Optimized preparation of activated carbon from furfural residue using response surface methodology and its application for bisphenol S adsorption Open Access

In recent years, the public has expressed intensive concern about environmental pollution caused by endocrine disrupting compounds (EDCs), which had a potential negative impact on animal and human health (Lopez-Ramon et al. 2019). Bisphenol S (BPS), a major EDC, could cause abnormal function of biological endocrine systems. In addition, BPS has been frequently found in surface water, municipal sewage, sediments, daily products, and human specimens like urine and blood. Continuous accumulation of BPS might increase the potential risk to the ecological balance and human health. In order to effectively remove BPS from aqueous solutions, several treatment technologies have been reported in the literature, including bio-degradation (Wang et al. 2019), membrane filtration (Ding et al. 2018), photolysis (Cao et al. 2013), advanced oxidation (Mehrabani-Zeinabad et al. 2016), adsorption (Wirasnita et al. 2018), and so on. The adsorption technology as a promising and effective method attracted widespread attention because of its advantages of easy operation, facile synthesis, reusability, and no highly toxic by-products.

Activated carbon (AC) was utilized in diverse areas as an adsorbent due to its well-developed internal porous texture, relatively high surface area, large adsorption capacity, and a wide variety of functional groups (Heidarinejad et al. 2020). There were many functional groups in the carbon structure responsible for the adsorption of contaminants, mainly including oxygen-containing and nitrogen-containing functional groups. The unique adsorption properties were related to the existing functional groups of AC, which were dependent not only on the nature of the precursor but also on the activation process of the precursor (Bhatnagar et al. 2013).

Generally speaking, AC was manufactured from different carbon-rich organic materials by chemical or physical activation. The chemical activation was conducted as a single step in which the carbonization and the activation processes were implemented together. More specifically, the raw materials were impregnated with a chemical reagent like H₃PO₄ (Shamsuddin et al. 2016), H₂SO₄ (Karagöz et al. 2008), ZnCl₂ (Kumar & Jena 2017), FeCl₃ (Fu et al. 2017), NaOH (Yahya et al. 2015), KOH (Liu et al. 2020), and so on. Subsequently, the impregnated biomass was heated in an inert atmosphere. The physical activation involved carbonization and activation processes. During the carbonization stage, the raw materials were pyrolyzed in an inert atmosphere at a medium-high temperature to remove the hydrocarbons, and then the carbonized products were activated at high temperature by oxidizing gases such as steam (Zhou et al. 2018) and carbon dioxide (Lan et al. 2019). Frequently, chemical activation could usually get a larger specific surface area than physical activation. However, the chemical activation method had some disadvantages, including secondary pollution, complicated manufacturing processes, and equipment corrosion (Kim et al. 2017). In physical activation, the AC produced by steam activation had better surface characteristics and more developed pore structures than that produced by CO₂ (Song et al. 2013). Larger molecular dimensions made CO₂ more difficult to diffuse through the pore of the carbon particle (Rodriguez-Reinoso et al. 1995). In addition, CO₂ activation required a higher temperature than steam activation, attributed to its lower reactivity. It meant that the requirement of energy consumption would increase (Fu et al. 2013). Therefore, physical activation as an inexpensive and green method for preparing AC has attracted remarkable attention. The procedure of steam activation could be considered as an uncomplicated and environment-friendly method.

AC was a versatile product with good market demand and the worldwide consumption of AC was steadily increasing (Sahu et al. 2010). Commercial AC was usually obtained using high-cost and non-renewable resources such as coal and wood as precursors (Hameed et al. 2008). With the shortage of non-renewable resources, as well as severe deforestation, the precursor for producing AC became increasingly scarce. Therefore, it was an inevitable trend to seek new renewable and low-cost substitutes for preparing AC. Biomass, as suitable raw material for preparing AC, has attracted more and more attention because of its easy accessibility and cheapness. Much previous literature has reported on the production of AC by agricultural wastes and industrial by-products, e.g., palm shell (Kittappa et al. 2020), sawdust (Okoya & Diisu 2021), bamboo (Ma et al. 2019), corncob (Amen et al. 2020), rice husk (Nche et al. 2017), coconut shell (Higai et al. 2020), and durian shell (Chandra et al. 2007), etc.

