All chemicals including SMX in the present study were purchased from Sigma-Aldrich Co. (Saint Louis, MO, USA) (Table 1). Based on the solubility of SMX in water at ambient conditions [26], 10–100 mg/L of SMX were used for the adsorption experiments in this study. Four commercial ACs used in the present study were obtained from Sigma-Aldrich Co. (Darco® G-60, Norit® GAC and Norit® GA1) and Calgon Carbon Corp. (Calgon F400).

Pine wood (Pinus taeda) was used for production of PBC as reported in Jang et al. [27]. Briefly, the air-dried and sieved (20 mesh) pine wood was carbonized using the preheated pyrolysis furnace (MTI Corporation, Richmond, CA, USA) under 300°C for 15 min with 1 L/min of N₂ flow. The pristine PBC was designated as “R-PBC”.

Activation of R-PBC was carried out as follows: (1) Soaking 3 g of R-PBC in 40 mL of 4 M NaOH solution for 2 h at room temperature, (2) Pyrolyzing the dried NaOH-treated R-PBC under 800°C for 2 h with a heating rate of 3°C/min while maintaining oxygen limited conditions (2 L/min of N₂ flow), and (3) Washing the BC with 0.1 M HCl solution (200 mL) followed by a deionized (DI) water until the pH of BC reached 7.0. The washed and dried NaOH-activated PBC was designated as “A-PBC”.

The physicochemical properties of the R- and A-PBCs were investigated in detail. Elemental composition analyzer (PerkinElmer 2400 Series II, MA, USA) and scanning electron microscope (SEM) (FEI Verios 460, Hillsboro, OR, USA) were used to evaluate the elemental compositions and the morphology, respectively. Thermogravimetric analysis (TA Instruments, New Castle, DE, USA) was used to evaluate the contents of fixed carbon, volatile compounds, and ash in R- and A-PBCs in accordance with ASTM standard D7582-15 [28]. Fourier transform infrared (FT-IR) spectroscopic analysis was conducted using a FT-IR spectrometer (Bruker Optik GmbH, Ettlingen, Germany). Brunauer-Emmett-Teller (BET) surface area and pore-size distribution were evaluated on the basis of the measurements relating to N₂ adsorption with an apparatus at 77 K (Micromeritics Gemini VII 2390p, Norcross, GA, USA). X-ray photoelectron spectroscopy (XPS) analysis was performed using XPS/UPS -SPECS System (SPECS Surface Nano Analysis GmbH, Berlin, Germany) equipped with a monochromatic Mg Kα radiation to quantify carbon-oxygen functional groups. All spectra were acquired from a sample area of 1 mm × 1 mm. The resolution of PHOIBOS 150 photoelectron analyzer is lower than 1 eV. The pH of zero point charges (pH_PZC) analysis was conducted as described previously [27]. Briefly, 0.01 M of 50 mL NaCl solution was used to maintain the ion strength and the initial pH of solution was adjusted ranged from 3.0 to 10.0. The final pH of solution was measured after 2 d incubation with 0.01 g of BC at 20°C and 150 rpm. The pH_PZC was calculated by the ΔpH = 0.

Energy loss spectroscopy (EELS) was conducted to analyze sp² content of the BCs (R- and A-PBCs). As described by Jang et al. [27], thin TEM samples of R- and A-PBCs were prepared by using UC7 Ultramicrotome (Leica Microsystems Inc. Buffalo Grove, IL, USA). Electron energy loss spectra for R- and A-PBCs were collected with the electron energy loss spectrometer (Gatan, Pleasanton, CA, USA) attached to probe aberration corrected scanning transmission electron Microscope (FEI Titan 80–300, Hillsboro, OR, USA). For further calculation, the carbon K-edge energy loss spectrum was deconvolved into three Gaussian spectra [29].

The sp² content is defined as a relative value between the sample 1 s to π* transition peak area ratio and that of graphite as calculated by Eq. (1) [30]:

where the transition at 285.0 eV indicates the electronic transition from carbon 1 s orbital to C = C π* bonding orbital, the second transition at 292.0 eV indicates the electronic transition from carbon 1 s orbital to C − C σ* bonding orbital, and the third transition at 298.0 eV indicates the electronic transition from carbon 1 s orbital to C = C σ* bonding orbital [29, 31, 32].

