High specific surface area N-doped activated carbon from hydrothermal carbonization of shaddock peel for the removal of norfloxacin from aqueous solution

As a kind of broad-spectrum antimicrobial drug, antibiotics are widely used in the precaution and treatment of infectious diseases of humans and animals, playing an essential role in the development of human health and animal husbandry (Allen et al. 2010; Li et al. 2021). Although antibiotics have made distinguished contributions to the treatment of diseases, there are also a series of environmental problems resulting from human abuse. Consequently, a large number of drug-resistant pathogenic bacteria, called ‘super bacteria’, have been induced in the environment. They also pose a severe threat to human health and the ecological environment (Yang et al. 2020).

Norfloxacin (NFX) is one of the fluoroquinolones (FQ) antibiotics, which have been used for the treatment of urinary tract infections, bacterial infections and burn wounds (Wang et al. 2019a). According to statistics, the consumption of NFX is 215 tons each year in China (Yuan et al. 2016). NFX in the natural environment poses a threat to human health. In the past decades, many treatment methods, such as adsorption (Ersan et al. 2017), biodegradation (Peng et al. 2019), photocatalysis (Chen & Chu 2015), ozonation (Chen & Wang 2021) and Fenton oxidation (Chen et al. 2021) have been used in the treatment of NFX wastewater. Among all, the adsorption was considered as a suitable method owing to its low cost, high efficiency and easy operation.

Activated carbon was one of the best adsorbents used in the adsorption process because of its high surface areas and developed microporosities (Li et al. 2017a; Barczak et al. 2018). However, high-cost and low-quality production has always been a challenge in the production process of activated carbon, and is the main constraint on its large-scale commercial application. Therefore, to reduce the cost of activated carbon, it is a good choice to product activated carbon using agricultural waste or organic solid waste as raw materials (Yan et al. 2017).

In addition to raw materials, the preparation methods also played essential roles, which influences the pore structure and surface chemistry property of activated carbon (Sun et al. 2016). To obtain an excellent activated carbon, biomass has been hydrothermally carbonized to improve the chemical characteristics of carbon precursors (Jain et al. 2015a). After hydrothermal carbonization in aqueous solution, the activated carbon precursors contained abundant functional groups, bestowing on it considerable activation potential for the production of high-quality activated carbon (Jain et al. 2015b; Kabakci & Baran 2019). Therefore, hydrothermal carbonization is feasible and effective in preparing activated carbon precursors.

At the same time, nitrogen-doped (N-doped) activated carbon has attracted significant attention because it could improve the adsorption capacity for pollutants from wastewater (Wang et al. 2018). N-doped activated carbon has abundant functional groups and unique electronic characteristics, which significantly increases its adsorption capacity (Liu et al. 2013; Liu et al. 2014; Wang et al. 2019b). The N-doped activated carbon is mainly obtained in two ways, namely post-treatment and in-situ doping (Cao et al. 2016; Lian et al. 2016). Compared with the traditional methods, if the nitrogen dopants and biomass are placed in the hydrothermal carbonization environment, they will fully contact with each other in a hot liquid phase. In this way, the reaction is supposed to be more competitive at the high temperature and pressure, which may be a more excellent pre-treatment method to produce N-doped activated carbon (Xiao et al. 2020).

As organic solid waste, shaddock peel is widespread in China. However, it is often treated as garbage and not fully utilized. In this study, a novel N-doped activated carbon was extracted from shaddock peels through incorporating urea as a nitrogen source during hydrothermal carbonization (HTC) pre-treatment, which can be applied to remove the NFX from aqueous solution. Besides, the adsorption characteristics and mechanisms of NFX on NAC were investigated.

The shaddock peel was collected from a local supermarket in Zhengzhou, China. Then the shaddock peel was washed with deionized water, dried at 378 K for 24 h, and ground to powder. The norfloxacin hydrochloride was purchased from Haizhengshenghua biotechnology Co., Ltd in Henan, China. Urea, KOH and HCl were purchased from Kermel Chemical Reagents Co., Ltd in Tianjin, China.

