Efficient treatment of acetaminophen-contaminated wastewater by a phenylboronic acid-functionalized magnetic expanded graphite nanocomposite

Abstract

Graphite-based adsorbents provide a sustainable and eco-friendly alternative to conventional metal-based adsorbents, yet their performance in removing pharmaceutical pollutants remains underexplored. In this study, a phenylboronic acid-functionalized magnetic expanded graphite (EG/Fe₃O₄-PBA) nanocomposite was developed as a novel adsorbent for the treatment of acetaminophen (ACT)-contaminated hospital wastewater. The functionalization endowed the composite with a porous structure, rapid magnetic separation, and strong affinity toward ACT molecules. The experiments were designed using a central composite design (CCD) and analyzed with response surface methodology (RSM) to assess the combined influence of operational parameters on adsorption performance and to optimize conditions for maximizing ACT removal efficiency. Adsorption behavior was systematically investigated through isotherm, kinetic, and thermodynamic studies. The equilibrium data fitted well to the Langmuir model, confirming monolayer adsorption with a high maximum capacity of 451.30 mg g⁻¹ achieved within 20 min. Kinetic analysis revealed that the process followed the pseudo-second-order model, highlighting chemisorption as the main mechanism. Thermodynamic evaluation showed negative Gibbs free energy values (–1.21 to − 15.67 kJ mol⁻¹), a positive enthalpy change (100.6 kJ mol⁻¹), and a positive entropy change (360.3 J mol⁻¹ K⁻¹), confirming that adsorption was feasible, spontaneous, and endothermic. The nanocomposite also exhibited excellent reusability, maintaining 80.5% efficiency over seven adsorption-desorption cycles with only a minor decline from the initial 98.8%. These findings demonstrate that EG/Fe₃O₄-PBA combines high adsorption capacity, rapid removal efficiency, and good recyclability with environmental compatibility. This work highlights the significance of functionalized graphite-based nanocomposites as a promising and sustainable solution for the treatment of pharmaceutical-contaminated wastewater.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-24309-9.

Introduction

Acetaminophen (ACT), often known as paracetamol, is one of the most consumed non-narcotic pharmaceuticals, extensively used to treat fever and mild to moderate pain¹. ACT is a medication with a water solubility of 12.900 mg L⁻¹, which is used to treat aches, such as colds, joint and muscular discomfort, fever, and in rare cases, peripheral nerve problems worldwide². According to studies, the body eliminates 58–68% of acetaminophen after consumption³. As a result, there is a substantial quantity of ACT in surface and subsurface water, urban effluent, and the pharmaceutical industry. In recent years, numerous unique chemical, physical, and biological approaches, including membrane filtration, biological degradation, adsorption, coagulation, oxidation processes, etc., have been used to remove ACT residues in water and wastewater treatment processes^4–6. Among these approaches, adsorption is one of the most promising methods widely utilized to remove different organic and inorganic contaminants from aqueous solutions due to its simplicity, efficiency, and low cost^7–9. To date, many adsorbent materials have been proposed and applied for their ability to remove ACT residues from contaminated water. Expanded graphite (EG) is a developed two-dimensional nanostructured modified graphite, which has drawn much attention due to its excellent properties, including large specific surface area, low density, multi-porosity, and hydrophobic nature¹⁰. Although many advantages of EG have been found for the adsorption of organic molecules from wastewater, the existence of some drawbacks, such as its low adsorption efficiency, small particle size, low-density, and high dispersion, causes difficulties in its collection and regeneration, which limit its application in a large-scale water environment¹¹. In recent years, the incorporation of EG with inorganic and organic components has emerged as a promising strategy to overcome the limitations of pristine EG in water treatment applications. Among these strategies, the integration of EG with magnetic particles to form magnetic expanded graphite (MEG) has gained considerable attention as an innovative approach that not only enables rapid magnetic separation from aqueous media but also enhances the functional applicability of EG. For instance, Ding et al.^12,13 synthesized a superparamagnetic MEG composite by depositing the cobalt ferrite (CoFe₂O₄) nanoparticles on EG, which was applied to remove oil leakage. Wojciech et al.^12,14 synthesized a magnetic Fe@graphite nanocomposite, and it was used for the adsorption of Acid Red 88 and Direct Orange 26 anionic dyes from aqueous solutions. Damasceno et al.¹⁵ studied the potential of magnetic graphite nanocomposite (GR-Fe₃O₄) for the adsorption of reactive black 5 (RB5) dye from aqueous solution. Tian et al.^12,16 prepared the magnetic expanded graphite composites (Fe₃O₄/MEG), and their adsorption properties toward ionic dyes were studied. Li et al.¹⁷ prepared a nano-graphite/Fe₃O₄ composite (NG/FC), and its potential for the removal of methyl violet (MV) from aqueous solution was evaluated. Collectively, these studies highlight the versatility of magnetic graphite-based composites in addressing diverse classes of organic pollutants and provide a foundation for further functionalization strategies to expand their environmental applications. In recent decades, a wide range of functional organic materials has been engineered and employed for the chemical modification of adsorbents, with the primary goal of enhancing their performance, particularly by increasing adsorption capacity and broadening their applicability¹⁸. Boronic acid is a moderately weak Lewis acid, and there are more than 1500 commercially available boronic acid ligands and derivatives, including phenylboronic acids (PBA) with different substituent groups¹⁹. As a chemical functionalizing agent, PBA has various advantages, including low cost, small size, high stability, and easy chemical modification. Due to the characteristics of PBA, the materials functionalized with PBA can have different uses in different fields²⁰. All the advantages mentioned aside, the functionalization of adsorbents with PBA has received little attention, and further efforts are required to develop its use on solid supports. However, no systematic research has been conducted to create phenylboronic acid-functionalized adsorbents.

