All chemical reagents (viz., KMnO₄, Fe(NO₃)₃•9H₂O, KOH, NH₄HCO₃, K₂Cr₂O₇, H₃PO₄, H₂SO₄, HCl, C₂H₅OH, etc.) used in this study were of analytical grade and purchased from Shanghai Guoyao Group Chemical Reagent Co. Ltd., China. Stock solution [1,000 mg/L Cr(VI)] was prepared with K₂Cr₂O₇ and ultrapure water and was used to prepare diluted chromium-containing solutions for adsorption experiments. 1,000 mg/L of Cr(VI) standard solution was purchased from the National Nonferrous Metals and Electronic Materials Analysis and Testing Center, China, and stored at 4°C. Ultrapure water was used to prepare all solutions.

The preparation methods of MBC and MBC-MFC were based on Chinese invention patents (Liang et al., 2019).

Preparation of MBC: the peeled mulberry was crushed into particles of less than 2 mm and finally dried in an oven (80°C) until it reached the constant weight. The dry mulberry was pyrolyzed at 450–650°C in a muffle furnace at the heating rate of 5°C/min until it reached a set temperature and maintained specified temperature for 3–6 h. The final products were denoted as MBC, which was stored in airtight glass bottles.

Preparation of MBC-MFC: some MBC was added to a conical flask containing 100 ml of 9.0–15.0 mol/L NaOH solution and stirred for 12–36 h and then filtered and dried; the dried MBC was activated at 700–900°C for 1 h. After activation, MBC was washed with ultrapure water 3–5 times and then filtered, and the filter cake was placed in a porcelain dish and dried at 110°C for 6 h to obtain the MBC intermediate. 3–6 g MBC intermediate was added to a 500 ml conical flask and oscillated with an ultrasonic oscillator at a frequency of 25–45 kHz for 30 min; then, the conical flask was put into a magnetic constant temperature water bath at 25°C. Under the magnetic stirring condition, 50 ml of 0.05–0.15 mol/L KMnO₄ solution and 50 ml of 1.0–1.5 mol/L NH₄HCO₃ solution were added to the conical flask and stirred for 15 min; immediately after, 50 ml of 0.1–0.3 mol/L Fe(NO₃)₃ solution was added to the conical flask, and then the pH of mixture suspension was adjusted to 7.5–10.5 with 1.0 mol/L NaOH solution. Next, the conical flask was put into a water bath at 95°C reacted for 4–8 h, after being cooled to room temperature, filtered, and washed three times with ultrapure water and one time with absolute ethanol. The filter cake was placed in a porcelain dish and dried at 85°C for constant weight to get the MBC-MFC; then, the MBC-MFC was ground and sifted by a 200-mesh sieve.

RSM is employed to optimize the preparation conditions in this work to study the effects of pH, temperature, concentration, and time on T_Cr and Cr(VI) removal efficiency of MBC-MFC. RSM-PB was used to figure out the main factors from various factors that affect the removal rate of sorbent from lots of variables (He and Tan, 2006), and RSM-BBD was employed to optimize those main factors to obtain the best conditions for fabricating MBC-MFC with excellent performance to adsorb T_Cr and Cr(VI) (Liang et al., 2019).

The thermochemical pyrolysis temperature (Wang and Wang, 2019), residency time (Rogovska et al., 2012), heating rate (Rogovska et al., 2012), and heating method (Liang et al., 2017) of biochar have a great influence on the number of functional groups on biochar. Moreover, nanocomposite sorbents with nanosized metal oxyhydrates are attached to carbonaceous surfaces within biochar and increase the sorption of contaminants (Ahmed et al., 2016). In this paper, eight parameters for MBC-MFC preparation (Supplementary Table 1) have been evaluated for their effects on the removal rate of T_Cr and Cr(VI). All preparation factors in Supplementary Table 1 were set on the basis of previous single factor experiment to make the obtained results under critical conditions, such as the chosen variables ranges, batch experiment conditions, and the applied concentrations.