15 g shaddock peel and 10 g urea were dispersed in 150 mL deionized water and placed into a 200 mL stainless autoclave for hydrothermal pre-treatment. Under N₂ atmosphere, the mixture was heated to 593 K with an increasing rate of 283 K per minute in an electric heater and then held for 120 mins. In this process, the pressure of reactor turned into 4.5 MPa with an agitation speed of 200 rpm. The hydrothermal products were separated by filtration and dried at 353 K for 12 h to obtain the carbonization materials denoted as hydrocar (Xiao et al. 2020).

The hydrochar was thoroughly blended with KOH in an agate mortar. Afterwards, the mixture was heated in a vertical tube furnace with a heating rate of 283 K min⁻¹ to a specific temperature under a N₂ gas flow and then held for a specific time. The calcination product was repeatedly washed with deionized water until neutral pH. Finally, the activated product was dried at 353 K for 12 h in an oven, which was denoted as N-doped activated carbon (NAC-x), where x was the temperature of activation.

The specific surface area of NAC was determined from nitrogen adsorption-desorption isotherms by using the specific surface and aperture analyzer (JW-BK 132F, CN) at 77 K. The external surface morphological structures of NAC were detected by scanning electron microscopy (SEM, FEI Quanta 200). The samples were sprayed with gold before testing to increase the conductivity. The high voltage of electron beam of scanning electron microscope is 20 kV. The elemental contents of NAC were characterized by X-ray photoelectron spectroscopy (ESCALAB-250Xi, USA). The spectrum position of each element was calibrated using 284.8 eV as the line position of carbon. The infrared spectrums of NAC were detected by using Fourier transform infrared (FTIR) spectrometer (Perkin-Elemer, Spectrum Two, USA) equipped with a pressed KBr pellet within the wave range 500–4,000 cm⁻¹. The detector of the FTIR spectrometer is LiTaO₃ thermal detector. The Raman spectrum of NAC-x was operated at room temperature on a confocal micro-Raman spectrometer (HR-800, USA) with a 532 nm laser excitation. A minimum of two acquisitions were done for each sample.

The NFX stock solution was prepared by the dissolution of 0.5 g NFX in 1.0 L deionized water. The working solutions were obtained by diluting the stock solution to the desired concentrations. In batch adsorption experiments, a known amount of NAC and 50 mL of NFX solution were placed in a 100 mL conical flask. The mixtures were stirred in a thermostatic water bath shaker at 150 rpm for 5–360 min. The pH of NFX solution was adjusted by 0.1 mol L⁻¹ HCl or 0.1 mol L⁻¹ NaOH. The mixture was immediately centrifuged and the remaining NFX concentration was measured by an UV-vis spectrophotometer (TU-1810, CN) at λ_max 276 nm.

Response surface methodology (RSM) is a statistical method that uses reasonable experimental design methods such as central composite design (CCD) and Box-Behnken design (BBD) to solve multivariate problems. BBD model was an independent quadratic design compared with CCD, which could optimize the most influential experimental conditions with the least number of experiments (Niu et al. 2021). Therefore, the BBD model was applied to experiments with three factors (X₁, X₂, X₃) and three levels (−1, 0, 1). In this experiment, the BBD model was used to optimize the optimal preparation conditions of NAC.

Based on the previous research, the experimental factors and their code levels are listed in Table 1. The experimental design matrix and response values are shown in Table 2. It can be seen from Table 2 that the actual and predicted adsorptive values of NFX onto NAC revealed good closeness between the predicted and experimental results.