This work proposes a phenylboronic acid-functionalized magnetic expanded graphite (EG/Fe₃O₄-PBA) nanocomposite as an advanced adsorbent for the efficient removal of ACT molecules from contaminated water. The structural and functional features of the as-prepared adsorbent were investigated in detail, and the adsorption process was systematically optimized via response surface methodology (RSM) using central composite design (CCD). Adsorption behavior (isotherm, kinetic, thermodynamic) and adsorption–desorption (reusability) cycles were studied. Compared with many recent adsorbents, EG/Fe₃O₄-PBA is expected to show improved adsorption capacity, faster removal kinetics, easier separation (magnetic), and excellent reusability. By evaluating its performance in actual water samples (river and hospital wastewater) in addition to laboratory solutions, this study aims to demonstrate not only novelty but also practical relevance and superior efficiency over many conventional and recently reported protocols.

Experimental

Materials

Natural graphite flakes (99% carbon basis), sulfuric acid (H₂SO₄, 98%), nitric acid (HNO₃, ≥ 90.0%), ammonium ferrous sulfate hexahydrate (NH₄)₂Fe(SO₄)₂·6H₂O), hexamethylenetetramine (HMTA, C₆H₁₂N₄, ≥ 99%), sodium sulfate (Na₂SO₄, ≥ 99.0%), ethanol (CH₃CH₂OH, 99.8%), 4-Formylbenzoic acid (C₇H₇BO₃, 97%), sodium cyanoborohydride (CH₃BNNa, 95%), hydrochloric acid (HCl, 37%), and sodium hydroxide (NaOH, 98%) were purchased from Sigma-Aldrich, and used without purification.

Synthesis methods

Synthesis of the expanded graphite (EG)

Expanded graphite was synthesized by combining chemical oxidation and thermal treatment methods, as follows: Natural graphite flakes (˂2 μm) were first dried at 75 °C in a vacuum oven for 8 h to remove the moisture content. Subsequently, the dried graphite flakes were mixed with a saturated acid consisting of sulfuric and concentrated nitric acids (in volume ratios of 3:1) for 12 h to form the graphite intercalated compound (GIC). During the reaction, nitric acid is an oxidizer, and sulfuric acid is an intercalant. Afterward, the mixture was stirred occasionally to obtain uniform intercalation of each flake. The chemically treated flakes were then thoroughly rinsed with ultrapure water until the pH level of the solution reached about 7.0, and then the washed flakes were dried at 60 °C for 5 h in a vacuum oven. Finally, the chemically treated flakes were exposed to high temperatures of 800–900 °C in a muffle furnace, and the expanded graphite (EG) flakes were formed²¹.

Synthesis of the magnetic expanded graphite (EG/Fe₃O₄)

In a typical procedure, a certain amount of the as-prepared expanded graphite (4 g) was first sonicated for 20 min. After that, specific amounts of the ammonium ferrous sulfate (5 mmol), hexamethylenetetramine (2.5 mmol), and sodium sulfate (1 g) were dissolved in 30 mL of ultrapure water. The above mixture was stirred for 20 min at room temperature, and a homogeneous dark-green suspension formed. Afterward, the prepared suspension was transferred into a 50 mL Teflon-lined autoclave and heated at 90 °C for 24 h in an oven. Then, the autoclave was cooled in air after the heating process, and the resulting black precipitates were washed with ethanol and ultrapure water, respectively. In the end, the washed precipitates were dried in an oven at 60 °C for 24 h, and the resulting black powder was labeled as EG/Fe₃O₄¹⁶.

Synthesis of the phenylboronic acid-functionalized magnetic expanded graphite (EG/Fe₃O₄-PBA)

At the outset, a certain amount of the as-prepared EG/Fe₃O₄ powder was added to 50 mL of 4-formylbenzoic acid (5 mg mL⁻¹) diluted in ethanol, and the mixture was stirred at room temperature overnight. Subsequently, a certain amount of the sodium cyanoborohydride was added multiple times during the reaction. After that, the precipitate was separated with a magnet and washed with ethanol and ultrapure water. Finally, the washed precipitate was dried in the oven at 90 °C overnight, and the resulting powder was labeled as EG/Fe₃O₄-PBA²².