The effects of initial solution pH on adsorption of SMX was conducted at pH 1–10. Before mixing with the BCs, the solution pH was set to the desired pH by adding 0.1 M of HCl or NaOH solution. The foil covered glass bottles were agitated at 150 rpm and 20°C for 5 d. To quantitatively evaluate the contributions of individual SMX species to overall sorption at a tested pH value, the following empirical model was used (Eq. (2)) [33]:

where K_d (L/kg) is overall sorption coefficient and K_d⁻, K_d⁰ and K_d⁺ are the sorption coefficients and α⁻, α⁰ and α⁺ are the mass fraction for the SMX⁻, SMX⁰ and SMX⁺, respectively.

The adsorption kinetic study was conducted by stirring 0.01 g of A-PBC and 0.1 L of SMX solution (20 and 100 mg/L) at pH 4 which was determined as the optimum pH from the above experiments. The aqueous samples were taken at a regular interval for 2 d. In addition, the adsorption experiments for isotherm study were conducted by vigorously stirring 0.1 L of SMX solution (10 to 100 mg/L, pH 4) and 0.01 g of A-PBC for 5 d to reach equilibrium. The SMX concentration in water was evaluated and used for development of an appropriate isotherm model.

During the batch adsorption experiments, the samples were centrifuged and then filtered using a 0.45 μm syringe filter. The concentration of SMX in filtrated solution was measured by a HPLC (LC-2030C model, SHIMADZU Corp., Torrance, CA, USA) with a C-18 column (Aeris PEPTIDE 3.6 μm XB-C18, Phenomenex Inc., Torrance, CA, USA) and a UV-vis detector operating at wave-length of 265 nm. More detailed information about analysis of SMX by a HPLC was described in previously [27].

Adsorption capacity at time t, Q_t (mg/g) was calculated by Eq. (3):

where C₀ and C_t are the SMX concentration of initial and at time t (mg/L), respectively, V is the volume of SMX solution (L) and M is the g of BC used in study.

The sum of squared error (SSE) was determined according to Eq. (4):

where Q_e and Q_c are the observed and calculated value of adsorption capacity (mg/g), respectively, and N is the number of measurements.

The six adsorption kinetic models including pseudo-first order, pseudo-second order, Elovich, two-compartment first order, intra-particle diffusion and liquid film diffusion can be represented as follows (Eq. (5)–(10)):

where Q_e = adsorption capacity (mg/g) at equilibrium time, Q_t = adsorption capacity (mg/g) at time t (min), K₁ = rate constant of pseudo-first order, K₂ = rate constant of pseudo-second order, a = rate constant of chemisorption, b = constant of the surface coverage, K_i = intra-particle diffusion rate constant (mg·min^0.5/g), C_i = a constant (mg/g), K_fd = adsorption rate constant, F_fast = mass fraction of fast, F_slow = mass fraction of slow, K_fast = first order rate constant for transfer into fast (h⁻¹), K_slow = first order rate constant for transfer into slow (h⁻¹), h = hour.

In this study, three isotherm models including Langmuir, Freundlich and Temkin models were used to fit the experimental data (Eq. (11)–(13)):

Where Q_e = adsorption capacity (mg/g) at equilibrium time, Q_m = maximum adsorption capacity (mg/g), K_L = Langmuir constant, C_e = equilibrium concentration (mg/L), R_L = separation constant, K_f and n_f = Freundlich constant, R = universal gas constant (8.314 J/mol), T = temperature in terms of Kelvin, b_T = Temkin constant, K_T = equilibrium bond constant related to the maximum energy of bond, t = time (min).