The furfural was produced from agricultural waste corncob as raw material after high-temperature hydrolysis. The residual solid waste produced in the furfural production process was furfural residue (FR). It was reported that 12–15 tons of underutilized FR were generated per ton of furfural production, and there were around 23 million tons of FR annually in China (Wang et al. 2016). Most of the FR was incinerated as waste, resulting in the waste of useful energy and serious environmental pollution problems. Therefore, how to reuse it to transform value-added products attracted remarkable attention. The FR contained carbon-rich substances including lignin and cellulose, which made it a promising precursor for the preparation of AC (Ao et al. 2021). The preparation of AC by utilizing low-budget and plentiful industrial by-product FR might be a promising process for secondary recycling of wastes.

The most significant characteristic of the AC was highly influenced by its preparation conditions. The response surface method (RSM) was a statistical tool for the optimization of preparation conditions by assessing the interaction effects of several independent variables with a minimum number of experimental runs (Gupta et al. 2017). In this work, the RSM was utilized to optimize the preparation conditions of activated carbon from furfural residue (FRAC) by steam activation. Furthermore, the FRAC was characterized and the adsorption kinetics and thermodynamics of BPS were investigated to evaluate the adsorption performance of FRAC.

FR was obtained from Luyuanliangkang Biochemical Co., Ltd (Shandong, China). The FR was rinsed with deionized water to remove water-soluble impurities, and then dried under 105 °C to remove moisture and volatile impurities. BPS (≥99%, C₁₂H₁₀O₄S, FW = 250.27) was purchased from Macklin Biochemical Technology Co., Ltd (Shanghai, China). BPS stock solution of 2 mmol/L was obtained by dissolving in a certain proportion of ethanol-deionized water mixed solution. The solution was ultrasonically treated until the BPS was completely dissolved, and then the working solutions were stored at 5 °C under dark conditions. The concentration of adsorption experimental solution was obtained by suitable dilution with deionized water. The other chemical reagents used in this experiment, such as HCl and NaOH, were all of analytical grade.

The elemental analysis of the FR was conducted by an elemental analyzer (Vario EL cube, UNICUBE, Germany). The data of ultimate analysis are given in Table S1. The analytical results revealed that FR had a high carbon content, indicating that the FR was suitable for the preparation of AC.

The carbonization process was conducted by loading dried FR into the dry distillation kettle, and then the dry distillation kettle was maintained at 450 °C for 1 h to obtain carbonaceous precursor. Next, the carbonaceous precursor was heated in a vertical tubular furnace (KJ-T1200-S5010LK1-L, Zhengzhou Kejia Furnace Co., Ltd, China) under a nitrogen environment with a constant heating rate (10 °C/min). Afterward, N₂ flowing was replaced by overheated steam when the target activation temperature was reached. After activation for a certain time, steam was switched to N₂ flowing until the tubular furnace was cooled to the ambient temperature.

The iodine adsorption value was conducted according to China National Standards (GB/T12496.8–1999). The yield of FRAC was defined as the mass ratio of FRAC to the carbonaceous precursor. The thermal stability of FR was determined by Thermogravimetric Analyzer (TGA, SDT-Q600, USA). The morphology and microstructure of FR and FRAC were observed by a scanning electron microscope (SEM, Zeiss Gemini 300, GER) and the specific surface area of FRAC was measured by a surface analyzer (BET, JW-BK100A, CN). To identify the functional groups of FR and FRAC, the Fourier transform infrared spectroscopy (FTIR, Spectrum Two, USA) was conducted in the range 500–4,000 cm⁻¹ by mixing FRAC with KBr to get a translucent disk.

