The PAS was collected from the Yanghu Wetland, Hunan Province, China. Prior to the experiment, the raw material was dried at 105°C for 24 h, crushed into fractions with particle sizes of 0.15–0.25 mm, and stored in a sealed bag at 4°C. Tetracycline purchased from Shanghai Ryon Biotechnology Co., Ltd.

The oxidative torrefaction experiments were performed in a rotary tube furnace. During each run, approximately 50 g of dried PAS was placed in the reactor. In the heating procedure, the temperature was increased from room temperature to 270°C at 3°C/min and the temperature was maintained for 30 min; the carrier gas (flow rate = 200 SCCM) consisted of nitrogen containing 9 vol.% of oxygen. After torrefaction, the PAS was pyrolyzed in a rotary tube furnace, and pure nitrogen was used as the carrier gas. In the heating program, the temperature was increased from room temperature to 675°C at 3°C/min and then maintained for 120 min. For comparison, the raw PAS was pyrolyzed in a rotary tube furnace under the same conditions. Notably, torrefaction and pyrolysis are continuous processes. Therefore, the effect of property changes that occur during the storage of torrefied PAS on the biochar properties was avoided. The biochars derived from raw and oxidatively torrefied PAS pyrolysis were denoted PBC and TPBC, respectively.

The surface functional groups of the biochar were identified using Fourier-transform infrared (FTIR, Bruker Vertex 70) spectroscopy. An X-ray diffractometer (XRD, X’Pert PRO MPD) with a Cu Kα radiation source was used to investigate the crystalline structure of the biochar. The surface chemical composition was analyzed by multifunctional X-ray photoelectron spectrometry (XPS, Thermo ESCALAB 250XI). The specific surface area and pore diameter were measured using a surface area analyzer (TriStar II 3flex).

Table 1 shows that the specific surface areas of the PBC and TPBC were 307.579 and 350.855 m²/g, respectively. The average pore diameters of PBC and TPBC were 1.954 and 1.979 nm, respectively, indicating that the pore types of the biochars were primarily micropores and mesopores. Compared with other similar studies, the biochars obtained from this study had higher specific surface areas. Figure 1 shows that PBC and TPBC showed similar XRD diffraction peak patterns, indicating that oxidative torrefaction had no significant effect on the crystal type of the PAS biochar. The spikes appearing at 21.94°, 31.45°, and 36.08° were related to SiO₂. The TPBC peak was slightly lower than that of PBC, indicating that TPBC contained lower ash content than PBC, which may be more advantageous for TC adsorption. In summary, TPBC may have better adsorbent properties than PBC, which may facilitate TC adsorption.

The kinetic study was performed using pseudo-first-order, pseudo-second-order, and intraparticle diffusion models. The corresponding linear plots and kinetic parameters are presented in Figure 5 and Tables 2, 3, respectively. Table 2 shows that the equilibrium adsorption capacities of PBC calculated by the pseudo-first order and pseudo-second order models were 10.7340 and 14.4886 mg/g, respectively; the equilibrium adsorption capacities of TPBC calculated by the pseudo-first-order and pseudo-second-order models were 12.3010 and 17.9019 mg/g, respectively. The higher equilibrium adsorption capacities for TPBC are consistent with the related experimental data. Moreover, the R²-values of the pseudo-second-order model (0.9663 for PBC and 0.9157 for TPBC) were higher than those of the pseudo-first-order model (0.9773 for PBC and 0.9901 for TPBC), indicating that the adsorption kinetics of TC by PAS biochar were in better agreement with the pseudo-second-order model.

The intraparticle diffusion model is used to describe the diffusion and mass transfer processes of the adsorbate inside and outside the adsorbent (Módenes et al., 2021). According to Figure 5C and Table 3, two-stage fitting was used for the intraparticle diffusion model; the linear plot contained two parts with different slopes. The first linear segment of the curve belongs to the rapid adsorption stage, in which TC is transported to the surface of PAS biochar, while the other linear segment is considered to diffuse through small pores (Adebayo et al., 2014). The linear part of the first linear plot cannot pass through the origin, indicating that intraparticle diffusion is not the only controlling mechanism (Adesemuyi et al., 2020). The concentration gradient decreased as the contact time increased and the TC concentration decreased, resulting in a low external mass transfer. Therefore, the adsorption rate was controlled by external mass transfer and intraparticle diffusion. Furthermore, the intercept of the second part is not zero, confirming that the adsorption reaction is also controlled by other steps (Chabi et al., 2020).

The equilibrium study can explain the relationship between the concentration of adsorbate in the solution and the amount of adsorption on the surface of the adsorbent during solid-liquid equilibrium (dos Reis et al., 2016). In this study, TC adsorption was investigated using the Langmuir and Freundlich isotherms. The constants and parameters were estimated for the two models using non-linear regression. The corresponding linear plots and calculated parameters are presented in Figure 6 and Table 4, respectively. Table 4 shows that the maximum adsorption capacity of TPBC calculated by the Langmuir model was slightly higher than that of PBC, indicating that PBC and TPBC had similar maximum adsorption capacities. However, according to the actual experimental data, the adsorption capacity of TC by PBC was significantly lower than that of TPBC at the same adsorption residence time, indicating that TPBC had a higher adsorption rate than PBC. Therefore, oxidative torrefaction promoted the TC adsorption of PAS biochar.