The reliability of the model was measured by statistical parameters such as correlation coefficient R², adequate precision(AP) and coefficient of variation (C.V.%). The value of R² was 0.9879, which indicates that the experimental data of 98.79% can be interpreted by this binomial model. The coefficient of variation (C.V.%) is 2.97%(<10%), indicating that the experimental operation has high reliability (Kumar et al. 2015). The analysis of variance (ANOVA) for the NFX adsorption capacity is displayed in Table 3. The significance of related parameters is determined by ‘Prob > F’ and ‘F-value’. The ‘F-value’ > 5 and the ‘Prob>F’ < 0.05 indicated that the model was significant. It can be seen from Table 3 that the ‘Prob > F’ value of the model was <0.0001 and the‘F-value’ was 63.34, implying that the model fitting result is significant. The ‘P-value of lack-of-fit’ was 0.08, which demonstrates that the effect of the experimental error on the results is insignificant. In this case, X₁, X₂, X₃ are all significant, while the interactions term of X₁X₂, X₁X₃ and X₂X₃ are not significant. According to the F value, the order of significance for independent variables is: temperature > residence time > the mass ratio of KOH to hydrochar, and the regression model was valid for the NFX adsorption. The actual and predicted adsorption capacity of NFX (Fig. S1) show that the predicted results were very close to the experimental results.

Taking the adsorption amount of NFX as the response values, the three-dimensional response surface diagrams of the interaction between different variables are shown in Figure 1. It can be seen from the stereogram that the NFX adsorption uptake (q) rose rapidly at first and then declined with temperature increasing. Similarly, the values of q increased firstly and then decreased by increasing time or the mass ratio of KOH and hydrochar. Finally, the optimum preparation conditions of NAC were obtained by the Equation (2). The best experimental conditions were activation temperature of 1106 K, residence time 2.4 h, the mass ratio of KOH to hydrochar 2.3:1 and the optimal predicted adsorption capacity of NAC on NFX was 690.25 mg g⁻¹.

To verify the accuracy of the optimized conditions, five sets of parallel experiments were carried out under the optimal conditions, and the adsorption uptake of NFX were 695.39, 686.49, 692.01, 681.66, 688.31 mg g⁻¹, respectively. The experimental values were close to the predicted values, which can corroborate the accuracy of the prediction.

The surface morphology of NAC-1106 is shown in Figure 2. The outer surface of NAC-1106 has a well-developed pore structure and abundant micropores. The distribution of the pore structure on the activated carbon surface is due to the decomposition of organic matter and the activation of KOH.

The nitrogen adsorption-desorption isotherm of NAC-x is shown in Figure 3(a). The adsorptive capacity of N₂ rose rapidly at P/P₀ between 0 and 0.1, which demonstrates the existence of micropores. It displayed a hysteresis loop with the increased relative pressure. According to the IUPAC classification, it exhibited a Type I isotherm. The presence of the hysteresis loop was mainly due to the capillary condensation, implying the existence of mesopores in the NAC-x. Micropore volume of NAC-1106 was obtained with 0.883 cm³ g⁻¹ by the t-plot method.

Furthermore, the average pore diameter of NAC-1106 was 2.05 nm, and the pore size distribution of the mesopores was relatively concentrated. As shown in Figure 3(c), the correlational relationship between the adsorption uptake and BET specific surface area of NAC-x was established. The R² of the linear fitting curve was 0.76, indicating that the BET specific surface area had little correlation with the adsorption capacity of NAC-x. In other words, the BET surface area was not the only reason for the adsorption of NAC towards NFX, and there were other interactions between NFX and NAC-x.

