Banana peel (BP) was sourced directly from a farm in Ho Chi Minh City, Vietnam. The steps involved in making biochar are depicted in Fig. 1. As a first step, the BPs were washed multiple times with deionized water and dried at 80 °C in an oven (IF 110 plus, Memmert, Germany). Then, under oxygen-limited conditions, BP was pyrolyzed at 500°C for 2 hours at a heating rate of 10°C min⁻¹ [30]. To prepare the products of pyrolysis (BPB) products for use in subsequent experiments, the BPB was washed many times with deionized water, dried at 80°C for one day, and then kept in a desiccator. The BPB was pretreated with KOH solution as follows: BPB was impregnated with 4 M KOH solution at a ratio (KOH: BPB) of 4:1 (w:w) in a glass beaker, well-mixed under magnetic stirring at 25°C for 2 h and subsequently air-dehydrated at 80°C for 8 h. Thereafter, the dried mixture was repyrolyzed at 800°C for 2 h [31]. Upon cooling to room temperature, the biochar sample was rinsed with HCl and DI water to neutralize the biochar medium. The material was then dehydrated at 80°C and placed in a desiccator for the adsorption tests that followed. The biochar product was labeled KOH-BPB. Detailed information on the preparation of biochar is provided in the Supplementary Materials (Text S1).

The characterization of materials was conducted by several analysis techniques, including SEM, BET, FTIR, and pH_PZC, which are detailed in the Supplementary Materials (Text S2).

High-density 50-mL polyethylene tubes were utilized for all experiments with a 30 mL reaction volume. The solution's pH was adjusted as necessary using a 0.1 M HCl solution or 0.1 M NaOH solution, and the readings were taken with a Milwaukee Mi150 pH meter (Rumania). Using a Jeio Tech OS-2000 Dual Action Shaker (Republic of Korea), the solution was shaken under identical settings in each experiment: 150 rpm, 24 h, and 25°C. pH was adjusted between 3 and 10. 20 mg L⁻¹ of DOX and 1 g L⁻¹ of KOH-BPB were used to investigate the effect of pH on the adsorption capacity. Adsorption isotherm experiments were performed at initial DOX concentrations ranging from 10 to 50 mg L⁻¹ and KOH-BPB concentrations of 1 g L⁻¹. Using a 20 mg L⁻¹ initial DOX concentration, a KOH-BPB dosage of 1 g L⁻¹, and an ionic strength (Ca²⁺ and Na⁺) of 0.05–2 M, the effects of ionic strength were examined. Using a Dynamica Halo XB-10 UV-VIS spectrophotometer (United Kingdom), the DOX concentration in a solution filtered through a 0.45-mm Whatman® membrane filter after the reaction time was measured at a wavelength of 300 nm. The data given in this study are the result of three independent studies. Detailed information on the chemicals used is provided in the Supplementary Materials (Text S3, Table S1).

The isotherm models are detailed in the Supplementary Materials (Text S4).

All experiments were performed in triplicate. The parameters for the adsorption isotherms, and the correlation coefficients (R²) were calculated in Microsoft Excel 2010. The error bars displayed in the figures were calculated in Origin 2021. The adsorption experimental data were presented as average values of three independent replicate treatments. The adsorption data obtained in this study are presented as the mean ± standard deviations (SD).

SEM images showing the final products, BPB and KOH-BPB biochar, are displayed in Fig. 2a and 2b, respectively. The scanning electron microscope images showed that the biochar has different types of tiny pores on its surface, which could serve as efficient adsorption sites. Pore filling is an essential step in biochar's ability to adsorb organic pollutants [32]. The substantial pore-filling effect of carbonaceous materials, as shown by Zhu et al. [33], makes it easy for biochar materials to absorb organic pollutants due to their large pore volumes and high surface areas. The surface areas (S_BET) of the BPB and KOH-BPB samples were calculated to be 8.251 m² g⁻¹ and 710.241 m² g⁻¹, respectively (Fig. 2c). Table 1 shows the BPB and KOH-BPB FTIR spectra from 4000 to 400 cm⁻¹. As illustrated in Fig. 2d, the bands at 3430 and 3411 cm⁻¹ could represent O-H stretching [34, 35]. Oxygen-containing functional groups (OCG) on KOH-BPB, such as carboxyl and hydroxyl groups, contribute substantially to DOX adsorption via hydrogen bonding [36, 37]. The other peaks can be assigned as follows: (a) the peak at 2933 cm⁻¹ is the characteristic C-H stretching vibration of aliphatics [38, 39]; (b) the peak at approximately 2340 cm⁻¹ can be attributed to the stretching of the O=C=O group [40]; and (c) the peaks at 1040 cm⁻¹ and 579 cm⁻¹ are assigned to C=O, C-O-C and C-C group stretching, respectively [41]. A comparison of the BPB and KOH-BPB spectra (Fig. 3d) indicates that the peaks at 1382 cm⁻¹ and 774 cm⁻¹ are either absent or weak in the KOH-BPB sample.