Based on the results of RSM-PB, we selected three main factors for MBC-MFC preparation and their effects on the removal rate of T_Cr and Cr(VI), which was described as Eq.1 (Grady and Kenneth, 2006). The three selected factors were defined as X₁, X₂, and X₃, respectively. They were coded at three levels, −1 (minimum), 0 (central), and +1 (maximum), covering the whole range of interest (Supplementary Table 2). The data in Supplementary Table 2 represent the factors and their level ranges in RSM-BBD. Y = β 0 + β 1 X 1 + β 2 X 2 + β 3 X 3 + β 12 X 1 X 2 + β 13 X 1 X 3 + β 23 X 2 X 3 + β 11 X 1 2 + β 22 X 2 2 + β 33 X 3 2 , (1) where Y is the estimated response (i.e., the removal rate of Cr(VI) and T_Cr); β₀ is the intercept; β₁, β₂, and β₃ are the linear coefficients; β₁₂, β₁₃, and β₂₃ are the squared coefficients; β₁₁, β₂₂, and β₃₃ are the quadratic coefficients.

The morphological structures of MBC and MBC-MFC (under the optimal conditions’ preparation) were analyzed by scanning electron microscopy combined with energy dispersive spectrum analysis (SEM-EDS) (Hitachi, JSM-7900F, Japan). X-ray diffractometer (XRD) (PANalytical, X’Pert PROX, Netherlands) and X-ray photoelectron spectroscopy (XPS) (Thermo Fisher Scientific, ESCALAB 250Xi, United States) analyses were used to identify the adsorbent constituents. A Fourier transform infrared spectrometer (FT-IR, PerkinElmer, Llantrisant, United Kingdom) was employed to collect spectra in the 400 and 4,000 cm⁻¹ regions to identify the surface functional groups. The surface zeta potentials of MBC-MFC and MBC were measured by a Nano-ZS90 apparatus (Malvern Panalytical, Nano-ZS90, United Kingdom). Specific surface area, pore size distribution, and pore volume were performed by a JW-BK200C apparatus (BET-N₂) (Beijing Jingwei Gaobo, JW-BK200C, Beijing, China).

The batch adsorption experiments were investigated by adding 0.10 g MBC-MFC to a series of 100 ml polyethylene plastic centrifuge tubes containing 50 ml of 45 mg/L Cr(VI) solutions, which were adjusted to a certain pH with 0.1 mol/L NaOH or 1.0 mol/L HCl. The centrifuge tubes were sealed by paraffin. Then they were shaken (at 180 rpm) for 48 h at 25°C. After being shaken for 48 h, samples were taken out and filtered through 0.22 um nylon membrane into a vial. The remaining Cr(VI) or T_Cr concentrations in the diluted were measured by ICP-OES. Unless otherwise noted, initial pH, contact time, the temperature, initial Cr(VI) concentration, and adsorbents dosage of all batch experiments were 3.0 ± 0.1, 48 h, 25°C, 60 mg/L, and 0.10 g/50 ml, respectively. The batch adsorption experiments with different pH values (2.0–11.0), different contact time (0.25, 0.5, 1, 2, 3, 4, 6, 8, 10, 12, 18, 24, 30, 36, and 48 h), different temperature (25, 35, and 45°C), and different MBC-MFC dosage (0.02, 0.05, 0.10, 0.15, 0.2, 0.25, 0.3, 0.35, 0.40 g/50 ml) were carried out. The batch adsorption experiments for optimization of preparation conditions through RSM were carried out too, in which the initial Cr(VI) concentration was 45 mg/L.

In order to test the reusability of MBC-MFC, three consecutive adsorption-regeneration cycles were performed in triplicate. During the adsorption process, 0.10 g of MBC-MFC was added to 50 ml Cr(VI) solution and shaken for 48 h. The regeneration method was carried out as follows: the used adsorbent was collected and gently washed with deionized water, dispersed into 50 ml of a regenerative solution (0.1, 0.3, and 0.5 mol/L HCl or NaOH), and shaken for 48 h. The concentration of desorbed Cr(VI) and T_Cr in the solution was also measured. Lastly, the adsorbent after desorption was collected and washed with deionized water about pH = 7.0 for the next cycle.