Larger Langmuir constant values typically correspond to greater affinity of the adsorbate toward the adsorbent. The K_L of PBC (0.1096) was lower than that of TPBC (0.1796), indicating the greater affinity of TPBC toward TC. Similarly, the K_f of PBC (11.6650) was lower than that of TPBC (12.2860), indicating the higher TC adsorption capacity of TPBC. The R_L values of PBC and TPBC calculated by the Langmuir isotherm were 0.1543–0.9012 and 0.1002–0.8477, respectively, indicating a favorable adsorption process. Simultaneously, the n values of PBC and TPBC calculated by the Freundlich isotherm were 2.5804 and 2.4455 (1/n ≈ 0.39 and 0.41), respectively, indicating favorable adsorption. The Freundlich model assumes that the surface-active sites are not equivalent, showing a heterogeneous surface. In addition, the adsorption strength decreases with increasing number of occupied sites (Areco et al., 2012; Baghdadi et al., 2016).

The thermodynamic studies can explain the spontaneity, feasibility and other thermodynamic information of the adsorption process (Su et al., 2021). The effect of temperature on TC adsorption on the biochar is depicted in Figure 7. The adsorption capacity increased with increasing temperature, indicating that TC adsorption by the biochars is an endothermic process. In addition, the TC adsorption capacity of TPBC was higher than that of PBC at different temperatures.

Table 5 shows that the ΔH values of PBC and TPBC were 24.4765 and 20.0708 kJ/mol, respectively, confirming that the adsorption process is endothermic. Simultaneously, the ΔH values of PBC and TPBC were < 40 kJ/mol, indicating that the dominant adsorption mode was physical. Therefore, TC adsorption by PAS biochar was primarily physical. The ΔS values of PBC and TPBC were positive (0.1021 and 0.0888 kJ/mol, respectively), indicating that the adsorption process is irreversible. The Gibbs free energy (ΔG) values of PBC were calculated as −4.9312, −5.9377, and −6.9737 kJ/mol at 288, 298, and 308 K, respectively. The ΔG values of TPBC were calculated as −5.7156, −6.4627, and −7.5098 kJ/mol at 288, 298, and 308 K, respectively. The ΔG values of PBC and TPBC were negative, indicating that adsorption was a spontaneous process (Baghdadi et al., 2016). ΔG values decreased as the temperature increased, indicating enhanced adsorption at high temperatures. Moreover, the ΔG values of PBC were higher than those of TPBC at different temperatures, indicating that TPBC adsorbed TC more easily than did PBC.

The higher TC adsorption capacity of TPBC than of PBC was the result of torrefaction pretreatment. In the oxidative process, devolatilization and thermal decomposition of PAS and oxidation reactions occur, resulting in better physicochemical properties of oxidatively torrefied PAS (Huang et al., 2020). Therefore, after the subsequent pyrolysis process, the obtained TPBC had a higher specific surface area, stronger hydrophobicity, and lower ash content, which led to a higher TC adsorption capacity. Kinetic, isotherm, and thermodynamic data reveal that the TC adsorption rate by PAS biochar was controlled by external mass transfer and by intraparticle diffusion; the adsorption process is favorable and irreversible, and the dominant adsorption mode is physical. For a more detailed analysis of the adsorption mechanism, the FTIR and XPS data before and after adsorption were used to investigate the role of organic functional groups in the biochars during adsorption.

Figure 8 shows that PBC and TPBC had similar FTIR spectra. The observed peaks at 3,436 and 3,433 cm^–1 primarily correspond to the O-H stretching of water (Khan et al., 2020; Ntzoufra et al., 2021). The peak at 1624 cm^–1 is attributed to the stretching vibrations of C = O and aromatic C = C (Xu et al., 2020). The peaks at 1,581 and 1,578 cm^–1 were attributed to the stretching vibration of skeletal C = C (Zhang et al., 2019; Chen et al., 2020). The peaks at 1,092 and 1,087 cm^–1 were related to antisymmetric Si-O-Si vibration, and the peak at 800 cm^–1 was related to Si-O stretching vibration (Chen et al., 2021). After adsorption, the peaks at 1,624 cm^–1 moved to 1,581 cm^–1 (PBC) and 1,578 cm^–1 (TPBC), indicating that the biochar adsorbed TC. The peaks at 1,581 and 1,578 cm^–1 are attributed to the stretching vibration of skeletal C = C (Zhang et al., 2019; Chen et al., 2020). The TPBC peak was slightly clearer than that of PBC, which could explain the higher TC adsorption capacity of TPBC.

XPS was used to investigate the adsorption mechanism. Figures 9A,B show that the PBC peak at 288.12 eV corresponds to carbon in the O–C = O group (Lei et al., 2014), whereas the XPS spectrum of TPBC did not show this peak, indicating that oxidative torrefaction could lead to the removal of the O–C = O group of PAS. Compared with Figures 9C,D, Supplementary Figures 1A–D, 2A–D, the peaks of PBC and TPBC correspond to the C–C, C = C, C–O, C = O, C–N, and N-H groups shifted by varying degrees after adsorption, due to the increase in the corresponding functional group amount. Moreover, TPBC showed a peak at 288.03 eV after adsorption, which corresponds to carbon in the O–C = O group of TC. According to Supplementary Figures 3A–D, the peaks of PBC and TPBC shifted by varying degrees after adsorption, which may correspond to changes in the Si-O group amount, confirming that ash content could participate in the adsorption process.

The capacity of TPBC to remove TC from aqueous solutions was compared with that of biochar produced from different types of biomass (Table 6). The comparison of Q_m implies that the adsorption capacity of the prepared biochar in this work is higher than that of the other six biochars (some of which were chemically modified) reported earlier for TC removal. The TPBC prepared by PAS pyrolysis combined with oxidative torrefaction in this work had a higher TC adsorption capacity than PBC. Therefore, oxidative torrefaction could be an effective approach to improve the adsorption capacity of PAS biochar for TC; and the use of PAS biochar could be recommended for the efficient treatment of wastewater containing TC, which could effectively remove TC and reduce the cost of biochar production.