Instruments

The surface functional groups of the synthesized materials were investigated by Fourier transform infrared spectroscopy (FT-IR/Thermo Nicolet, Avatar 360, USA) spectra in the range of 4000–400 cm⁻¹. The surface morphology of the EG/Fe₃O₄-PBA was studied by field emission scanning electron microscopy (FESEM/MIRA3 TESCAN, Czech Republic) images. The same instrument was used for energy-dispersive X-ray spectroscopy (EDX) to investigate the elemental composition of EG/Fe₃O₄-PBA at 20 kV accelerating voltage. The porosity characteristics of the EG/Fe₃O₄-PBA was studied using nitrogen adsorption/desorption isotherm (BEL, Belsorp- mini II. Japan) according to the Brunauer–Emmett–Teller (BET) and Barrett-Joyner-Halenda (BJH) methods. The residual concentration of ACT in the supernatant after the adsorption process was determined by a UV–Vis spectrophotometer (DR6000, HACH, USA) at λ_max of 242 nm.

Static adsorption experiments

The static (batch) experiments were conducted to evaluate the adsorption capacity of the synthesized EG/Fe₃O₄-PBA towards ACT molecules. For each series of experiments, a known amount of the fresh adsorbent (8–40 mg) was added to the glass flasks (150 mL) with 100 mL of ACT solution (5–40 mg L⁻¹). The experimental ACT solutions were prepared by diluting a stock solution of ACT with ultrapure water to give the specific concentration of the desired solutions. The pH of the solutions was adjusted from 2.0 to 10 with HCl and NaOH (0.1 mol L⁻¹) solutions. Also, the experiments were conducted under various temperatures (2–10 °C) and contact times (3–15 min). Additionally, using the equation from our earlier study²³, the performance of EG/Fe₃O₄-PBA was estimated in terms of ACT removal percentage (% removal) and EG/Fe₃O₄-PBA equilibrium adsorption capacity (q_e, mg g⁻¹).

Experimental design

The experimental design (Table S1) used in this study has been moved to the “Electronic Supplementary Information.”

Results and discussion

Characterization

The FTIR analysis was used to investigate the surface functional groups of the samples, and the results are presented in Fig. 1. As can be seen, the FTIR spectrum of EG (Fig. 1a) show the absorption bands attributed to the stretching/bending vibrations of C-H (~ 680 cm⁻¹), C-O-C (~ 1200 cm⁻¹), C = C (~ 1600 cm⁻¹), C = O (~ 1670 cm⁻¹), C-H (~ 3710 cm⁻¹), and O-H (~ 3780 cm⁻¹) functional groups^11,24. The FTIR spectrum of EG-Fe₃O₄ (Fig. 1b) exhibits an additional absorption band attributed to the stretching vibration of the Fe-O (~ 548 cm⁻¹) functional group²⁵, in addition to the characteristic absorption bands of EG, which were slightly shifted in their wavelengths. The FTIR spectrum of EG/Fe₃O₄-PBA (Fig. 1c) displayed additional absorption bands associated with the C = C (~ 670 cm⁻¹) bending and B-O (~ 1298 cm⁻¹) stretching vibrations, compared to the EG-Fe₃O₄ spectrum²⁶. Furthermore, the intensity of the C = C absorption band at around 1600 cm⁻¹ increased. These changes may all be attributed to the interactions between the PBA and EG/Fe₃O₄ functional groups²⁷. The microstructure, surface morphology, and elemental composition of the bare EG and EG/Fe₃O₄-PBA at different magnifications are shown in Fig. 2. The bare EG (Figs. 2a,b) exhibits twisted-like sheets with a size of more than 10 μm and semi-smooth surfaces. After the composition with Fe₃O₄-PBA, the images of EG/Fe₃O₄-PBA (Figs. 2c,d) reveal a significant change in morphology compared to EG. This indicates that Fe₃O₄ nanoparticles have grown and aggregated on the EG surfaces and the spaces between the EG sheets, resulting in a rough surface characterized by numerous nano- and micrometric pores. This roughness may enhance the specific surface area of the EG/Fe₃O₄-PBA, thereby increasing its adsorption capacity. The chemical composition of the EG/Fe₃O₄-PBA surface was analyzed using the EDX spectrum (Fig. 2e) and EDX mapping (Fig. 2f). The results revealed the presence and uniform distribution of carbon (C), oxygen (O), and iron (Fe) elements within the EG/Fe₃O₄-PBA structure. The porosity of the EG/Fe₃O₄-PBA was analyzed using BET-BJH, with the nitrogen adsorption-desorption isotherm and the corresponding BJH pore size distribution illustrated in Fig. 3. The nitrogen adsorption-desorption isotherm (Fig. 3a), covering a relative pressure from 0.4 to 1, exhibits a type III behavior with H3 hysteresis loops, indicating the presence of mesoporous and microporous structures²⁸. In addition, the BJH curve (Fig. 3b) revealed that the EG/Fe₃O₄-PBA consists of pores ranging from 5 to 100 nm. The BET-specific surface area and total pore volume of the EG/Fe₃O₄-PBA were determined to be 62.33 m² g⁻¹ and 0.412 cm³ g⁻¹, indicating a favorable surface area that can offer more active sites for adsorption reactions²⁹. The surface charge of the EG/Fe₃O₄-PBA nanocomposite was evaluated by zeta potential measurements over pH 2.0–10, as shown in Fig. 4. The results revealed that the EG/Fe₃O₄-PBA exhibited a positive charge at acidic pH (+ 25 ± 5 mV at pH 2.0), decreased to approximately zero at pH 5.9 (pHzpc ≈ 5.9), and became progressively negative at alkaline pH (reaching about − 27 to − 30 mV at pH 9–10). The pH effect on acetaminophen (ACT) adsorption showed that the adsorption rate increased with rising pH, and equilibrium was reached at pH 7.5. At low pH (< pHzpc), adsorption was suppressed, likely because high proton concentration competes for active binding sites and the strongly positive surface disfavors the optimal orientation and hydrogen-bonding interactions of ACT molecules. As the pH increases toward neutral, the surface charge moves through the pHzpc and electrostatic repulsion is reduced, which facilitates stronger hydrogen bonding, π–π stacking, and specific Lewis acid–base interactions between ACT molecules and the EG/Fe₃O₄-PBA surface, resulting in enhanced uptake and the observed optimum at pH 7.5. Electrostatic attraction plays a modulatory role: while it is not the dominant mechanism at the experimental optimum (pH 7.5, where ACT is largely neutral, pKa ≈ 9.5), electrostatic forces can become important under conditions where ACT is ionized or where local surface charge heterogeneity creates oppositely charged domains; overall, changing surface charge with pH indirectly controls the balance between electrostatic effects and specific chemical interactions. These combined effects, diminished electrostatic barriers near neutral pH, plus strong PBA-mediated affinity and π-interactions, explain the increasing adsorption with pH and the practical optimum at pH 7.5.