The optimal values of the selected variables were got by solving the equation (Eqs. 2, 3) using Design-Expert 8.0.6 software: A, B, C, D, E, F, G, and H were set at 5.70 mol/L and 9 mol/L, 450°C, 3 and 12 h, 790°C, C _Fe = 0.28 mol/L, C _Mn = 0.14 mol/L, 4 h, and 9.0, respectively. The theoretical removal rates predicted under the above conditions were 95.13 and 87.2% to Cr(VI) and T_Cr, respectively. However, the experiment mean removal rates were 96.47 and 88.63% to Cr(VI) and T_Cr, respectively. The relative deviation between the predicted value and the experimental value was 1.40 and 1.64%, respectively, proving the validity of our model.

According to the results measured by BET-N₂, the surface area of MBC and MBC-MFC was 179.29 and 318.53 m²/g, respectively, and the total pore volume of MBC-MFC (0.337 cm³/g) was 2.6 times that of MBC (0.132 cm³/g). Increasing surface area and total volume are considered as one of the favorable conditions for improving the adsorption capacity of MBC-MFC.

The SEM-EDS images (Figure 2) revealed that there were lots of pores with different sizes on the surfaces of MBC and MBC-MFC. Some Fe-Mn oxides were loaded on the surface of MBC-MFC with the porous structure not being disrupted (Figure 2B), which will be of great benefit to Cr(VI) and T_Cr adsorption. The chemical composition of MBC-MFC showed that the elemental molar ratio of Fe to Mn was close to 2:1, the same as that of the preparation precursor. The EDS analysis showed that the content of O (17.75%) in MBC-MFC was significantly higher than that of MBC (13.46%), suggesting that MBC-MFC owned more adsorption sites for heavy metal complex adsorption (Liu et al., 2020a).

XRD patterns of MBC and MBC-MFC are illustrated in Figure 3A. As shown, the peaks of the MBC-MFC at 24.13°, 33.12°, 35.60°, 40.82°, 49.41°, 54.00°, and 62.38° were corresponding to lattice planes of (012), (104), (110), (113), (024), (116), and (214) of Fe₂O₃ (JCPDS NO: 089–0,597), and the peaks at 20.35°, 30.92°, 33.53°, 41.75°, and 45.01° were corresponding to lattice planes of (021), (110), (023), (131), and (132) of Mn₃O₄ (JCPDS NO: 065–1,123), confirming the presence of crystalline Fe₂O₃ and Mn₃O₄ within MBC-MFC. In addition, the broad peak around at 2θ = 26.60° was corresponding to amorphous carbon (Liu et al., 2020b). The XPS spectra of MBC and MBC-MFC are shown in Figure 3B. The distinct peaks of Fe and Mn could only be observed on MBC-MFC, which confirmed the successful fabrication of Fe₂O₃ and Mn₃O₄ on MBC.

The FT-IR spectra of MBC and MBC-MFC were as shown in Figure 3C. The peak appearing at 3,391–3,422 cm⁻¹ and 1,569–1,639 cm⁻¹ was related to the stretching of -OH groups and C=O, respectively, indicating that both MBC and MBC-MFC contained C=O and -OH groups. The bending vibration at about 1,383 cm⁻¹ was mainly contributed by the stretching vibration of Fe-O-Mn, and the bands at 1,118 cm⁻¹ were associated with the bending vibration of hydroxyl groups of metal oxides (M-OH) (Du et al., 2017). MBC-MFC showed additional broadband at 595 cm⁻¹, relating to the stretching vibration of Fe-O-Fe (Luo et al., 2013). Therefore, the SEM-EDS, FT-IR, and XRD spectra confirmed that iron oxide and manganese oxide were successfully coated on the surface of MBC-MFC.