RSM was utilized to fit the functional relationship between the preparation conditions of FRAC and iodine adsorption capacity (Q), and the optimal conditions were found by analyzing the regression equation (Mourabet et al. 2017). Based on the preliminary single-factor experiments, the interaction of each variable was investigated by using Box-Behnken Design (BBD) in the software of Design-Expert 8.0.6 (Liu et al. 2021). Generally, the BBD consists of 17 experiments, including k² factorial points, k axial points, five repetitive runs at the central point for the estimating of the experimental error, and the k indicated the number of independent variables.

The actual values of the iodine adsorption capacity obtained from the experiments versus the predicted values gained from Equation (4) are shown in Figure S1. From Figure S1, all the points were located on the diagonal between the predicted values and the actual values. It could be observed that the correlation between the FRAC preparation conditions and the iodine adsorption capacity was perfectly captured by the developed model due to the good closeness between the predicted and experimental values.

The applicability of the model was estimated based on correlation coefficients, which are given in Table S2. The high determination coefficient (R²) value (R² = 0.9876) implied that 98.76% of the total variation in the iodine adsorption capacity was attributed to the preparation conditions of FRAC, indicating that the established quadratic model was appropriate for predicting the iodine adsorption capacity of FRAC. In addition, the significance of the quadratic model was further justified with fairly good consistency between the adjusted R² (0.9716) and predicted R² (0.9084) (Ani & Ochin 2018). The Adequate precision (AP) and Coefficient of variance (C.V.%) are a measurement of the accuracy of experiments. As shown in Table S2, the AP (22.56) of the model was much higher than 4 and the C.V.% (1.90%) of the model was much less than 10%, suggesting the reliability of the established model (Zhu et al. 2014).

Analysis of variance (ANOVA) was further applied to justify the adequacy of the quadratic model, and the results are listed in Table 3. The statistical significance of the established model in explaining the relationship between the iodine adsorption capacity and the preparation conditions of FRAC was estimated by the Prob. > F (p-value) and F-value. The value of F-value more than 5 and a small p-value (<0.05) indicated that the model was significant. Otherwise, the model was insignificant. According to Table 3, the F-value was 61.92 and p-value <0.0001, implying that the quadratic model was successful in predicting the interaction between the iodine adsorption capacity and the preparation conditions of FRAC (Gani et al. 2017). In addition, the ‘Lack-of-fit’ with a p-value (0.5144) and F-value (0.90) was not significant, showing that the residual error caused by random error could be ignored and the quadratic model had superior performance within the scope of the experimental data. In this study, X₁, X₂, X₃, X₁X₃, X₂X₃, X₁², X₂², and X₃² were significant for the iodine adsorption capacity of FRAC, which could be inferred by all p-values being less than 0.05. On the contrary, X₁X₂ was an insignificant model term. X₁², with an F-value of 117.93, had the most significant impact on the iodine adsorption capacity of FRAC. Based on the F-value, the intensity sequence of the independent variables was: X₁ > X₃ > X₂. Therefore, the ANOVA proved that the established model was adequately fitted to represent the interaction between the preparation conditions of FRAC and iodine adsorption capacity.

The three-dimensional (3D) response surface plots, illustrated in Figure 1, were applied to identify the relationship between the experimental variables and iodine adsorption capacity. It was observed from Figure 1(a) and 1(b) that the iodine adsorption capacity firstly increased and then decreased gradually. This may be attributed to the uniform micropores developed by the carbon-water reaction. However, with the increase of activation temperature, the micropores were enlarged to mesopores and macropores, thereby reducing the adsorption capacity of iodine to some extent (Sureshkumar & Susmita 2018). The iodine adsorption capacity of FRAC initially increased and thereafter declined with the prolongation of activation time, which can be seen from Figure 1(a) and 1(c). The burn-off of the existing pores or even collapse occurred with an increment in the activation time, leading to a decrease in iodine adsorption capacity (Kadir et al. 2014). According to Figure 1(b) and 1(c), the same behavior was observed in the effect of the mass ratio of char to H₂O, i.e., the iodine adsorption capacity increased first and then decreased with increasing mass ratio of char to H₂O. The original pores would be collapsed, attributed to superabundant steam, causing the reduction of iodine adsorption capacity.