Fig. 1.

FTIR spectra of EG (a), EG/Fe₃O₄ (b), and EG/Fe₃O₄-PBA (c).

Fig. 2.

FESEM images of EG (a, b), and EG/Fe₃O₄-PBA (c, d). EDX spectrum (e), and EDX mapping (f) of the EG/Fe₃O₄-PBA.

Fig. 3.

Nitrogen adsorption-desorption isotherm (left), and BJH pore size distribution (right) of the EG/Fe₃O₄-PBA.

Fig. 4.

Zeta potential as a function of pH for the EG/Fe₃O₄-PBA nanocomposite.

Statistical analysis

The statistical Design-Expert software was employed based on CCD-RSM to conduct a polynomial regression analysis. A second-order quadratic polynomial regression equation (Eq. 1) was performed to predict the response based on the independent variables X₁, X₂, X₃, X₄, and X₅, which were derived as follows:

1

This equation can calculate the influence of variables on the response. According to the equation, positive coefficient values indicate that significant factors positively interact and affect the sorption process. In contrast, a negative sign for parameters indicates that these parameters have a negative effect on adsorption. The analysis of variance (ANOVA) was conducted to evaluate the model’s acceptability, and a quadratic model was selected to fit the data and predict the responses, as shown in Table S2. According to the results, the model’s P-value was less than 0.05 (P-value < 0.0001), indicating that the model was statistically significant at the 5% confidence level. The lack of fit (0.1531) was not substantial compared to the pure error (1.334). This non-significant lack of fit further confirms the model’s strong predictability. The summary statistics for the predicted models are presented in Table S3. As can be seen, the model’s standard deviation (SD) was low (1.942), and the R-squared (R²) value was very close to 1 (0.997), indicating the accuracy of the model’s predictions. The coefficient of variation (C.V. %) was 2.904, a low value that underscores the reliability of the fitted model. This low C.V. value should instill confidence in the audience about the model’s reliability. The slight difference between the adjusted R-squared (Adj-R²) and predicted R-squared (Pred-R²), less than 0.2, further supports the good correlation between the actual and predicted values. The adequate precision (AP), the signal-to-noise ratio, was above 4 (48.22), indicating the model’s sufficiency³⁰. The Pareto chart (Figure S1a) effectively shows the absolute values of the experimental independent variables against their standard values. Variables significantly influencing the results are indicated by bars that cross the red vertical line (P = 0.05). All factors had P-values below 0.05, demonstrating their statistical significance at a 95% confidence level and highlighting their critical role in the observed outcomes. The relationship between the observed experimental values and the predicted values from the model is illustrated in Figure S1b. The observed values align closely with the expected values, indicating consistency between the experimental data and the model’s prediction. The raw residuals of the model are presented in Figure S1c, corresponding to the related observations. These residuals are randomly distributed within the range of + 2 to −2. A residual distribution graph based on the case numbers is presented in Figure S1d. This illustration lacks a clear pattern for the expected model, suggesting that the errors are random.