The pH effects on the surface electronegativity and corresponding Cr(VI) adsorption of adsorbent were studied. When the pH ≤ 2.0, Cr contaminant mainly existed in the form of H₂CrO₄. With the increase of pH (2.0 ≤ pH ≤ 6.4), it mainly existed in the form of HCrO₄ ⁻ and Cr₂O₇ ²⁻. By further increase in the pH (pH ≥ 6.4), it mainly existed in the form of CrO₄ ²⁻ Liu et al., 2021. As illustrated in Supplementary Figure 2, the pH_ZPC of MBC-MFC was 5.64. When pH < pH_ZPC (5.64), the functional groups of absorbent were protonated and positively charged, promoting the adsorption of chromium ions onto the MBC-MFC surface. When pH > pH_ZPC (5.64), the surface of the adsorbent was negatively charged, which weakens the attraction for Cr(VI). In addition, a higher pH would tend to generate more OH⁻, which would occupy some of the adsorption sites, thereby reducing the amount of Cr(VI) adsorption on MBC-MFC (Chen et al., 2018).

The effects of pH on the adsorption of Cr(VI) by MBC-MFC are shown in Figure 4A. In Figure 4A, when the initial Cr(VI) concentration was set at 30 and 60 mg/L, the adsorption capacity of Cr(VI) increased slowly as the pH raised from 2.0 to 3.0. The uptake of cationic metal ions by biochar at pH 2.0 was lower than that at pH 3.0 (Senthilkumar et al., 2019). Further enhancement of equilibrium pH negatively influenced the adsorption capacity of MBC-MFC (3.0 < pH < 9.0). It shows that ion exchange and electronic interaction may be involved in the adsorption mechanism of Cr(VI) (Choudhary and Paul, 2018). The optimal pH value for MBC-MFC to adsorb Cr(VI) varies with the initial Cr(VI) concentration. When the initial Cr(VI) concentration was 30 mg/L, the best initial pH range was 2.0–5.0. When the initial Cr(VI) concentration was 60 mg/L, the best initial pH range was 2.0–3.0. When the pH value was in the range of 8–11, the adsorption capacity of Cr(VI) was relatively low and basically remains unchanged, which was similar to the results of Chen et al. (2018). The pH of the subsequent experiments was 3.0.

The effects of contact time on adsorption are shown in Figure 4B. It can be seen from Figure 4B that, with the increase of the adsorption time, the adsorption capacity of Cr(VI) increases, but with the extension of time, the adsorption capacity gradually reaches equilibrium. In the initial stage of adsorption, due to the large number of pores on the surface of MBC-MFC, it can provide more binding sites, and it was easy to adhere to the surface of MBC-MFC after adsorption. As the contact time increases, the number of adsorption binding sites decreases, making it difficult for the adsorption reaction to continue, thus reaching adsorption equilibrium (Zhang et al., 2018). When the initial concentrations were 30, 45, and 60 mg/L, the adsorption equilibrium was reached after 420, 960, and 1,440 min, and the removal rates were 99.2, 99.7, and 99.3%, respectively. With the increase of Cr(VI) concentration, the electrostatic repulsion between ions increases, so the adsorption equilibrium time was prolonged (Mondal et al., 2007; Park, 2020). After reaching the adsorption equilibrium, the equilibrium adsorption capacity of initial concentrations of 30, 45, and 60 mg/L was about 14.97, 22.46, and 29.97 mg/g, respectively.

The effects of temperature on adsorption are shown in Figure 4C. As shown in Figure 4C, the maximum adsorption capacity of MBC and MBC-MFC for Cr(VI) was 12.84 mg/g and 54.31 mg/g at 25°C, respectively. Compared with MBC, the adsorption capacity of MBC-MFC on Cr(VI) has been greatly improved, which was 4.16 times that before modification. At 25, 35, and 45°C, the equilibrium adsorption capacity of MBC-MFC for Cr(VI) was about 54.31, 58.57, and 62.60 mg/g, respectively.