The FTIR spectra of NFX, NAC-1106 before and after adsorption NFX are shown in Figure 4. As observed in the spectra of NFX, the spectral adsorption band at 3,422 cm⁻¹ is typical of the existence of the hydroxyl functional group (Acuna et al. 2017). The band at 2,982 cm⁻¹ is fitted to the -NH- stretching vibration of piperazine groups (Gurjar et al. 1996). The peaks loaded at 2,734–2,849 cm⁻¹ corresponds to the ethyl groups. The peak at 2,481 cm⁻¹ is assigned to carboxylic OH groups. The band at 1,715 cm⁻¹ is due to the C = O stretching of the carbonyl. The peaks centering at 1,617 cm⁻¹ can be assigned to the N-H bending vibration of quinolones (Fang et al. 2020). The peaks centering at 1,483 cm⁻¹ and 1,386 cm⁻¹ were assigned as the skeleton vibrations of the benzene rings (Peng et al. 2012). Furthermore, the spectral adsorption band at 1,035 cm⁻¹ are described as the C-F group, which represents the characteristic peak of NFX (Sahoo et al. 2012). The peak at 958 cm⁻¹ might be attributed to the NH bending vibration of amines. As for NAC-1106, the broad band at 3,443 cm⁻¹ might be attributed to the N-H group, O-H group of NAC-1106 or absorbed water (Zheng et al. 2014). The peaks loaded in 2,924 cm⁻¹ are due to the C-H stretching of the aliphatic series. The 1,630 cm⁻¹ might be attributed to C = O group vibration of NAC-1106 (Jia et al. 2013). In addition, the bands in 1,068 cm⁻¹ and 1,390 cm⁻¹ are considered to be stretching vibrations of C-O and -CH₃, respectively (Xu et al. 2013). Compared to NAC-1106, it can be seen from NAC-1106-adsorbed NFX that some peaks of NAC-1106-adsorbed NFX had position shifts and intensity changed. In addition, the new peaks at 1,037 cm⁻¹ and 1,492 cm⁻¹ indicate that NFX was successfully adsorbed on the surface of NAC-1106.

Raman spectroscopy was used to quantitatively examine the graphitization degrees. From Figure 4(b), all NAC-x showed two characteristic peaks at approximately 1,350 cm⁻¹ (D-band) and 1,600 cm⁻¹ (G-band). The D-band represents C atoms of defects or disordered structures, while G-band represents C atoms of graphite structures (Lee et al. 2016). Therefore, the lower intensity ratio (I_D/I_G) corresponds to a higher degree of graphitization (Liu et al. 2013). Generally, defects in the activation carbon could provide adsorption sites, which is conducive to the adsorption process (Li et al. 2017b). The I_D/I_G values of NAC-973, NAC-1073, NAC-1106 and NAC-1173 were 0.975, 1.034, 1.039, 1.045, respectively, indicating that the higher activation temperature made more defects at the activation carbon frameworks (Peng et al. 2021). Besides, all the I_D/I_G values of NAC-x were very close to 1 (values of graphene-based materials), further testifying that all NAC-x had a high degree of graphitization.

XPS was used to characterize the surface electronic states and chemical states of NAC-1106 before and after adsorption, which are displayed in Figure 5. Figure 5(a) shows the surface element of NAC-x. It can be seen that NAC-1106 had the highest N content, which may be related to its highest adsorption of NFX. Figure 5(b) shows that the obvious increase in N content indicates that NFX was successfully adsorbed by NAC-1106. For the C_1s spectrum (Figure 5(c)), three peaks were assigned to C = C (284.8 eV), C-C&C-N (285.6 eV), C = O/N-C-O (288.6 eV) respectively (Xu et al. 2019). Besides, the binding energy of C = C changed from 284.7 eV to 284.9 eV after NAC-1106 adsorbed NFX. The existence of π-π EDA interaction between the NAC-1106 and NFX may cause this result. At a neutral pH, the benzene ring on NFX had a strong electron-withdrawing ability due to the presence of the fluorine group, so it was generally considered as the π-acceptor (Keiluweit & Kleber 2009; Chen et al. 2015). NAC-1106 was an electron donor because of its abundant hydroxyl groups, so the π-π EDA interaction was between NAC-1106 and NFX. In addition, the binding energy of O-C-N/C = O changed from 288.6 eV to 289.9 eV after NAC-1106 adsorbed NFX, implying that there could be hydrogen bonding interactions. The O-H, NH₂ and -NH- groups of NFX interacted with oxygen-containing functional groups on the surface of NAC through hydrogen bonding. Similarly, the -OH, -NH- groups on the surface of NAC can interact with oxygen-containing functional groups of NFX (Teixido et al. 2011).

