Adsorption behaviors and mechanisms of Cu2+, Zn2+ and Pb2+ by magnetically modified lignite

The study aims to solve the problems of limited capacity and difficult recovery of lignite to adsort Cu2+, Zn2+ and Pb2+ in acid mine wastewater (AMD). Magnetically modified lignite (MML) was prepared by the chemical co-precipitation method. Static beaker experiments and dynamic continuous column experiments were set up to explore the adsorption properties of Cu2+, Zn2+ and Pb2+ by lignite and MML. Lignite and MML before and after the adsorption of heavy metal ions were characterized by scanning electron microscopy (SEM), X-ray diffraction (XRD) and Fourier transform infrared spectrometer (FTIR). Meanwhile, the adsorption mechanisms of Cu2+, Zn2+ and Pb2+ by lignite and MML were revealed by combining the adsorption isotherm model and the adsorption kinetics model. The results showed that the pH, adsorbent dosage, temperature, initial concentration of heavy metal ions, and contact time had an influence on the adsorption of Cu2+, Zn2+ and Pb2+ by lignite and MML, and the adsorption processes were more in line with the Langmuir model. The adsorption kinetics experiments showed that the adsorption processes were jointly controlled by multiple adsorption stages. The adsorption of heavy metal ions by lignite obeyed the Quasi first-order kinetic model, while the adsorption of MML was chemisorption that obeyed the Quasi second-order kinetic model. The negative ΔG and positive ΔH of Cu2+ and Zn2+ indicated the spontaneous and endothermic nature reaction, while the negative ΔH of Pb2+ indicated the exothermic nature reaction. The dynamic continuous column experiments showed that the average removal rates of Cu2+, Zn2+ and Pb2+ by lignite were 78.00, 76.97 and 78.65%, respectively, and those of heavy metal ions by MML were 82.83, 81.57 and 83.50%, respectively. Compared with lignite, the adsorption effect of MML was better. As shown by SEM, XRD and FTIR tests, Fe3O4 was successfully loaded on the surface of lignite during the magnetic modification, which made the surface morphology of lignite coarser. Lignite and MML removed Cu2+, Zn2+ and Pb2+ from AMD in different forms. In addition, the adsorption process of MML is related to the O–H stretching vibration of carboxylic acid ions and the Fe–O stretching vibration of Fe3O4 particles.


Results and discussion
Effect of pH, adsorbent dose and temperature. Effect of pH. According to the actual pH of AMD, the effects of different pH (2)(3)(4) on the adsorption process of Cu 2+ , Zn 2+ and Pb 2+ by lignite and MML were studied as shown in Fig. 1a,b. With the increase of pH value, the adsorption of Cu 2+ , Zn 2+ and Pb 2+ by lignite and MML gradually increases. When pH = 4, the removal rates of heavy metal ions reaches the maximum. Among them, the removal rates of Cu 2+ , Zn 2+ and Pb 2+ by lignite are 83.02, 78.79 and 80.34%, respectively, and those of heavy metal ions by MML are 91.61, 89.60 and 98.00%, respectively. It can be seen that MML has higher adsorption capacity than lignite. In addition, the results show that pH has a great influence on the adsorption process. Because large amount of H + in a strong acidic solution would compete with heavy metal cations for adsorption sites, leading to the low removal rates under lower pH value. Therefore, the optimum pH value is 4.
Effect of adsorbent dose. The effects of different adsorbent dosage on the adsorption process of Cu 2+ , Zn 2+ and Pb 2+ were studied as shown in Fig. 1c,d. It can be seen that with the increase of adsorbent amount, the removal rates also increases. For MML, when the adsorbent amount is 4 g/L, the removal rates of heavy metal ions exceeds 89%. When the amount of MML is increased to 6 g/L, the reaction tends to be balanced, and the removal rates is not significantly improved. Therefore, considering the adsorption effect and economic cost, the optimum adsorbent amount is 1 g/L.