The effects of MBC-MFC dosage on chromium adsorption are shown in Figure 4D. In Figure 4D, when the initial concentration of Cr(VI) was set at 30 mg/L, with the increase of the dosage from 0.02 g/50 ml to 0.40 g/50 ml, the adsorption capacity decreased from 46.97 to 3.75 mg/g, and when the initial concentration of Cr(VI) was set at 60 mg/L, the adsorption capacity decreased from 68.50 to 7.45 mg/g. The adsorption capacity of Cr(VI) decreased with the increase of the dosage, while the removal rate of Cr(VI) increases with the increase of the dosage. When the initial concentration of Cr(VI) was set at 60 mg/L, the removal rate increased from 45.66 to 99.98%. When the dosage was greater than 0.10 g/50 ml, the removal rate did not increase. The reason may be that, with the increase of the dosage, the total reaction sites on the sorbent surface were increased (Tang et al., 2021). Therefore, the effective dosage for Cr(VI) was selected at 0.10 g/50 ml for the latter experiments.

In this study, we employed a pseudo-first-order kinetic model and pseudo-second-order kinetic model to study the Cr(VI) adsorption kinetic onto MBC-MFC and MBC. The fitting plots are shown in Figures 5A,B, respectively, and the kinetic parameters are summarized in Supplementary Table 7. The pseudo-second-order kinetic model better fitted the experimental data of the Cr(VI) and T_Cr removal by the MBC-MFC and MBC (R ² = 0.957–0.989) and the Cr(VI) equilibrium adsorption capacities (14.97, 14.96, 9.61, and 9.50 mg/g) calculated by the pseudo-second-order kinetic model was approximately equal to the Cr(VI) experimental adsorption capacities data (14.98, 14.97, 8.90, and 8.82 mg/g) (Pap et al., 2018).

Figure 6 shows the application of Langmuir and Freundlich isotherms to the Cr(VI) adsorption on MBC-MFC and MBC. Supplementary Table 8 lists the isotherms parameters. The Langmuir model better fitted the experimental data of the MBC-MFC and MBC (R ² = 0.909–0.998) than the Freundlich model, which may imply monolayer adsorption of Cr(VI) and T_Cr (Wen et al., 2018). The MBC-MFC exhibited superior T_Cr and Cr(VI) adsorption capacity than MBC. The maximum T_Cr and Cr(VI) adsorption capacities (q_m) of the MBC-MFC reached 54.97 and 56.18 mg/g, respectively. It was about 4.11 times that of MBC, which is 5.61 times that of dolomite adsorbent (Albadarin et al., 2012), 2.64 times that of Eucalyptus globulus bark biochar (Choudhary and Paul, 2018), 8.03 times that of biochar derived from municipal sludge (Chen et al., 2015), slightly higher than 2.64 times that of Eucalyptus globulus bark biochar (Zhang et al., 2018), and 2.14 times of Fe/Mn metal oxide nanocomposites (Yacou et al., 2018; Supplementary Table 9). The adsorption capacity of Cr(VI) was calculated based on the remaining amount of Cr(VI) in solution, while the adsorption capacity of T_Cr was calculated based on the remaining amount of chromium in solution. According to our result, the maximum adsorption capacity of Cr(VI) was 2.20% higher than that of T_Cr, indicating the existence of a little Cr(III) in the solution. The Cr(III) in solution after adsorption was originated from the redox reaction of Cr(VI) (Lv et al., 2012).

Strong acids and base are the most common agents used to elute heavy metals, and the choice of eluent depends on the adsorption mechanism of heavy metals and the nature of the adsorbents (Markovski et al., 2014). The results of desorption experiments are shown in Figure 7A. It can be seen from Figure 7A that 0.1, 0.3, and 0.5 mol/L HCl were used as the desorption solvent, and the adsorption capacity of Cr(VI) by the adsorbent after desorption was 23.03, 26.17, and 27.67 mg/g, respectively. However, 0.1, 0.3, and 0.5 mol/L NaOH were used as the desorption solvent, and the adsorption capacity of Cr(VI) by the adsorbent after desorption was 28.79, 29.43, and 29.43 mg/g, respectively. Therefore, NaOH solution performed a better desorption effect than HCl solution as the regeneration reagents. Considering the cost factor, a 0.3 mol/L NaOH solution was selected as the desorption solution for the desorption regeneration of MBC-MFC.