The physicochemical properties of R-PBC and A-PBC are presented in Fig. 1. For the elemental compositions, the C content increased from 56.4 to 85.8% after activation, whereas the O and H contents decreased from 37.6 to 13.5% and from 5.9 to 0.5%, respectively. With decomposition of O and H under high temperature [27], significant structural changes occurred in A-PBC. As shown in Fig. 2, the ratio of O/C decreased from 0.67 to 0.16 while the ratio of H/C decreased from 0.10 to 5.8 × 10⁻³, respectively. The decrease in the ratios of O/C and H/C represented decarboxylation (i.e., loss of CO₂) and demethylation (i.e., loss of CH₃) [34], respectively, suggesting the formation of porous polyaromatic structures with rich carbon content after the NaOH activation of R-PBC. Interestingly, the O/C and H/C of A-PBC were lower than those of other BCs made from grasses, crop residues, manure and sewage sludge, but were close to that of powder activated carbon (Fig. 2). The low O/C and H/C of A-PBC indicated high hydrophobicity and aromaticity in A-PBC which would be closely related to adsorption characteristics of A-PBC for hydrophobic contaminants.

Fig. 1 showed the calculated sp² content of R-PBC and A-PBC. Interestingly, A-PBC showed higher sp² content (73.7%) than that of R-PBC (58.0%). Similary, Yoo et al. [30] reported the increase in sp² content as a function of pyrolysis temperature during the pyrolysis of pine wood. Thus, the increase in sp² content in A-PBC might be due to the increase in pyrolysis temperature (800°C) during NaOH activation of R-PBC. Also, this was consistent with ¹³CDP/NMR data (i.e. a growing aromatic structure after NaOH activation of pine BC) which reported by Park et al. [35]. Furthermore, FTIR spectrum of A-PBC exhibited the negligible functional groups and more similar to commercial ACs which have high hydrophobicity (Fig. S1).

After NaOH activation, changes in the morphology of BC surface were observed (Fig. S2). The sponge-like structure in A-PBC indicated the development of micro/meso-pore on the surface. This is highly supported by the data from the textural analysis (i.e., decreased in average of pore width and increased in volume of pores) (Fig. 1). With the well-developed structure, high BET surface area (959.9 m²/g) was achieved in A-PBC. Thus, the activation of PBC increased hydrophobicity and surface area which can lead to more heterogeneous adsorption characteristics [21, 36].

Recently, a strong relationship between SMX species and surface charge of CMs has been reported in the literatures [8, 37]. To take the SMX property into consideration, adsorption capacities (Q_e, mg/g) of SMX on the R-PBC and A-PBC were investigated at various initial solution pH (Fig. 3). Both BCs exhibited similar pH-dependent adsorption patterns. The Q_e values increased with increasing pH from 1 to 4 and then reached the maximum at pH 4 for both BCs (58.91 mg/g for R-PBC and 437.36 mg/g for A-PBC). Meanwhile, the significantly decreased Q_e values were observed at pH higher than 7. Similar results were found from the pH-dependent SMX adsorption on bamboo-derived BC [19].

The higher K_d⁰ than both K_d⁺ and K_d⁻ at the pH of 2–5 (for R-PBC) and 1–6 (for A-PBC) was observed in both R-PBC and A-PBC (Table S1), indicating the importance of SMX⁰ species to the overall adsorption. At the pH ranges between 1 and 5, the contribution of SMX⁰ species to the overall adsorption was higher than 51% (Fig. S3). This was consistent with the data from Zheng et al. [8] who reported significant contribution of SMX⁰ species to the overall adsorption on giant reed BC. Interestingly, between pH 1 and 5, the contribution of SMX⁰ species to the overall adsorption on A-PBC was higher than that of R-PBC, suggesting that activation process enhanced hydrophobic interaction as well π-π interaction which was predominant adsorption mechanisms for SMX species on A-PBC [16, 38].