Interactive effects of experimental variables

The interactive effects of experimental independent variables on the adsorption efficiency of ACT by the EG/Fe₃O₄-PBA were analyzed using response surface methodology. In this regard, three-dimensional (3D) response surface plots were created to evaluate the interactive relationships, illustrating the connections between the adsorption variables and the corresponding ACT adsorption percentages, as shown in Fig. 5. The interaction effect of solution pH and initial ACT concentration on the adsorption efficiency is shown in a 3D plot (Fig. 5a). The results show that the highest percentage of ACT adsorption occurred at a pH range of 6.5 to 7.5 and at lower ACT concentrations, particularly at 10 mg L⁻¹. The 3D plot illustrating the interaction effect of solution pH and temperature (Fig. 5b) indicates that both increasing the temperature (up to 50 °C) and decreasing it (down to 10 °C) did not positively impact adsorption. The maximum adsorption was observed within a temperature range of 20–30 °C and a pH between 6.5 and 7.5. The 3D plot of the interaction effects of contact time and adsorbent mass on ACT adsorption (Fig. 5c) demonstrates that the adsorption percentage was increased with an increasing dose of adsorbent (12 mg). Initially, extending the contact time increased the amount of adsorption; however, after reaching a time range of 15 to 25 min, further increasing the contact time negatively impacted the adsorption process. The 3D plot illustrating the interaction effect of contact time and ACT concentration (Fig. 5d) shows that higher ACT adsorption occurs at lower ACT concentrations and increases with longer contact time.

Fig. 5.

Three-dimensional (3D) response surface plots of the binary interactive effects of ACT concentration and pH (a), pH and temperature (b), contact time and mass of adsorbent (c), and ACT concentration and contact time (d) for ACT adsorption efficiency by the EG/Fe₃O₄-PBA nanocomposite.

Optimization of the adsorption process

The profile for the predicted values and the desirability function in the optimization process is illustrated in Figure S2. The desirability function is represented as a global function with maximum and minimum values related to ACT adsorption. This function uses a scale ranging from 0.0 (undesirable) to 1.0 (desirable). The adsorption experimental matrix results showed that a desirability of 1.0 was assigned for maximum removal (95.89%), 0.0 for minimum removal (18.66%), and 0.5 for moderate removal (57.28%). The optimum values for the experimental independent variables were found to be solution pH (X₁) = 7.5, adsorbent mass (X₂) = 12 mg, ACT concentration (X₃) = 80 mg L⁻¹, temperature (X₄) = 30 °C, and contact time (X₅) = 20 min. Under optimal conditions, the predicted removal percentage for ACT was 99.11%, with a desirability score of 1. An experiment was conducted using the identified optimal conditions to verify this prediction. The results showed the ACT removal percentage of 98.89 ± 2.37%, which closely aligns with the predicted value. This demonstrates the high accuracy of the model.

Adsorption isotherm models

The behavior of the adsorption mechanism can be expressed through the analysis and fitting of experimental equilibrium data with well-known adsorption isotherm models. In this regard, the experimental data were fitted to four isotherm models, and Table S4 lists the corresponding calculated parameters and correlation coefficient (R²) values. The best-fit isotherm model was selected based on its corresponding R² value, as a model with a higher R² value (closer to 1) produces a better fit than the other models³¹. The results revealed that the Langmuir model (R² = 0.999) was found to be the best-fit isotherm for the equilibrium adsorption of ACT molecules by EG/Fe₃O₄-PBA, which suggests that ACT molecules were uniformly adsorbed (homogeneous adsorption) onto the surface of EG/Fe₃O₄-PBA in a monolayer form (chemisorption)³². Furthermore, the Langmuir maximum adsorption capacity (Q_m) of ACT was 451.30 mg g⁻¹, which was very close to the experimental value (q_{e exp}) of 332.0 mg g⁻¹. The percentage difference between the theoretical Langmuir maximum adsorption capacity and the experimental value was approximately 26.5%. This deviation is consistent with literature reports, as theoretical models predict idealized monolayer adsorption, while experimental conditions may be influenced by multiple factors such as surface heterogeneity, diffusion limitations, etc. Nevertheless, the close alignment of the values, together with the high correlation coefficient, confirms that the Langmuir model provides a reliable description of the adsorption process. In the Langmuir model, a dimensionless separation factor (R_L) was also employed to describe the tendency for adsorption, where R_L>1 denotes unfavorable, R_L = 0 denotes irreversible, R_L = 1 denotes linear, and 0 < R_L < 1 denotes a favorable adsorption process³³. The calculated R_L value was within the range of 0.002–0.016 (0 < R_L < 1), indicating a favorable adsorption process. The Freundlich model employs 1/n as a constant to represent surface heterogeneity, which reflects the energy relative to the distribution and heterogeneity of the adsorbate sites³⁴. The 1/n value determines whether adsorption is considered favorable (0 < 1/n < 1), unfavorable (1/n > 1), or irreversible (1/n = 0)³⁵. The results demonstrated that the value of 1/n was 0.154 (0 < 1/n < 1), indicating that the adsorption process can be considered favorable. The D-R isotherm model estimated the energy of adsorption (E) to be 5.3558 kJ mol⁻¹ (below 8 kJ mol⁻¹), which indicates that physical adsorption may be considered to be effective in the adsorption process³⁶.