The pH_PZC is the pH at which the negative charges on an adsorbent surface are equal to the positive charges [42]. In Fig. 3b, the pH_PZC value was 7.7 which is approximately the same as that observed in other studies. Zhang et al. [43] found a pH_pzc of 7.5 for rice husk biochar, while Bagheri et al. [44] found the pH_pzc of 7.09 for Moringa seed biochar.

Fig. 3a shows the calculated DOX removal efficiencies of the biochar samples. The removal efficiency was approximately 94% in the KOH-BPB sample. It has been observed that raw BP (15.89 mg g⁻¹) has the lowest DOX adsorption capacity of the three materials studied (BP, BPB, and KOH-BPB), whereas BPB (44.61 mg g⁻¹) show a higher value, and KOH-BPB (89.11 mg g⁻¹) has the highest. Because of its larger volume of micro- and mesopores, as well as its larger surface area, KOH-BPB has a stronger capacity for adsorption. Hence, the higher adsorption affinity of KOH-BPB can be attributed to its pore volume and surface area, as the pore filling allows for more suitable adsorption sites to be available. High-temperature pore formation after KOH activation is explained by the following chemical processes, as shown by Abioye and Ani [45].

This increased the number of pores because K₂CO₃ is reduced by carbons to generate CO, CO₂, K, and K₂O. Increases in pore volume can also be attributed to the metallic K produced by Eq. (1) and Eq. (2), as the contact between these atoms and the carbon matrix can enlarge the interlayer distances [46]. Due to its porous structure, large pore volume, high specific surface area, and abundant functional groups, biochar has been successfully utilized as an adsorbent for the removal of antibiotics [47]. Furthermore, the change in the surface characteristics of the three materials seemed to be compatible with the substantially stronger adsorption of DOX on KOH-BPB relative to BPB and BP. The surface area (S_BET) was measured to be 0.164 m² g⁻¹ for BP, 8.251 m² g⁻¹ for BPB and 710.241 m² g⁻¹ for KOH-BPB. The results suggested that KOH-BPB could be used to treat water contaminated with DOX.

The effectiveness of DOX adsorption is significantly impacted by the solution's pH. Adsorption studies of DOX onto KOH-BPB were conducted at pH values between pH 3 and 10 to evaluate the impact of pH on DOX adsorption. The charges on DOX changed depending on the pH of the solution, as illustrated in Fig. 3d. DOX molecules contain tricarbonylamine, phenolic diketone, and dimethylamine groups; depending on the pH, DOX can be found in cationic (pH<3.50), zwitterionic (3.50<pH<7.07), neutral, and anionic forms (pH>7.07) [15]. The pH_PZC of KOH-BPB is 7.7 (Fig. 3b). The surface charge of the sorbent was negative at pH > pH_PZC while it was positive at pH < pH_PZC [48]. This can facilitate electrostatic attraction between the KOH-BPB surface and the cationic or anionic forms of DOX. A pH of 6 is favorable for DOX adsorption onto KOH-BPB (Fig. 3c). When the pH was adjusted from 3 to 6, the amount of DOX that could be adsorbed by KOH-BPB increased from 20.15 to 59.11 mg g⁻¹. Both DOX and KOH-BPB have positive charges at pH > pH_PZC = 7.7. DOX is negatively charged at pH > pK_a =6.22, and KOH-BPB is negatively charged at pH > pH_PZC = 7.7, hence raising the pH from 7 to 10 decreased the amount of DOX adsorbed on KOH-BPB from 50.73 to 44.09 mg g⁻¹. Consistent with earlier reports, the results demonstrate that as the pH increases from 2 to 6, the amount of DOX adsorbed increases, with the maximum capacity (59.11 mg g⁻¹) at pH 6. However, the adsorption capacity of DOX on biochar decreases as the pH is further increased from 7 to 11.