Effect of temperature. Temperature is an important parameter in the adsorption process. The effects of temperature on the adsorption process of heavy metal ions at different temperatures (298.15, 308.15 and 318.15 K) were shown in Fig. 1e,f. The removal rates of Cu 2+ and Zn 2+ by lignite and MML increases with the increase of temperature, indicating that the adsorption process is endothermic. However, the removal rate of Pb 2+ decreases with the increase of temperature, which indicates that the adsorption process of Pb 2+ by lignite and MML is exothermic. Although the removal rates of Cu 2+ and Zn 2+ by lignite and MML reaches the maximum at 318.15 Initial concentration and adsorption isotherm. Effect of initial concentration. The adsorption capacity and removal rates of lignite and MML for different initial concentrations of heavy metal ions were shown in Fig. 2. The adsorption capacity of heavy metal ions by lignite and MML increases with the increase of initial concentration. This is because the higher the initial concentration of heavy metal ions, the higher the chance of collisions with adsorption sites on the surface of the adsorbent. Moreover, the driving force of mass transfer is better, which is conducive to reduce the mass transfer resistance and increase the adsorption capacity 26 . However, the removal rate of heavy metal ions by lignite and MML decreases with the increase of initial concentration. Especially when the initial concentration of Cu 2+ , Zn 2+ and Pb 2+ are 30, 30, and 50 mg/L respectively, the slope of the removal rate curve increases significantly. This is because for the fixed amount of adsorbent, the number of adsorption sites on the surface is limited, and the adsorption effect will achieve the best adsorption at a certain concentration of heavy metal ions.
Comparing the adsorption effects of lignite and MML on heavy metal ions, it can be seen that the adsorption capacity and removal rates of Cu 2+ , Zn 2+ and Pb 2+ by MML are higher than that of lignite at the same concentrated  www.nature.com/scientificreports/ heavy metal ion degree. In the initial concentration range of 10-90 mg/L, the removal rates of Cu 2+ and Zn 2+ by lignite and MML shows similar trends, and the difference in removal rates tend to increase. In contrast, the removal rates of Pb 2+ by MML in the range of 10-50 mg/L is almost constant with increasing initial concentration, indicating that Pb 2+ in AMD is well removed by MML in this range. In addition, In the initial concentration range of 10-90 mg/L, the difference of adsorption of Cu 2+ , Zn 2+ and Pb 2+ between MML and lignite increases with the increase of initial concentration. This phenomenon indicates that MML has greater adsorption potential than lignite when the concentration of heavy metal ions in AMD solution is higher.
Adsorption isotherm. The adsorption isotherm describes the relationship between the adsorbent and the amount of analytical substances in the solution 27 . To clarify the mechanism of Cu 2+ , Zn 2+ and Pb 2+ adsorption by lignite and MML, the Langmuir model and Freundlich model were used to fit the experimental data. The Langmuir model assumes that monolayer adsorption occurs on the uniform adsorbent surface, with no interaction between adsorbates. The Langmuir model is expressed in the following form 28 : where q m and q e are the maximum adsorption capacity and the adsorption capacity at equilibrium (mg/g), respectively, C e is the adsorbate concentration in solution at equilibrium (mg/L), and K L is the Langmuir adsorption constant (L/mg). The values of q m and K L can be calculated by a linear relationship. In addition, the equilibrium constant R L of the Langmuir model can be used to describe the adsorption effect of the adsorption process. The R L equation is of the following form 29 : where C 0 is the initial concentration of metal ions. The value R L < 1 indicates good adsorption performance.
Based on multilayer adsorption on non-homogeneous surfaces, the empirical Freundlich equation without assumptions is expressed in the following form 30 : where K F (L/mg) and n (dimensionless) are constant indications of the adsorption capacity.
The adsorption isotherms and corresponding parameters of Cu 2+ , Zn 2+ and Pb 2+ adsorption by lignite and MML were shown in Fig. 3 and Table 1, respectively. The correlation coefficient (R 2 > 0.99) of the Langmuir model is higher, indicating that the processes of Cu 2+ , Zn 2+ and Pb 2+ adsorption by lignite and MML are more consistent  31 . This indicates that a monolayer adsorbent is formed on the surface of lignite and MML, and no further adsorption will be carried out after the surface is completely covered 32,33 . In addition, the R L value obtained according to Langmuir adsorption constant K L is between 0 and 1, indicating that lignite and MML have good adsorption of heavy metal ions 34 . In Freundlich model, the 1/n value less than 1 also confirms that the adsorption conditions is good, and a smaller 1/n value indicates that MML is more favorable for the adsorption of Cu 2+ , Zn 2+ and Pb 2+ in the solution 35,36 . By comparing the parameters of lignite and MML adsorption isotherms, it can be seen that the Langmuir model of MML has a large correlation coefficient, which may be due to the uniform specific adsorption sites generated in the magnetization process 37 . It has been reported that the larger the adsorption capacity of the adsorbent, the greater the K F value 38 . The K F value of MML in this study is larger than that of original lignite, indicating that the adsorption capacity of MML is larger than that of lignite. The maximum adsorption capacity of MML for Cu 2+ , Zn 2+ and Pb 2+ are 16.2127, 15.8053 and 18.3870 mg/g, respectively, while the maximum adsorption capacities of lignite are13.3618, 14.7929 and 17.5274 mg/g, respectively. It is further confirmed that magnetic modification of lignite improves the adsorption capacity of Cu 2+ , Zn 2+ and Pb 2+ .