In this study, preparation conditions of FRAC were optimized based on iodine adsorption capacity as a response. The optimal operating conditions were obtained by solving partial derivatives of independent variables according to Equation (4). The optimal conditions for the preparation of FRAC were activation temperature of 922 °C, activation time of 62 min, the mass ratio of char to H₂O of 1:4.5, and the corresponding predicted value of the iodine adsorption capacity was 1,503.58 mg/g. Five repeated experiments were performed under optimal conditions to verify the accuracy of the prediction, and the average iodine adsorption capacity was 1,501.84 mg/g. The result showed a good closeness between the predicted and experimental values, suggesting the accuracy of the prediction.

Thermogravimetric analysis was conducted in a nitrogen atmosphere at a pyrolysis rate of 10 °C/min, and the final pyrolysis temperature was 800 °C. Thermogravimetry (TG) and its derivative thermogravimetry curves (DTG) of FR, shown in Figure S2, suggested that four weight-loss stages were observed during thermal decomposition. The weight loss of the first drying stage (25–110 °C) was mainly ascribed to the liberation of moisture content. In the second pre-heating transitional stage, which occurred at temperatures ranging from 110 to 225 °C, the weight loss curve almost became a plateau, implying that the pyrolysis rate tended to be relatively stable. The third pyrolysis stage with the weight loss of approximately 51.29% occurred at 225–450 °C. In addition, a large weight loss peak with a maximum pyrolysis rate of 0.837%/min was generated at 338 °C, corresponding to the escape of a large amount of hemicellulose and cellulose (Yang et al. 2019). In the fourth carbonization stage (450–800 °C), residues of around 33.18% were found according to the TG curve of FR, while the DTG curve was maintained horizontally with little change. The thermogravimetric analysis results showed that FR can be used as a good precursor for the production of AC.

The scanning electron microscopy (SEM) micrographs of carbonaceous precursor (a) and FRAC (b) are given in Figure 2. As seen from Figure 2(a), the surface morphology of carbonaceous precursor was rough without deeper pore structures. However, the FRAC (b) revealed an irregular and heterogeneous surface morphology with lots of small crevices. During the process of activation, the reaction of carbon with steam generated lots of pores through the loss of volatile components, which remarkably increased the surface area of FRAC.

The N₂ adsorption-desorption isotherm and the pore size distribution plots of the FRAC are given in Figure 3. It was evident from Figure 3 that the volume of adsorbed N₂ increased sharply at P/P₀ < 0.1, which was caused by the microporous structure of FRAC. The slope of the isotherm decreased gradually with the increase of P/P₀, and a hysteresis loop was observed at high P/P₀, which was related to the capillary condensation in mesoporous structures. According to the IUPAC classification scheme, the isotherm conformed to the combination of standard type I and standard type IV (Ma et al. 2020).

The pore size distribution plot that was used to characterize the internal structures of FRAC was obtained by using the Barrett–Joyner–Halenda (BJH) method. The special surface area (S_BET) of the FRAC was estimated according to the Brunauer–Emmett–Teller (BET) equation. The total pore volume (V_tot) was evaluated by single-point adsorption of N₂ at P/P₀ = 0.993 (Sun et al. 2016). Micropore volume (V_mic) was determined by the t-plot method (Passe-Coutrin et al. 2008). The mean pore diameter D_p was calculated from D_P = 4V_tot/S, in which V_tot and S were the total pores volume and S_BET, respectively (Han et al. 2014). S_BET and the pore volume of the carbonaceous precursor were measured by the same method. The S_BET of FRAC (1,662.41 m²/g) was higher than that of carbonaceous precursor (23.88 m²/g). The V_tot of carbonaceous precursor and FRAC were 0.075 cm³/g and 1.20 cm³/g, respectively. In addition, the mean pore size of FRAC was estimated to be 2.89 nm, indicating that it was typically a mesoporous carbon sorbent.