The N_1s could be divided into three peaks at 397.8 eV, 399.5 eV, 401.3 eV, which were attributed to pyridinic-N, pyrrolic-N and graphitic-N (Figure 5(d)). The shifts of pyridinic-N, pyrrolic-N and graphitic-N were observed from 397.8 eV, 399.5 eV, 401.3 eV to 399.1 eV, 399.8 eV, 401.6 eV after NAC-1106 adsorbed NFX. The binging energy of graphite N did not have a large deviation because graphite N is the most stable of the three kinds of N and it does not exhibit chemical properties (Sheng et al. 2011). Pyrrolic N is formed by incorporation of N atom into a heterocyclic ring. It is mainly considered as the lewis base site to combine with the lewis acid site on NFX (Wang et al. 2010; Lian et al. 2016). The pyridinic N combined with two adjacent sp² hybridized carbon atoms at the defects or edge of activation carbon, and it can donate one p electron to the π system, making the lonely pair electron of pyridinic N not confined. Therefore, pyridinic N can generally improve the hydrophobicity of NAC and enhance the hydrophobic effect during the adsorption process (Kumar et al. 2015). Among them, the binding energy of pyridinic N changed most obviously, so the hydrophobic effect may play a vital role in the adsorption of NFX.

The comparison of maximum monolayer adsorption capacity of NFX onto various adsorbents is listed in Table 5, which showed that the NAC-1106 had greater adsorption capacity than other adsorbents from the published literature.

The Koble-Corrigan (K-C) isotherm is a combination of Langmuir and Freundlich isotherms. From Table 5, the higher R² values and the lower χ² values of K-C models also indicated that this model was suited to represent the NFX adsorption process. In addition, all the constants (A_K, B_K and M) were positively proportional to temperature. However, the Temkin model and Freundlich model could not describe the adsorption process well because of the lower R² and higher χ² values.

The values of ΔG were −33.81, −35.96 and −37.83 kJ mol⁻¹ for NFX adsorbed on NAC-1106 at 298, 308 and 318 K, respectively. The negative values of ΔG at the different temperatures indicated that the adsorption process of NFX on NAC-1106 was spontaneous. The ΔH was 34.73 kJ mol⁻¹, confirming that the adsorption process was endothermic in nature, while the positive value of ΔS (ΔS = 0.23 kJ (mol K)⁻¹) confirms the increasing disorderliness at the adsorbent-adsorbate interface in the adsorption process of NFX onto NAC-1106.

To understand the relationship between the adsorption capacity and time, the pseudo-first-order, pseudo-second-order, and intraparticle diffusion (Equation (S5)–(S7)) were utilized to fit the adsorption data.

The non-linear plots of the adsorption capacity of NFX versus time were described in Figure 7(a), and the parameters of kinetics models were recorded in Table 6. The R² values of the pseudo-second-order model were all close to 1 and the values of q_exp were close to the values of q_cal,2, indicating that the pseudo-second-order model was more favorable for expressing NFX adsorption onto NAC-1106. These results implied that the chemisorption played an essential role during the adsorption process. Nevertheless, the pseudo-first-order model cannot describe the adsorption process of NFX onto NAC-1106 well because of the pretty low correlation coefficient. Based on the theory of diffusive and mass transfer, the diffusion mechanism of the adsorption process is discussed using the intra-particle diffusion model (Han et al. 2011). From Figure 7(b), the adsorption process was divided into three portions. This result showed the adsorption process of NFX onto NAC-1106 include three stages: (i) NFX was transferred from the solution to the surface boundary layer of NAC-1106; (ii) NFX was partitioned from the boundary layer to the solid surface or inside the particle; (iii) NFX reached adsorption equilibrium on the NAC-1106 surface active sites. At the same temperature, the k_ti values gradually lowered as the increase of time during the whole process. Besides, each fitted line didn't pass through the origin point, implying that adsorption process was driven by the boundary diffusion and intra-particle diffusion.

