adsorbent for caffeine. The heterogeneity factor (n = 11.68), exceeding 1, indicated that the adsorption was physical and favorable. However, the model exhibited a comparatively lower correlation coefficient (R² = 0.964), indicating a limited fit.

The Temkin model suggested that physical adsorption was dominant, as evidenced by the low heat of adsorption (b = 985.5 J/mol). Nevertheless, the model’s lower correlation coefficient (R² = 0.942) compared to other models indicated a limited fit.

The Dubinin–Radushkevich (D-R) model (R² = 0.891) indicated that the adsorption was a low-energy physical process (E = 0.67 kJ/mol), consistent with a microporous structure. However, this model also exhibited a lower correlation coefficient compared to the others, suggesting a limited fit.

Overall, the superior fit of the Langmuir model confirmed that monolayer adsorption was the dominant mechanism for caffeine adsorption onto the bean shell adsorbent. The other models provided complementary information regarding heterogeneous surface interactions and low-energy physical adsorption. These findings suggest that bean shell-based adsorbents are an effective option for removing contaminants like caffeine, and they provide recommendations for process optimization and adsorbent modification in future studies.

3.5 Thermodynamic evaluation

Since the interpretations made with a small amount of experimental data in thermodynamic studies may be incorrect, the amount of temperature data obtained from the experiments should be increased [75]. Therefore, the thermodynamic behavior of the adsorption of caffeine onto the adsorbent obtained from bean shells was investigated at five different temperatures (290, 300, 305, 310, and 315 K) to explain the spontaneity, energy changes, and mechanism of the process. The calculated thermodynamic parameters – Gibbs free energy change (ΔG ⁰), enthalpy change (ΔH ⁰), and entropy change (ΔS ⁰) – are summarized in Table 4.

The negative ΔG ⁰ values at all temperatures (−10.34 to −11.79 kJ/mol) confirm the spontaneous nature of caffeine adsorption. The increasing negative ΔG ⁰ with increasing temperature (290 → 315 K) indicates that the favourability of adsorption increases at higher temperatures [76]. The magnitude of ΔG ⁰ (<−20 kJ/mol) indicates a physisorption mechanism dominated by weak van der Waals forces or electrostatic interactions [77,78]. The negative ΔH ⁰ value (−5.22 kJ/mol) suggests an exothermic process and supports the observed temperature-dependent efficiency decrease. The positive ΔS ⁰ (17.65 kJ/mol) reflects greater randomness at the solid–liquid interface during adsorption. This entropy-driven behavior suggests that caffeine molecules adopt a less ordered arrangement on the adsorbent surface, possibly due to solvent displacement or multilayer adsorption [79]. Thermodynamic analysis shows that caffeine adsorption onto the bean shell-derived adsorbent occurs:

Spontaneous (ΔG ⁰ < 0),

Exothermic (ΔH ⁰ < 0), and

Entropy-driven (ΔS ⁰ > 0),

with physiosorption as the dominant mechanism. These findings are in agreement with previous studies on biochar-based adsorbents [77,78,79,80] and emphasize the potential of this low-cost material for organic pollutant removal.

3.6 Adsorption mechanism

Biochar derived from phaseolus vulgaris pod husks (BFB) is an effective material for caffeine adsorption. The biochar possesses a micro- and mesoporous structure, facilitating the retention of small molecules such as caffeine on its surface. Typically exhibiting a surface area ranging from 100 to 1,000 m²/g, it provides numerous adsorption sites. The biochar surface contains oxygen-containing functional groups, such as carboxyl (–COOH), hydroxyl (–OH), phenolic, and lactone groups, which form physical and chemical interactions with caffeine. Additionally, the carbon-based hydrophobic structure of biochar promotes hydrophobic interactions with organic molecules. BFB, with its high cellulose, hemicellulose, and lignin content, is well-suited for biochar production. During pyrolysis, these components carbonize, forming a porous structure [81]

Caffeine (C₈H₁₀N₄O₂), a methylxanthine alkaloid, is characterized by a heterocyclic structure. Containing nitrogen and oxygen atoms, this structure enables polar interactions. The aromatic purine ring contributes to hydrophobic interactions, while caffeine’s nitrogen atoms confer weak basic properties, allowing protonation depending on pH. With a molecular size of approximately 0.6–0.8 nm, caffeine can easily penetrate the micro- and mesopores of biochar [82]

The adsorption of caffeine onto biochar results from a combination of physical and chemical interactions. The primary mechanisms are outlined below.