As shown in Fig. 3, the SMX⁺ and SMX⁻ are the dominant species at pH < pK_a1 = 1.6 and pH > pK_a2 = 5.7, respectively, and SMX⁰ is dominant at pH between pK_a1 and pK_a2. In addition, it is assumed that the surface charge of BC was protonated (i.e., positive charge) when the solution pH was lower than the pH_PZC of BC, otherwise it was the negative. Considering the pH_PZC of R-PBC (5.08) and A-PBC (6.83) (Table S1), a strong electrostatic repulsion between the SMX species and surface of BCs would occur at high pH (> 7) or low pH (< 1) resulting in relatively low value of Q_e. However, it is interesting to note that some portion of SMX (72–98 mg/g) was still adsorbed on the A-PBC at high pH (> 7), although a strong electrostatic repulsion would be expected between the SMX⁻ and negative charged surface of BCs. This was explained by a strong negative charged assistant hydrogen bond (CAHB) along with oxygen-containing group on surface [20]. Teixidó et al. [33] and Yu et al. [16] reported that -NH₂ and -SO₂NH- groups in sulfonamides can interact with oxygen-containing group on surface of CM_S. Thus, it is hypothesized that A-PBC has a high potential for hydrogen bonding accepting or donating due to its plenty of oxygen-containing group on surface of A-PBC (Table S1). As expected, the significant decrease in Q_e values (about 291 mg/g at pH 2 and 104 mg/g at pH 1) were also observed at low pH due to the much less hydrophobicity of A-PBC with abundant ionized forms and the inhibited Lewis acid-base interaction [1].

π-π electron donor-acceptor (EDA) interaction is one of the possible mechanisms for the adsorption of sulfonamides on BC surface [19, 39]. SMX has a strong π-electron acceptor nature due to its p-amino rings (donates loan pair electrons to the benzene ring) and N and/or O-hetero-aromatic rings (contribution to electronic resonance), while the graphite structure π-electron of BC can act as electron donor [16, 19]. The ability for π-π EDA interaction of different SMX species follows the order of SMX⁺ > SMX⁰ > SMX⁻ due to the decrease in π-withdrawing ability resulted from deprotonation of -SO₂NH- groups [40], while increased by protonation of –NH₂- groups [41]. At pH ranges between 4 and 6, this was not consistent with the order of K_d (K_d⁰ > K_d⁻ > K_d⁺) in this study. Thus, π-π EDA interaction was not the major mechanism for adsorption of SMX onto A-PBC at these pH ranges. This was also consistent with the previous research [16] which reported the strong hydrophobic interaction between TC and CNTs rather than π-π EDA interaction. In contrast, previous researches [42, 43] reported the possible π-π EDA interaction between SMX and CMs. Recently, Yoo [44] reported the increasing π-π* transition (plasmon) intensity after the chemical activation of PBC. Similarly, Peng et al. [45] reported that adsorption is determined mainly by the number or areas of aromatic rings in both antibiotics and adsorbent when the main adsorption mechanism is the π-π interaction. Collectively, it is reasonable to conclude that the A-PBC in this study with the high surface area (959.9 m²/g), hydrophobicity and sp² after the NaOH activation resulted in high SMX adsorption capacity via possible adsorption mechanisms including hydrophobic interaction, π-π EDA interaction and hydrogen bonding.

To understand the possible adsorption mechanisms and rate-limiting step, the kinetic study for adsorption of SMX onto A-PBC was conducted as described in section 2.5 (Fig. 4). Investigation of adsorption kinetics of SMX and A-PBC interaction using PFO, PSO, Elovich and two-compartment first order are summarized in Table 2. Most of studies reported that kinetic data from sulfonamides adsorption experiment was well-fitted by PSO indicating chemisorptive interaction [7]. As shown in Fig. 4 and Table 2, PSO (R² of 0.91–0.92) was more suitable than PFO (R² of 0.65) to described the interaction between SMX and A-PBC. In addition, Elovich was the best fitted kinetic model with lowest SSE (3.25–7.96) and highest R² (0.98) suggesting that strong chemisorptive interaction occurred on energetically heterogeneous surface of A-PBC [46].

Additionally, the time-courses of SMX adsorption data was analyzed using a two-compartment first order model (Fig. 4(a) and (b)). This model well fitted the dynamics of SMX (20 and 100 mg/L) on A-PBC with low SSE (3.18–7.38) and high R² (0.98). The significantly higher F_fast (0.83–0.85) and K_fast (116.16–163.7) value than those of F_slow (0.15–0.17) and K_slow (0.25–0.26), indicating that the fast sorption stage was predominant during adsorption process. This was supported by the calculated values from Elovich model (i.e. the higher a (adsorption rate) than b (desorption rate)).