In this investigation, four widely used models - the Langmuir, Freundlich, Tempkin, and Dubinin-Radushkevich models - were utilized to characterize the adsorptive behavior of DOX on the KOH-BPB. It is evident that the selected models satisfactorily described the data for adsorption equilibrium (R² > 0.90), in the following order: Langmuir (R² = 0.9879) > Tempkin (R² = 0.9541) > Freundlich (R² = 0.9121) > Dubinin-Radushkevich models (R² = 0.9098). A regression technique was used to determine the corresponding parameters of the applied models (Table S2). Fig. 4b shows that the linear relationship between 1/qe and 1/Ce had a good fit (R² = 0.9879). The results indicated that DOX was absorbed by KOH-BPB via a monolayer adsorption. The maximum adsorption capacity (q_max) for DOX was 121.95 mg g⁻¹. In addition, the K_L value was 0.02 so the value of R_L was less than 1, revealing that the adsorption processes were favorable. Fig. 4c shows that the linear relationship between logq_e and logC_e had a good fit (R² = 0.9121). The parameter 1/n indicated the ability of the pollutant to fill the pores on the surface of biochar [49, 50]. The parameters 1/n and K_f of the Freundlich model were 0.55 and 6.81, respectively. The value of 1/n was less than 1 indicating that the pollutant is easily adsorbed by KOH-BPB [51]. Fig. 4c reveals that the linear relationship between q_e and lnC_e had a good fit (Table S2). The linear relationship between q_e and ɛ² of the D-R model is shown in Fig. 4e. The q_DR and β values were 67.66 mg g⁻¹ and 2.03x10⁻⁵ mol² kJ⁻², respectively. Using Eq. (10) (Text S4) to determine the average adsorption energy E (kJ mol⁻¹) was determined to be 156.97. If E < 8 kJmol⁻¹, the adsorption is physical adsorption; if E is between 8 and 16 kJmol⁻¹, the adsorption process can be classified as ion exchange; and if E is > 16 kJmol⁻¹, the adsorption process is chemical adsorption [52]. Therefore, the adsorption of DOX to KOH-BPB was a chemisorption process. Fig. 4a, b, c, d and e show the observed data and model-fitted results for DOX adsorption onto KOH-BPB. The results were best fitted with the Langmuir adsorption model (R² > 0.9879), and the R² value was lower for the Dubinin-Radushkevich model. The q_max reached 121.95 mg g⁻¹ based on the Langmuir model (Table S2), indicating that the superior DOX adsorption performance of KOH-BPB compared to other adsorbents in Table 2. The q_max of KOH-BPB is higher than that of other materials, such as pumpkin seed shell activated carbon (PSSAC) [16], granular activated carbon [53], spent black tea leaves [54], and rice husk ash (RHA) [17], but lower than that of iron-loaded sludge biochar [15], and rice straw biochar [14]. As a result, KOH-BPB biochar is a very attractive adsorbent for DOX.

The intensity of adsorbate-adsorbent interactions may be modified by the presence of electrolytes in the solution [55] hence the influence of salt concentration on adsorption was investigated. Fig. 5 a, b depicts the influence of Na⁺ and Ca²⁺, two common background cations, on DOX adsorption. The adsorption capacity decreases by 22.8 and 45.3 mg g⁻¹ for NaCl and CaCl₂, respectively, when the salt concentration is raised from 0 to 2 M. During the adsorption process, the increasing of ionic concentration decreased the active sites on the adsorbent, and electrolytes may compete with DOX molecules for adsorption on KOH-BPB, which confirmed the existence of electrostatic interactions between KOH-BPB and DOX. Increasing the ionic concentration decreased the active sites on the adsorbent [35]. The results revealed that the DOX adsorption capacity decreased due to the influence of Na⁺ and Ca²⁺. The findings were in agreement with the results that were provided by Wan et al. [56].