Contact time and adsorption kinetics.
To clarify the adsorption mechanism of lignite and MML, the adsorption kinetics of Cu 2+ , Zn 2+ and Pb 2+ by lignite and MML were analyzed by Quasi first-order kinetic model, Quasi second-order kinetic model, Elovich model and Intra-particle diffusion model.
Quasi first-order model. Lagergren proposed an adsorption analysis method based on solid adsorption capacity 20 , which is the Quasi first-order kinetic equation in the following form 29,31 : where q e and q t are the amounts of adsorbed metal ions at equilibrium and at time t (mg/g), respectively, and k 1 is the Quasi first-order rate constant (min −1 ).
Quasi second-order model. The Quasi second-order kinetic model is based on the assumption that the adsorption rate is controlled by chemisorption 24 . The Quasi second-order kinetic model is expressed in the following form 39 : where q e and q t are the amounts of adsorbed metal ions at equilibrium and at time t (mg/g), respectively, and k 2 is the Quasi second-order rate constant (min −1 ).
The results of Quasi first-order model and Quasi-second-order kinetic fits for the adsorption of Cu 2+ , Zn 2+ and Pb 2+ by lignite and MML were shown in Fig. 4 and Table 2.
From Fig. 4a,b, it can be seen that the tangent slope of the curve is larger at the beginning of the adsorption, indicating that the adsorption rate of MML and lignite to adsorb heavy metal ions is faster. Then the slope gradually decreases and the adsorption rate decreases. This is because there are enough effective adsorption sites on the surface of the adsorbent at the initial stage. As the reaction progresses, the adsorption sites are gradually occupied, resulting in the reduction of the adsorption efficiency. On the other hand, the experimental results ( Fig. 4a) show that for the single-component mode, as reported by Jellali et al. 18 , the adsorption efficiency of the studied heavy metal ions by lignite is as follows: Pb 2+ > Cu 2+ > Zn 2+ . However, for the single-component mode, the adsorption efficiency of the studied heavy metal ions by MML is as follows: Pb 2+ > Zn 2+ > Cu 2+ (Fig. 4b). This may be because the magnetic modification affected the physico-chemical properties of the lignite, such as the donor atoms abundance (oxygen, nitrogen, sulfur) 40 .
As can be seen from Table 2, the Quasi first-order kinetic parameters R 2 for the adsorption of Cu 2+ , Zn 2+ and Pb 2+ by lignite are higher (R 2 > 0.96), indicating that the adsorption process followe the Quasi first-order kinetic model and is dominated by physisorption 41 . The fitted equations of the Quasi first-order kinetics model of lignite for Cu 2+ , Zn 2+ and Pb 2+ are: y = 9.2602*(1 − e −0.00607x ), y = 10.2839*(1 − e −0.00468x ), and y = 11.8456*(1 − e −0.01265x ), respectively. In contrast, the Quasi second-order kinetic parameter R 2 for the adsorption of heavy metal ions by MML is higher than that of the Quasi first-order kinetic, indicating that the Quasi second-order kinetic model fits well the experimental data for three heavy metal ions 25 . Moreover, the Quasi second-order kinetic equilibrium adsorption capacity is closer to the experimental adsorption capacity. Therefore, under the used experimental conditions, the Quasi second-order kinetic model is more suitable for fitting the adsorption process of Cu 2+ , Zn 2+ and Pb 2+ by MML. The Quasi second-order kinetic model shows that the adsorption process of the studied heavy metal ions by MML is mainly chemisorption, and the adsorption rate is affected by the coordination between the surface active site of adsorbent and the heavy metal ions 42 . The fitted equations for the Quasi secondorder kinetics model of MML for Cu 2+ , Zn 2+ and Pb 2+ are: y = 0.09599x + 8.81861, y = 0.09333x + 10.01582, and y = 0.05103x + 5.02836, respectively.
Elovich model. The Elovich model assumes that with the increase of the amount of heavy metal ions, the adsorption rate decreases exponentially, following a chemisorption mechanism. The model is expressed in the following form 43 :  The experimental data were fitted by the Elovich model, and the results were shown in Fig. 4 and Table 2. As can be seen, Elovich model is in good agreement with the experimental data of adsorption of Cu 2+ , Zn 2+ and Pb 2+ by lignite and MML, and R 2 is between 0.87 and 0.97 (Fig. 4e,f). It shows that there is chemisorption between adsorbents (lignite and MML) and three kinds of heavy metal ions.