Physical adsorption (van der Waals forces): Caffeine molecules are physically attracted to the porous structure of biochar through van der Waals forces. The micro- and mesopores trap caffeine molecules, enhancing adsorption capacity. This low-energy, reversible process is influenced by environmental factors such as temperature and concentration.

Hydrophobic interactions: The aromatic purine ring of caffeine interacts with the hydrophobic carbon surfaces of biochar. This interaction promotes the orientation of caffeine toward the biochar surface, particularly in aqueous solutions. The carbonized structure of BFB-derived biochar provides a suitable environment for hydrophobic interactions.

π–π interactions: The graphitic structure of biochar contains aromatic rings that form π–π stacking interactions with the purine ring of caffeine. These interactions enable strong binding of caffeine to the biochar surface, increasing adsorption efficiency.

Hydrogen bonding: Oxygen-containing functional groups (–OH, –COOH) on the biochar surface form hydrogen bonds with nitrogen or oxygen atoms of caffeine. For instance, the carbonyl groups (C═O) of caffeine can form hydrogen bonds with the hydroxyl groups of biochar.

Electrostatic interactions: The pH of the solution affects the charge states of both the caffeine and biochar surfaces. At low pH, caffeine may become protonated, adopting a cationic form. Negatively charged functional groups, such as carboxylate ions, on the biochar surface, can form electrostatic attractions with protonated caffeine. However, the surface chemistry of BFB-derived biochar varies depending on pyrolysis conditions; at higher pH, the surface becomes more negatively charged, influencing electrostatic interactions [83].

These mechanisms collectively enable the high efficiency of BFB biochar in caffeine adsorption. The adsorption process depends on both the surface properties of the biochar and the environmental conditions of the solution.

3.7 Reusability of BFB adsorbent

The reusability of the biochar-based adsorbent (BFB) was evaluated through five consecutive adsorption–desorption cycles to assess its practical applicability and economic viability. The experimental procedure consisted of the following steps: caffeine adsorption was conducted under optimized conditions (pH 7, 50 mg/L initial concentration, 60 mg adsorbent dose, and room temperature). The caffeine-saturated adsorbent was filtered and separated from the solution. Ethanol (50 mL) was used as the desorption solvent due to its high caffeine solubility, low boiling point, and ease of removal. The adsorbent was washed at 150 rpm for 10 min. The regenerated adsorbent was dried at 80°C until a constant weight was achieved. The adsorption–desorption process was repeated five times, with removal efficiency recorded for each cycle. The reusability performance of BFB is summarized in Table 5.

The fresh BFB adsorbent exhibited a high removal efficiency of 96.50%, demonstrating its effectiveness under optimized conditions. A gradual decrease in efficiency was observed over successive cycles, with 94.72% (Cycle 1), 93.10% (Cycle 2), 90.70% (Cycle 3), and 80.36% (Cycle 4). The adsorbent maintained > 90% efficiency for three cycles, indicating robust reusability. By the fifth cycle, efficiency dropped to 80.36%, likely due to partial pore blockage by residual caffeine or ethanol impurities and structural degradation of active sites after repeated regeneration. The use of ethanol as a desorption solvent proved effective due to its ability to dissolve caffeine without severely damaging the adsorbent structure. The drying step (80°C) ensured complete solvent removal, preventing interference in subsequent cycles.

BFB’s ability to retain high efficiency (>90%) for multiple cycles reduces operational costs in wastewater treatment. The adsorbent’s reusability aligns with circular economy principles by minimizing waste generation. Future studies could explore alternative regeneration methods (e.g., thermal or chemical treatment) to enhance longevity beyond five cycles. BFB demonstrates excellent reusability, with >90% efficiency retained for three cycles and >80% even after five cycles. While a gradual decline occurs, the adsorbent remains viable for practical applications, offering an economical and sustainable solution for caffeine removal. Further optimization of regeneration protocols could extend its lifespan. Table 6 lists recent studies on adsorption for caffeine removal from aqueous solutions.