The MBC-MFC desorption and resorption experiment is shown in Figure 7B. It can be seen from Figure 7B that as the number of desorption cycles increases, the adsorption amount of Cr(VI) also decreases. It can be attributed to the following reasons: ① The specific surface area (effective adsorption point) of the adsorbent was gradually decreasing during the adsorption-desorption cycle. ② During the adsorption-desorption cycle, the functional groups on the surface of the adsorbent were gradually weakened, but there were still a certain number of functional groups that play a role in adsorption (Li et al., 2016). After three cycles of adsorption/desorption experiment, the adsorption capacity of Cr(VI) showed little effect on Cr(VI) adsorption, which was only slightly reduced from 29.95 to 26.81 mg/g.

It shows that MBC-MFC can be regenerated and reused by NaOH solution, which also fully shows that the interaction between adsorbent and adsorbate mainly depends on electrostatic adsorption, which was consistent with the aforementioned characterization analysis and kinetic experiment analysis results (Yu et al., 2020).

Figure 8A presents the survey XPS spectra of MBC-MFC before and after Cr(VI) adsorption. The element of Cr (576.2; 578.4 eV) was absent in the MBC-MFC result. However, it appeared on the XPS spectra of MBC-MFC after Cr(VI) adsorption, demonstrating the successful adsorption of Cr onto MBC-MFC. To further understand the Cr(VI) adsorption on MBC-MFC, high-resolution Cr2p spectra were measured, shown in Figure 8B. Obviously, no chromium peak was found on MBC-MFC before Cr(VI) adsorption. The broad peak of Cr2p_3/2 was found on the XPS result of MBC-MFC, which could be deconvoluted into peaks at binding energies of 578.4 and 576.2 eV, corresponding to the characteristic peaks of Cr(VI) and Cr(III) (Chen et al., 2017; Zhong et al., 2020). The areas of the two peaks indicate that Cr(VI) is partially reduced to Cr(III) (Wang et al., 2019). The high-resolution Cr2p spectra demonstrated the coexistence of Cr(III) and Cr(VI) on the microspheres of MBC-MFC. MBC-MFC can adsorb both Cr(III) and Cr(VI). We believe that there was a reduction of Cr(VI) to Cr(III) in the experiments.

In order to understand the adsorption mechanism of Cr(VI) on MBC-MFC, Fe2p and Mn2p peaks of MBC-MFC and after Cr(VI) adsorption were measured (Figures 8C–F). The observed two asymmetric peaks located at 710.3 and 724.3 eV could be ascribed to Fe2p_3/2 and Fe2p_1/2, respectively. For Fe2p_3/2, the peak at 710.3, 712.4, and 717.5 eV was attributed to Fe(II) and two forms of Fe(III), respectively (Li et al., 2019b). The binding energy of 710.0 eV (Fe2p_3/2) of Fe(II) peaks was shifted to 709.70 eV after Cr(VI) adsorption (Figures 8C,E), implying that the Cr(VI) adsorption occurred through bidentate ligand or monodentate interaction with iron oxide (Peng et al., 2019). The observed two asymmetric peaks located at 641.4 and 640.3 eV could be ascribed to Mn2p_3/2, respectively (Figure 8D). The peak at 641.4 eV was ascribed to Mn(IV), and another peak at 640.3 eV was related to Mn(II) and Mn(III) in different forms (Li et al., 2019b; Lin et al., 2017). The binding energy of 641.4 eV (Mn2p_3/2) of Mn2p peaks was reduced by 0.3 eV after Cr(VI) adsorption (Figure 8F), suggesting that the Cr(VI) adsorption occurred through surface complexation with Mn (Xiong et al., 2017).

In addition, chromium (CrO₄ ²⁻, Cr₂O₇ ²⁻, and HCrO₄ ⁻) could be also adsorbed on the surface of MBC-MFC by electrostatic interaction or forming outer-sphere surface complexes. As illustrated in Supplementary Figure 2, by regulating the solution pH to be lower than the pH_zpc value of MBC-MFC (5.64), the zeta potential of MBC-MFC was positive and desirable for adsorption of chromium anions by electrostatic attraction.