To further investigate the adsorption mechanism of NFX on NAC-1106, the solid addition method was used to obtain the point of zero charge (pH_pzc) of NAC-1106 (Rivera-Utrilla et al. 2001). As illustrated in Figure 8(a), the pH_pzc of NAC-1106 was around 8. The surface net charge of the NAC-1106 was positive at pH < pH_pzc; on the contrary, the surface net charge was negative when pH > pH_pzc. The pKa values of NFX were 6.22 and 8.51, which suggested that NFX had two proton-binding sites (carboxyl and piperazinyl group). Therefore, NFX can exist as NFX⁺(cationic form), NFX^± (zwitterionic form) or NFX⁻(anionic form) at different pH (Wang et al. 2017). The adsorption uptake of NFX at different pH values and the existing species of NFX in aqueous solutions are shown in Figure 8(b). As shown in Figure 8(b), the adsorption uptake of NFX first increased, then tended to flatten, and finally decreased as the pH increasing. According to the previous research, the high adsorption capacity of NFX was mainly dependent on the surface morphology and physical/chemical properties. Firstly, the high specific surface area and high porosity of NAC-1106 provided better conditions for the NFX pore/size-selective adsorption. Meanwhile, NFX adsorption was carried out by the hydrogen bonding between the -OH, -NH₂ and -NH- of NFX and the carbonyl groups of the NAC-1106, as well as via bonding between the oxygen-containing functional groups of NFX and the hydroxyl groups of the NAC-1106. In addition, the pyrrolic N in the NAC-1106 is mainly considered to be the lewis base site to combine with the lewis acid site of NFX, which may also be the reason for the high adsorption capacity of NFX. Besides, the π-π EDA interaction between NAC-1106 and NFX was another important crucial position for the high adsorption capacity of NFX, because NFX could be considered as the π-acceptor due to the presence of the fluorine group and NAC-1106 was an electron donor due to the abundant hydroxyl groups (Luo et al. 2018).

Several possible adsorption mechanisms are shown in Figure 8(c) to investigate the adsorptive behaviors existing in adsorption process. At pH < 5, the cationic form of NFX decreases with the increase of pH. Meanwhile, the adsorption capacity of NFX increases with the rise of pH, which is mainly attributed to the reductions of electrostatic repulsion between NFX⁺ and the surface positive charge of NAC-1106. At 5 < pH < 7, the adsorption process was also influenced by the hydrophobic effect. This was because water solubility reaches the lowest and lipophilicity was the strongest at pH = 7 (Yang et al. 2012). On the other hand, the high graphitization degree of NAC-1106 provides hydrophobic sites, improving the hydrophobic effect of NFX and NAC-1106 (Kumar et al. 2015). In addition, from the XPS spectrum analysis of NAC-1106, the hydrophobic effect made a significant contribution to adsorption uptake of NFX because the binding energy of pyridinic N shifted most obviously. At pH > 7, the hydrogen bond interaction is gradually weakened due to the ionization of NFX, and thus the adsorption capacity of NFX decreases slowly. In addition, the strong electrostatic repulsion between NFX⁻ and the negative surface charge of NAC-1106 at pH > 8 is one of the factors that reduces the adsorption capacity.

In summary, an NAC was prepared successfully using RSM. The NAC exhibited a rich pore structure and an enormous specific surface area (2,481.81 m²g⁻¹), which related to its high adsorption capacity. Besides, it was found that the experimental data of NAC-1106 matched better with Langmuir and K-C isotherm models and the adsorption kinetics process followed the pseudo-second-order model. The maximum adsorption capacity for NFX was 746.29 mg g⁻¹ at 298 K, suggesting that NAC is an excellent adsorbent in wastewater treatment for NFX removal.

This study was supported by National Natural Science Foundation of China (52006200).

The authors declare no competing financial interests or personal relationships to this work.

All relevant data are included in the paper or its Supplementary Information.