Adsorption kinetic models

The adsorption kinetics were investigated to elucidate the adsorption mechanism of ACT molecules onto EG/Fe₃O₄-PBA. Hence, the adsorption experimental results were analyzed using four kinetic models: pseudo-first-order (PFO), pseudo-second-order (PSO), intra-particle diffusion, and Elovich, as shown in Table S5. The kinetic parameters indicated that the adsorption process adheres more to the PSO model. This is emphasized by its related R² value of the PSO model (0.995), which is remarkably close to 1. Additionally, the calculated equilibrium adsorption capacity from the PSO model was 372.7 mg g⁻¹, which closely matches the experimental capacity of 332.0 mg g⁻¹. This correlation underscores the PSO model’s reliability in describing the adsorption process. The strong alignment of the results with the PSO kinetic model demonstrates that the adsorption of ACT molecules was predominantly driven by chemisorption³⁷.

Adsorption thermodynamics

The thermodynamic behavior of ACT adsorption onto EG/Fe₃O₄-PBA was evaluated using the temperature-dependent thermodynamic parameters, including standard Gibbs free energy change (∆G°), standard enthalpy change (∆H°), and standard entropy change (∆S°), using the following equations:

2

3

4

where K_D is the distribution coefficient, R is the universal gas constant (8.314 J mol^–1 K^–1), T is the absolute temperature in Kelvin, is the amount of ACT adsorbed on the surface of the adsorbent, and is the equilibrium concentration of ACT in the solution.

The values of ΔH° and ΔS° were determined from the Van’t Hoff plot of ln Kc versus 1/T as shown in Figure S3. The corresponding calculated parameters are presented in Table S6. The negative value of ΔG° at all temperatures means that the adsorption was feasible and spontaneous³⁸. The increase in the absolute value of ΔG° from − 1.21 to −15.67 kJ mol⁻¹ with an increase in temperature from 283.15 to 323.15 K demonstrates that adsorption becomes more efficient at higher temperatures³⁹. In addition, ΔGº values ranged from 0 to −20 kJ mol⁻¹ (from − 1.21 to −15.67 kJ mol⁻¹), indicating that the adsorption process primarily occurred through physical interactions, specifically Van der Waals forces. In contrast, the positive ΔH° value suggested that the adsorption process was endothermic, consistent with a chemisorption process⁴⁰. Furthermore, the ΔH° value (100.6 kJ mol⁻¹) fell within the range of 20.9 < ΔH° < 418.4 kJ mol⁻¹, providing additional evidence that the adsorption process can be attributed to chemisorption⁴¹. Based on the total data obtained regarding ΔG° and ΔH°, it can be concluded that both physical and chemical interactions drove the ACT adsorption process. The positive ΔS° value (360.3 J mol⁻¹ k⁻¹) indicated the affinity of EG/Fe₃O₄-PBA for adsorbing ACT molecules and suggests an increase in randomness during the adsorption process⁴².