Intra-particle diffusion model. The adsorption process usually involves two main mechanisms: film diffusion and particle diffusion. In order to determine the way of metal ions entering the adsorbent material from the solution, the Intra -particle diffusion model (Eq. 7) was used to determine the adsorption rate control steps and the results are shown in Fig. 4 and Table 2 44 .
where q t is the amount of metal ions adsorbed at any moment t (mg/g), k 3 is the diffusion rate constant within the particle (min −1 ), and C is the constant involving thickness and boundary layer. The larger the value of C, the greater the contribution of the boundary layer. Figure 4g,h show the linear relationship between q t and t 1/2 . Among them, the parameters of the Intra-particle diffusion model for Cu 2+ , Zn 2+ and Pb 2+ adsorption by lignite and MML were shown in Table 2. According to reports, if the plots are linear and pass through the origin, indicating that Intra-particle diffusion is the only rate control step; if the linear plot of the fitted results does not pass through the origin, indicating that the adsorption rate is also controlled by other adsorption stages 45 . As can be seen in Fig. 4, the fitted results for the adsorption of Cu 2+ , Zn 2+ and Pb 2+ by lignite and MML are linear and do not pass the origin, indicating that the rates of Cu 2+ , Zn 2+ and Pb 2+ adsorption by lignite and MML are jointly controlled by multiple adsorption stages. In addition, linearity indicates spontaneous utilization of available adsorption sites on adsorbent surfaces 38 .
Adsorption thermodynamics. Based on the adsorption experimental data affected by temperature, the thermodynamic parameters (Gibbs free energy changes, ΔG; entropy, ΔS and enthalpy change, ΔH) determined in the following forms 37 : where R is the ideal gas constant (8.314 J/mol·K), T is the Kelvin temperature (K), and K is the thermodynamic equilibrium constant.
According to the experimental results, the thermodynamic parameters were determined, as shown in Fig. 5 and Table 3. The positive ΔH values of Cu 2+ and Zn 2+ indicates that the adsorption of Cu 2+ and Zn 2+ by lignite and MML is endothermic adsorption, while the negative ΔH values of Pb 2+ indicates that the adsorption is exothermic adsorption. With the increase of temperature, the negative ΔG values of Cu 2+ and Zn 2+ become more negative, which indicates that the adsorption efficiency is higher at higher temperature and the adsorption is spontaneous. The negative ΔG values of Pb 2+ tend to positive, indicating that the adsorption is spontaneous but its adsorption efficiency is lower at higher temperature. The results show that the temperature increase is beneficial to the adsorption of Cu 2+ and Zn 2+ by lignite and MML, but not conducive to the adsorption of Pb 2+ . In addition, the ΔG values of an adsorbent for the adsorption of some heavy metal ions at different temperatures are very close, which indicates that the adsorption process of Cu 2+ , Zn 2+ and Pb 2+ by lignite and MML is not obviously affected by temperature. were used to analyze the adsorption process. The desorption agents commonly used for desorption of adsorbents are NaOH, HCl, HNO 3 , EDTA, CaCl 2 and organic solvents such as methanol and ethanol 46 . In this study, 0.1 mol/L H 2 SO 4 was used as desorption agent for desorption experiment. Figure 6 shows that the desorption rates of Cu 2+ and Pb 2+ are less than 50%, indicating that the retention of MML for both metals is very strong under acidic conditions. Therefore, the adsorption process of Cu 2+ and Pb 2+ by MML is chemisorption. This result is consistent with the adsorption kinetics. However, the desorption rate of Pb 2+ is as high as 84.62%, which may be due to the removal of Zn 2+ in the form of Zn(OH) 2 in solution. When the pH value of the desorption   Fig. 7. The dynamic removal effects of both lignite and MML on Cu 2+ , Zn 2+ and Pb 2+ shows a similar trend (Fig. 7). Heavy metal ions are removed rapidly in the first 13 days, with removal rates of Cu 2+ , Zn 2+ and Pb 2+ exceeding 95, 92 and 97%, respectively. Then the removal rates gradually decreases from the 13th day to the 22nd day, and the removal rates is only about 10% at the 22nd day. This phenomenon is attributed to the fact that there are enough binding sites on the surfaces of adsorbent lignite and MML for metal ions to occupy at the initial stage, which make the adsorption process easier. However, the number of effective adsorption sites on the surfaces of lignite and MML gradually consumes with time, resulting in a decrease in the removal rates. During the whole dynamic removal cycle, the average removal rates of Cu 2+ , Zn 2+ and Pb 2+ by lignite are 78.00, 76.97 and 78.65%, respectively, and the average removal rates of the studied heavy metal ions by MML are 82.83, 81.57 and 83.50%, respectively. Apparently, magnetic modification increases the adsorption capacity of heavy metal ions by the lignite. This may be because the surface of MML is loaded with Fe 3 O 4 particles, which increases the specific surface area of lignite 23 . In addition, the adsorption capacity of the three heavy   Fig. 8. From Fig. 8a,b, it appears that the raw lignite presents a smooth and porous surface with a lager pore size. However, MML presents a slightly rough surface, which is mainly because the successful loading of Fe 3 O 4 on the lignite surface. A large number of Fe 3 O 4 particles were scattered on the surface of lignite, which increased the specific surface area of the lignite and facilitated the removal of Cu 2+ , Zn 2+ and Pb 2+ by MML. From Fig. 8c, d, surface voids of lignite after adsorption of Cu 2+ , Zn 2+ and Pb 2+ are filled. Compared with lignite, the granular material on the surface of MML was significantly increased, indicating that more sediment was generated on the surface of MML.