DOX removal testing with KOH-BPB in different water sources was evaluated in three kinds of waters, namely, pure water, tap water, and river water. River water samples were collected from the Sai Gon River, Vietnam. Filtration was used to remove suspended particles from a river water sample using experimental conditions identical to those described above. In Fig. 5c, the DOX removal effectiveness only marginally dropped in the natural water samples, with pure water showing the highest removal efficiency, followed by tap water, and finally river water. The findings suggested that KOH-BPB was an excellent choice for antibiotic elimination from the aquatic environment. Real wastewater should be studied in order to discover the application potential of KOH-BPB for the control of pharmaceutical contaminants in the future.

The pore size distribution, specific surface area, and surface functionality are shown in Fig. 6 to explain how DOX is adsorbed onto KOH-BPB. The specific surface area of KOH-BPB (710.241 m² g⁻¹) was improved relative to that of BPB (8.251 m² g⁻¹) (Fig 1c), as KOH treatment introduced more adsorption sites. In addition, the stronger DOX adsorption to KOH-BPB than BPB appeared to be attributable to the change in surface properties upon alkaline treatment. The SEM analysis demonstrated that KOH-BPB had many more micro and fine holes than BPB. The increased surface area and pore volume of KOH-BPB led to its high DOX adsorption capacity through a process known as pore filling [14]. The π-π stacking interactions are one of the key mechanisms of absorption for a chemical such as DOX. This is because DOX is a polycyclic aromatic compound (with 4 interconnected benzene rings). They are electron-rich areas on the adsorbent surface that allow for electron acceptor-donor interactions with aromatic groups. To further assess the adsorption mechanism of DOX adsorption on the biochar, FTIR spectra of KOH-BPB before and after DOX adsorption were obtained to qualitatively assess the qualitative adsorption mechanisms (Fig. 6). FTIR spectra of biochar after adsorption clearly showed new bands at 1660 cm⁻¹. The peak at 1750–1630 cm⁻¹ was due to C=O stretching, ester, amide, ketone groups [41] and the peak at 1525 cm⁻¹ was due to C=C in aromatic hydrocarbon [40]. Additionally, variations in C=O and C=C were observed, suggesting that π-π interactions between biochar (π-donor) and the benzene rings of DOX (π-acceptor) may participate in the adsorption process [57]. Due to the high electron-withdrawing capacity of the ketone group, DOX may act as a π-electron acceptors and strongly interact with biochar as a π-electron donor [57, 58].

Ahmed et al. [59] showed that the cost to estimate the price of biochar, including (1) the total capital investment (equipment, civil work, direct installation, and the auxiliary equipment), (2) operating costs (labor, storage, feedstock acquisition, transportation, and maintenance), and (3) electricity consumption. We synthesized a very small amount of biochar in the laboratory, so the feedstock purchase cost, HCl acid, KOH, and N² gas cost, and electricity usage was used to estimate the biochar cost. Banana peel was acquired from agricultural waste, and HCl acid and KOH were utilized sparingly. The feedstock, N² gas, KOH and HCl acid costs amounted to approximately 1.05 USD. Energy usage rates (1.3 USD/pyrolysis, Vietnam power rates) were used to estimate cost related to furnace electricity use. During each synthesis process, the furnace produced 40.0 g of KOH-BP biochar (10.0 g × 4 ceramic pots). Hence, banana peel biochar costs $58.75 kg⁻¹ ($2.35 per 40 gram). Detailed information on the chemicals used is provided in the Supplementary Materials (Table S3). Banana peel biochar costs more than biochars made from the corn cob ($56 kg⁻¹) [40], the rice straw ($0.55 kg⁻¹) [60], the rice husk ($0.58 kg⁻¹) [61]. In this work, banana peel biochar was synthesized on a modest scale in the laboratory, but large-scale synthesis could reduce the costs. Ha et al. [62] showed that the biochar cost was significantly reduced via the use of mobile pyrolysis units and improved logistics. Moreover, banana peel biomass is available in large quantities. According to the statistics of the Ministry of Agriculture and Rural Development, in 2020, banana production amounted to 2.19 million tons. Banana production in Vietnam increased from 470,000 tons in 1971 to 2.19 million tons in 2020. Bananas, which can be harvested picked year-round, average 2.1 million metric tons in Vietnam. Therefore, the adsorbents made from banana peel require low financial input. Based on the results of this research, the inexpensive, plentiful, and environmentally friendly adsorbent KOH-BPB can effectively remove DOX from water.