Possible adsorption mechanisms

The adsorption of ACT onto the EG/Fe₃O₄-PBA nanocomposite was governed by multiple interactions, including π–π stacking, hydrogen bonding, Lewis acid–base interactions, and electrostatic attractions between the surface functional groups of EG/Fe₃O₄-PBA and ACT molecules. FTIR spectra (Figs. 6a–c) revealed new peaks and shifts in characteristic bands associated with –OH, –COOH, and B–O groups, indicating their active involvement in the binding process. UV–Vis spectroscopy (Fig. 6d) further confirmed adsorption, showing a significant reduction in the intensity of the ACT absorption peak along with a slight red shift. These spectroscopic observations clearly demonstrate the formation of specific interactions, underscoring the selectivity of EG/Fe₃O₄-PBA for ACT molecules. Adsorption modeling and thermodynamic analysis provide additional insights into the mechanism. Langmuir isotherm fitting (R² = 0.999) indicates monolayer adsorption on heterogeneous sites, while the pseudo-second-order kinetic model (R² = 0.995) suggests that chemisorption, involving electron sharing or exchange, governs the rate-limiting step. Thermodynamic parameters (ΔH° = 100.6 kJ mol⁻¹, ΔG° = 0 to − 20 kJ mol⁻¹) confirmed that the process was spontaneous and endothermic. Furthermore, the Dubinin–Radushkevich model (E = 5.36 kJ mol⁻¹ < 8 kJ mol⁻¹) reveals that physical adsorption also contributes, particularly at weaker surface sites. Together, these analyses indicate a dual adsorption mechanism, with chemisorption as the dominant process and physisorption supporting initial uptake and surface coverage. Although the BET surface area of the EG/Fe₃O₄-PBA was relatively modest (62.33 m²/g), the adsorption capacity was remarkably high, demonstrating that surface-area-dependent physical adsorption was not the primary driver. Instead, the functional groups introduced by PBA modification, particularly the boronic acid moieties, provide abundant active sites and strong affinity toward ACT molecules, enabling highly efficient adsorption even on a moderate surface area. Collectively, the spectroscopic evidence, adsorption modeling, and thermodynamic data indicate that the adsorption of ACT onto EG/Fe₃O₄-PBA was a synergistic process, dominated by chemisorption with supplementary physisorption. This highlights the critical and innovative role of PBA functionalization in achieving enhanced adsorption efficiency and selectivity.

Fig. 6.

FTIR spectra of EG/Fe₃O₄-PBA before ACT adsorption (a), pure ACT (b), and after ACT adsorption (c). UV-Vis spectra before and after ACT adsorption (d).

Regeneration and reusability performance

Investigating the reusability of an adsorbent is crucial for ensuring its efficiency and cost-effectiveness. To this end, the effectiveness of different eluents in desorbing ACT molecules from spent EG/Fe₃O₄-PBA was tested. The eluents selected for this study included ultrapure water, ethanol (96%, v/v), and isopropanol. The results are shown in Fig. 7a. The data indicate that ultrapure water was ineffective, achieving only a 37% desorption efficiency for ACT molecules. In contrast, both ethanol and isopropanol demonstrated greater desorption efficiencies of 87% and 70% for ACT molecules, respectively. This difference can be attributed to the fact that acetaminophen has limited solubility in ultrapure water, whereas it dissolves much better in moderately polar organic solvents. Ethanol and isopropanol reduce solvent polarity, disrupt the hydrogen bonds between the EG/Fe₃O₄-PBA and ACT molecules, and offer enhanced solvation of the aromatic ring and polar groups of ACT molecules. As a result, desorption is more efficient with these solvents than with ultrapure water. Consequently, ethanol was chosen as the optimal eluent for regenerating the EG/Fe₃O₄-PBA. The reusability performance of the EG/Fe₃O₄-PBA was assessed over seven cycles of adsorption and desorption, as shown in Fig. 7b. The results indicated that the EG/Fe₃O₄-PBA retains high adsorption efficiency even after six cycles of the adsorption-desorption process. In the seventh cycle, the removal efficiency of ACT was 80.5 ± 3.24%, which represents a decrease of approximately 18.3% from the first cycle, where the efficiency was 98.89 ± 2.37. This suggests that EG/Fe₃O₄-PBA could be a cost-effective adsorbent for ACT removal in practical applications.

Fig. 7.

Compares the desorption efficiency of different eluents for the desorption of ACT from EG/Fe₃O₄-PBA (a), and the adsorption efficiency of ACT during reusability tests (b).

Performance of EG/Fe₃O₄-PBA

Table 1 summarizes the adsorption performance of EG/Fe₃O₄-PBA for ACT in comparison with other reported adsorbents. In addition to maximum adsorption capacity (Q_max) and adsorption time, other parameters such as pH, temperature, isotherm model, and kinetic model are included to provide a comprehensive assessment. As shown, EG/Fe₃O₄-PBA exhibits a Q_max of 451.30 mg g⁻¹ within only 20 min, which is not only the highest capacity among the listed sorbents but also one of the fastest adsorption rates under near-mild conditions (pH 7.5, 30 °C). These results demonstrate that EG/Fe₃O₄-PBA was a promising and efficient adsorbent for ACT removal from aqueous solutions. On the other hand, to validate the effectiveness of the functionalization, the pristine EG/Fe₃O₄ was also tested under identical conditions. The EG/Fe₃O₄ exhibited a Q_max of 300.6 mg g⁻¹ within 30 min, while the functionalized EG/Fe₃O₄-PBA achieved a much higher Q_max of 451.3 mg g⁻¹ within only 20 min. This improvement highlights the role of PBA functionalization in enhancing the affinity and interaction sites for ACT molecules, leading to both higher adsorption capacity and faster kinetics.

Table 1.

Comparison of the maximum adsorption capacity, operating parameters, and fitted models of isotherm and kinetics for EG/Fe₃O₄-PBA with various adsorbents reported for ACT removal from aqueous solutions.