The specific surface areas of AC obtained by various raw materials and preparation methods were different from each other. The comparison of S_BET between FRAC and other biomass-derived AC recorded in the literature is given in Table S3. In general, the chemical activation was more effective than the physical activation. However, the FRAC prepared using steam activation had a relatively higher S_BET than AC produced from other biomass wastes after the optimization of the preparation conditions by RSM.

The Fourier transform infrared (FTIR) spectra of precursor (a) and FRAC (b) are illustrated in Figure 4. From Figure 4(b), the band at 3,447 cm⁻¹ was ascribed to the stretching vibration of -OH groups, which originated from hydroxyl functional groups or adsorbed water in FRAC (Zhang et al. 2020). The peaks at 2,924 and 2,853 cm⁻¹ corresponded to C-H stretching in the methyl and methylene groups (Shi et al. 2020). A strong peak at 1,632 cm⁻¹ could belong to the C = O stretching vibration of carbonyl or carboxyl groups (Zhu et al. 2016). The stretching vibration of -CH₃ at 1,384 cm⁻¹ was related to methyl structures (Kyzas et al. 2016). The band in the range of 1,300–1,000 cm⁻¹ was probably due to the C-O surface group stretching (Benadjemia et al. 2011). It also could be observed from Figure 4 that compared with the carbonaceous precursors, the position of some peaks of FRAC had shifted and the intensity had changed.

Adsorption kinetics was utilized to elucidate the adsorption process of BPS onto FRAC. Specifically, three sets of experiments were performed at 298, 308, and 318 K, respectively. Four different kinetic models, including pseudo-first-order, pseudo-second-order, Elovich, and intraparticle diffusion models, were applied to analyze the experimental data.

The fitting plots of four kinetic models to experimental data are illustrated in Figure 5(a) and 5(b), and the related kinetic parameters are depicted in Table 4. From Figure 5(a), the adsorption capacity of BPS increased sharply at first; thereafter it continued to increase at a relatively slow rate until the adsorption equilibrium was reached after 180 min. Based on the above result, a contact time (t) of 180 min was selected for the following adsorption experiments to ensure complete equilibrium. In addition, the adsorption amount of BPS decreased with the increase of temperature, revealing that lower temperature was beneficial to the adsorption of BPS on FRAC. According to Table 4, the correlation coefficients R² (>0.9904) of the pseudo-second-order model were found to be higher than other kinetic models and the calculated values of q_e,cal estimated from the pseudo-second-order model equation were close to the experimental values of q_e,exp, suggesting that the adsorption process followed the pseudo-second-order model well.

As shown in Figure 5(b), the intraparticle diffusion model was applied to further explain the adsorption behavior of BPS onto the FRAC. The intra-particle diffusion fitting implied that the adsorption process included three stages. The first stage was the rapid adsorption stage, in which BPS rapidly diffused from the solution to the surface of FRAC with a faster rate. The second stage was the intra-particle diffusion stage, in which the BPS diffused from the FRAC surface to the interior, and the adsorption rate decreased with the increase of diffusion resistance. In the third stage, the final equilibrium of BPS on FRAC was achieved. From Table 4, the values of the adsorption rate constant k decreased with the prolongation of time under the same temperature, showing that the mass transfer rate decreased. However, the increased values of C revealed that the thickness of the boundary layer increased accordingly. Meanwhile, none of the lines passed through the origin, indicating that intra-particle diffusion was not the only rate-controlling step, and the adsorption rate of BPS onto FRAC was affected by both intra-particle diffusion and membrane diffusion.