In general, transfer of solutes to adsorbents is usually characterized by liquid film diffusion or intraparticle diffusion. In this study, two diffusion models (intra-particle diffusion and film diffusion) were used to gain insight into the rate-limiting steps affecting the kinetics of SMX adsorption onto A-PBC. As shown in Fig. 4(c)–(f), intra-particle diffusion model was more fitted (R² of 0.84) to the experimental results than liquid film model (R² of 0.65). Thus, intra-particle diffusion would be considered as the major limitation for the adsorption of SMX onto A-PBC. More detailed corresponding rate constants (i.e., K_i1-K_i3) (Fig. 4(c) and (d)) derived from linear regression represented the highest constant at the first stage (4.27–10.55 g/mg min^1/2), while rate constants of 2^nd (1.45–3.91 g/mg min^1/2) and 3^rd stages (0.46–1.27 g/mg min^1/2) attributed to the intra-particle diffusion showed low value. Especially, the rate constants of 3^rd stage assigned to micropore diffusion was significantly lower than those of the previous two stages. This phenomenon is frequently observed during adsorption of organic contaminants resulted from decreased active adsorption sites on BCs [46].

Several literatures reported that isotherms of SMX adsorption onto various BCs could be fitted into either Langmuir [21] or Freundlich [8, 18–20] model. Based on the type of isotherm of A-PBC (i.e., Type I) (Fig. 5), three phenomenological models including Langmuir (no decrease of heat), Freundlich (a logarithmic decrease of heat) and Temkin (a linear decrease of heat) were applied in this study. The results in Fig. 5 showed Freundlich model was more appropriate than Langmuir and Temkin models indicating that the adsorption may take place on a heterogeneous surface with varied affinities. Based on the n_f values derived from Freundlich model, the adsorption of SMX was favorable (i.e. the n_f value within the range 1–10) onto A-PBC with high energetic heterogeneity. Besides, when the results from the isotherm experiments were fitted to Langmuir and Temkin models, the Langmuir constant (K_L = 0.82) and Temkin constant (b_T = 57.63) suggested the SMX adsorption onto A-PBC would be favorable (i.e., 0 < R_L <1) and adsorption reaction would occur exothermically (i.e., b_T > 1) in the concentration range used in this study.

Table 3 lists the maximum adsorption capacities for SMX and surface areas of CMs. The Q_m value for A-PBC (397.29 mg/g) was higher than those of various CMs including BC, AC, CNT and graphene oxide. For example, Zheng et al. [8] and Sun et al. [21] reported the Q_m value of 1.93–4.99 mg/g and 0.25–7.40 mg/g using the BCs from giant reed and crop residues, respectively. In addition, Çalışkan and Göktürk [15] and Chen et al. [14] achieved the Q_m value of 185.19 mg/g and 240 mg/g using commercial AC and graphene oxide, respectively. From the statistical results (i.e., Pearson correlation) (Fig. S4), the listed surface area showed positive correlation (r = 0.904, p < 0.01) with Q_m value, suggesting that high SMX adsorption capacity on A-PBC could be attributed to its highly increased surface area (959.9 m²/g) with well-developed porous structure after activation.

To evaluate the feasibility of PBC for adsorption of SMX, the batch adsorption experiments were conducted using four commercial ACs (Calgon F400, Darco^® G-60, Norit^® CA1 and Norit^® GAC) under the same conditions (0.1 g/L, 100 mg/L of SMX and pH 4). As shown in Table 3, the Q_m of A-PBC (397.29 mg/g) was comparable to four commercial ACs (312.14 mg/g for Calgon F400, 328.83 mg/g for Darco^® G-60, 377.5 mg/g for Norit^® GAC and Norit^® CA1 for 399.94 mg/g). Thus, these results supported the A-PBC would have high potential for removal of SMX in wastewater and water after process optimization.

Recently, Ahmed et al. [19] reported that adsorption of SMX onto BCs decreased when containing competing compounds (i.e., sulfamethazine and sulfathiazole) while it was enhanced with low molecular weight organic acids (e.g., citric- and malic acid) [21]. Thus, further investigations will include adsorption of SMX onto A-PBC in various wastewater containing multiple contaminants. In addition, appropriate regeneration processes (e.g., chemical VS thermal) of the A-PBC should be studied to increase their cost effectiveness.