Adsorbent	Qmax (mg g−1)	Contact time (min)	pH	Temperature (°C)	Isotherm model	Kinetic model	Ref.
Chitosan-coated MWCNT	205	60	5.0	25	Freundlich	Second order	43
Fe3O4 nanoparticles	68.9	480	6.0	25	Langmuir	-	44
Double-oxidized graphene oxide	704.0	300	8.0	25	Langmuir	Second order	45
Silica microspheres	89.0	30	5.0	30	Freundlich	Second order	46
AC synthesized from spent tea leaves	59.2	60	3.0	25	Langmuir	Second order	47
Amine-functionalized superparamagnetic silica	58.00	30	6.0	25	Langmuir	Second order	48
AC derived from Quercus Brantii (Oak) acorn	45.45	150	3.0	45	Freundlich	Second order	49
Carbon-based pine gasification residues	270.3	300	5.0	40	Langmuir	Second order	50
Commercial wood-based PAC	267	360	5.8	30	Langmuir	Second order	51
Cork waste-based PAC	200	360
Peach stone-based PAC	204	360
PET waste-based PAC	113	360
Commercial coal-based PAC	255.0	1440
Magnetic expanded graphite (EG/Fe3O4)	300.6	30	7.5	30	Langmuir	Second order	This work
Phenylboronic acid-functionalized magnetic expanded graphite (EG/Fe3O4-PBA)	451.3	20

Application of EG/Fe₃O₄-PBA to actual water samples

The application of EG/Fe₃O₄-PBA in removing ACT molecules from actual water samples was explored using four real water samples. Before conducting the experiments, the concentration of ACT in the samples was adjusted to 80.0 mg L⁻¹ by adding an ACT stock solution (1000 mg L⁻¹). The experiments were conducted under optimal conditions without altering the native pH values. The chemical composition of the actual water samples is presented in Table S7, and the adsorption results are summarized in Table 2. The results showed that the adsorption percentages of ACT by EG/Fe₃O₄-PBA in river water samples (1 and 2) were 98.03% and 97.47%, respectively. In contrast, the adsorption efficiencies in hospital effluents (1 and 2) were slightly lower, at 88.23% and 91.53%. As indicated in Table S7, hospital effluents exhibited higher conductivity, total dissolved solids (TDS), total suspended solids (TSS), turbidity, and total hardness compared with river waters, reflecting a greater load of dissolved and suspended organic matter. These coexisting pollutants, particularly aromatic and pharmaceutical residues, can compete with ACT molecules for active sites and thereby reduce adsorption efficiency. Nevertheless, EG/Fe₃O₄-PBA still achieved high removal performance in all cases, demonstrating its robustness in complex water matrices.

Table 2.

Efficiency of EG/Fe₃O₄-PBA for adsorbing ACT molecules in real water samples; (N = 4).

Real sample	ACT adsorption% ± SD
River water (1)	98.03 ± 1.84
River water (2)	97.47 ± 1.66
Hospital effluent (1)	88.23 ± 3.58
Hospital effluent (2)	91.53 ± 2.67

SD standard deviation, N number of repetitions.

Conclusions

This study presents the successful development of a novel EG/Fe₃O₄-PBA nanocomposite with outstanding adsorption performance for the removal of ACT molecules from contaminated water. The functionalization of EG/Fe₃O₄ with PBA significantly enhanced both adsorption capacity (Q_max = 451.3 mg g⁻¹) and kinetics, achieving over 95% removal efficiency within just 20 min under optimized conditions. Adsorption followed the Langmuir isotherm and pseudo-second-order kinetics, indicating a chemisorption-driven monolayer mechanism. Thermodynamic analysis confirmed that the adsorption process was spontaneous and endothermic. Crucially, the EG/Fe₃O₄-PBA demonstrated strong efficacy not only in controlled laboratory conditions but also in real water samples, with removal efficiencies of approximately 98% in river water and above 88% in hospital effluents. Furthermore, its ability to retain over 80% adsorption capacity after seven regeneration cycles highlights its practical potential as a cost-effective and reusable adsorbent. These findings suggest that EG/Fe₃O₄-PBA was a highly promising adsorbent for the rapid and efficient removal of pharmaceutical contaminants from wastewater. Its high adsorption capacity, fast kinetics, and durability position it well for integration into existing water treatment processes, particularly for addressing emerging contaminants in environmental waters. Future research could focus on scaling this technology and exploring its application in diverse wastewater treatment settings, contributing to improved water quality and public health protection.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

All authors contributed to the study conception and design. P. A.: Writing - original draft, Investigation, Data curation, Formal analysis, Writing - review & editing. A. A.: Supervision, Conceptualization, Writing - original draft, Writing - review & editing, Methodology; Data curation; Validation. F.S.: Writing - Original Draft; Validation.

Data availability

The authors declare that the data supporting the findings of this study are available within the paper, a request for more detailed data should be sent to the corresponding authors with the permission of all authors.

Declarations

Competing interests

The authors declare no competing interests.