The dosage of FRAC was an important parameter because the active sites were provided from the FRAC for the adsorption of BPS. The impact of FRAC dosage on the removal rate (%R) and equilibrium adsorption capacity (q_e) was investigated by varying the FRAC dosage, and the result is presented in Figure 6(a). From Figure 6(a), it was found that the removal rate increased from 72.95% to 98.90%, while q_e decreased from 2.9025 mmol/g to 0.1430 mmol/g with the increase of FRAC dosage. The improvement of %R was attributed to the more adsorption sites and bigger surface areas of FRAC. On the other hand, with the increasing FRAC dosage, aggregation of particles and the adsorption competition among FRAC took place, resulting in the decrease of q_e. Considering the q_e and %R of FRAC comprehensively, 0.2 g/L with a better %R (97.13) and q_e (0.9440 mmol/g) was selected as the optimal dosage for further experiments.

The pH was another key factor in the adsorption process, which might affect both the forms of BPS in aqueous phases and the surface net charge of FRAC. The distribution coefficient of BPS is depicted in Figure 7(a), and the pK_a of BPS was 8.2 (Ntuli & Hapazari 2013). When pH < pK_a, the uncharged BPS molecule was the dominant species in the aqueous solution. The proportion of anionic form (BPS⁻) increased due to the deprotonation of the hydroxy group at pH> 6. When the pH >10, the anionic form (BPS⁻) was the primary ionic form. The zero point charge (pH_pzc) of BPS was measured by salt addition method (Slimani et al. 2014). As illustrated in Figure 7(b), the pH_pzc of FRAC was found around 8.23. It suggested that the BPS had a positive charge on its surface at pH < 8.23 and a negative charge on its surface at pH> 8.23.

The influence of pH on the removal of BPS from aqueous solutions was studied, and the results are illustrated in Figure 8(a). According to Figure 8(a), the adsorption capacity of FRAC was in the range of 0.9355 mmol/g to 0.9655 mmol/g as the pH increased from 2 to 7. At pH <7, the FRAC exhibited a high adsorption capacity for BPS, which was presumably attributed to the hydrogen bonding between hydroxyl groups of BPS and the carbonyl groups of the FRAC as well as via bonding between the sulfonyl group of BPS and the hydroxyl groups of the FRAC. The electrostatic attraction between the BPS⁻ and the positively charged FRAC also promoted the adsorption process when pH < pH_pzc. In addition, π-π interaction between BPS and the FRAC also occurred to a certain extent. When the pH >7, the adsorption capacity declined sharply from 0.9355 mmol/g to 0.1845 mmol/g, which was mainly ascribed to the hydrogen bonds decreasing with the deprotonation of the BPS (Couto et al. 2020). Besides, the electrostatic repulsion between the BPS⁻ and negatively charged FRAC gradually increased with an increase of pH (Goyal et al. 2016). By experiments, the pH of the original BPS solution (pH = 5.77) showed a relatively high adsorption capacity. Therefore the pH of the original BPS solution was selected as a suitable pH for all other batch experiments.

To better elucidate the interaction mechanism of BPS onto FRAC, the adsorption behaviors of BPS were investigated at three different temperatures by varying the initial concentration of BPS. Experimental results of BPS adsorption were regressively simulated with Langmuir, Freundlich, Koble-Corrigan, and Redlich-Peterson isotherm models.

The nonlinear fitting curves of the isotherm model to experimental data are given in Figure 9(a) and 9(b), and the corresponding parameters are reported in Table 5. It could be observed from Figure 9(a) that a decrease of the adsorption capacity with increasing temperature was found, which was consistent with the conclusion obtained by adsorption kinetics. However, an opposite trend was observed between BPS concentration and BPS adsorption capacity. From Table 5, both the q_m and K_L decreased with the increasing temperature. K_L is the Langmuir constant related to the affinity between adsorbent and adsorbate. A higher value of K_L indicates a favorable adsorption process. Therefore, the values further verified that low temperature was favorable for the adsorption of BPS on FRAC. Moreover, the Langmuir model showed a higher R² (>0.9877) and lower χ² (<0.0716). Therefore, the adsorption process of BPS on FRAC was successfully represented by the Langmuir model. In addition, the R_L values presented in Figure 9(b) at three temperatures were all between 0 and 1, confirming that the FRAC was suitable for the removal of BPS from aqueous solutions. Moreover, the value of R_L at 298 K was lower than those at other temperatures, which further proved that the BPS adsorption was more propitious at a lower temperature. The maximum adsorption capacity obtained from the Langmuir isotherm was found to be 3.2848, 3.1254, and 2.9773 mmol/g at 298, 308, and 318 K, respectively.

The comparison of adsorption uptake of BPS onto different materials is given in Table S4. Compared with other materials reported in the literature, the FRAC had a relatively higher BPS adsorption capacity with a larger specific surface area. The high BPS adsorption capacity was attributed to the high specific surface area providing substantial accessible adsorption sites. Higher R² (>0.9864) and lower χ² (<0.0711) also could imply that the BPS adsorption process followed the Koble-Corrigan isotherm model. In contrast, the lower R² and higher χ² value for the Freundlich model indicated that it was inappropriate for describing the BPS adsorption process. Even though the R² values were higher than 0.98 for the Redlich-Peterson model, the values of g at the three temperatures were all greater than 1, indicating that the Redlich-Peterson model could not be applied to describe the adsorption process of BPS on FRAC. Overall, the adsorption behavior of BPS on FRAC could be well described by Langmuir and Koble-Corrigan isotherm models.

The values of ΔG for the BPS adsorption process were −37.72, −38.04, and −38.68 kJ/mol at 298, 308, and 318 K, respectively. The negative ΔG suggested that the adsorption process was favorable and spontaneous. The value of ΔH (−23.36 kJ/mol) revealed that the adsorption reaction was an exothermic process, which was confirmed from the increased adsorption amount of BPS with the decrease of temperatures. Furthermore, the degree of randomness at the FRAC-BPS interface during the adsorption process increased due to the positive ΔS (48.00 J/mol K) value.

The stability and reusability of FRAC are the key factors for industrial and practical applications. Five continuous adsorption-desorption cycles were conducted to study the reusability of FRAC, and the result is shown in Figure 10. In this study, the regeneration of the FRAC after BPS adsorption was performed by mixing BPS-loaded FRAC with 50 mL of ethanol for 180 min, and then the mixture was filtrated and the FRAC was dried at 105 °C. From Figure 10, the adsorption efficiency of FRAC was decreased from 99.21 to 97.23% after five cycles of reuse, which proved that FRAC was a promising adsorbent with excellent regeneration and reusability.

In this paper, a solid waste FR was utilized to prepare high-quality AC by steam activation for the removal of BPS from aqueous solutions. The RSM was successfully implemented to optimize the preparation conditions, and the optimum conditions have been identified as activation temperature of 922 °C, activation time of 62 min, and mass ratio of char to H₂O of 1:4.5. The optimum conditions resulted in the FRAC with a specific surface area up to 1,662.41 m²/g and adsorption capacity of iodine of 1,501.84 mg/g, respectively. Batch adsorption experiments showed that the adsorption kinetics data obeyed the pseudo-second-order model and the adsorption isotherm data were successfully represented by Langmuir and Koble-Corrigan isotherm models. Thermodynamic parameters confirmed the spontaneous and exothermic nature of the BPS adsorption process. The adsorption of BPS onto FRAC could be achieved through the hydrogen bond, π-π interaction, and electrostatic interaction. Furthermore, the FRAC was regenerated effectively by ethanol and the decrease in the adsorption efficiency was not significant after five cycles. The monolayer saturation adsorption capacity of BPS onto FRAC was 3.2848 mmol/g at 298 K, implying that the FRAC derived from FR could be a promising adsorbent because of the low cost and large adsorption capacity for the removal of BPS from aqueous solutions.

This project is funded by the Program of Processing and Efficient Utilization of Biomass Resources of Henan Center for Outstanding Overseas Scientists (No. GZS2018004).

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