XRD analysis. The XRD test results of lignite and MML before dynamic test are shown in Fig. 8e. The XRD patterns of lignite have relatively wide diffraction peaks (Fig. 8e) The occurrence of Zn(OH) 2 may be due to the increase of pH value caused by the consumption of H + in the reaction process. In addition, PbS diffraction peaks appeared at 2θ = 30.1° and 43.1°. This may be due to the fact that lignite contains a certain amount of S 2− , which is dissolved and released as the reaction progresses and reacted with the free Pb 2+ in AMD. The results show that MML can remove Pb 2+ and S 2− at the same time, and prevent the oxidation of S 2− in lignite to SO 4 2− , causing secondary pollution. The appearance of phases such as CuFe 2 O 4 , ZnFe 2 O 4 , Zn(OH) 2 and PbS confirms that the adsorption process of Cu 2+ , Zn 2+ and Pb 2+ by MML is mainly chemisorption.
FTIR analysis. The lignite and MML before and after the dynamic test were taken for FTIR detection, and the results are shown in Fig. 8g,h. The peaks of lignite and MML at 3400 cm −1 was caused by O-H stretching vibrations of carboxylic acid groups, and the peak at 2920 cm −1 was attributed to the presence of -CH 2 in the stretching of aliphatic compounds, and the peak at 1600 cm −1 was related to the stretching vibrations of carboxylic acid functional groups 47 (Fig. 8f, h). Some peaks of lignite changed after magnetic modification, especially the generation of new peaks at 584 cm −1 , was attributed to Fe-O stretching vibrations of Fe 3 O 4 particles 48 . It shows that Fe 3 O 4 was successfully loaded onto the lignite surface, which is consistent with the XRD results.
After the dynamic test, the peak position and intensity of some functional groups in lignite changed slightly. For instance, the peaks value at 3384, 2921 and 1599 cm −1 moved to 3392, 2923 and 1597 cm −1 , respectively. It shows that the adsorption of Cu 2+ , Zn 2+ and Pb 2+ by lignite is due to physisorption under van der Waals forces 28 . The peak value of MML at 586 cm −1 shifted to 575 cm −1 , indicating the possibility of Fe-O combining with Cu 2+ and Zn 2+ through Fe-O-Cu and Fe-O-Zn. In addition, the peak shape of the hydroxyl group corresponding to 1111-1270 cm −1 also changed, which may be due to the reaction of Zn 2+ and OH − to produce Zn(OH) 2 precipitation. These phenomena are consistent with XRD results. The stretching vibration of Zn 2+ by hydroxyl group of carboxylic acid group in lignite generated metal compound C 4  Compare the removal of Cu 2+ , Zn 2+ and Pb 2+ by different adsorbents. Although the Langmuir model assumes that monolayer adsorption occurs on the uniform adsorbent surface and there is no interaction between adsorbates, many researchers have used Langmuir constant q m to evaluate the adsorption capacity of adsorbents. To compare the maximum adsorption capacity of Cu 2+ , Zn 2+ and Pb 2+ during the adsorption process, various adsorbents for Cu 2+ , Zn 2+ and Pb 2+ removal were prepared, as shown in Table 4  www.nature.com/scientificreports/ indicates that MML is a potential adsorbent for Cu 2+ , Zn 2+ and Pb 2+ adsorption from AMD. On the other hand, the optimum amount (4 g/L) of MML is lower than that of other sorbents. It shows that MML can be used as a low-cost adsorbent to remove Cu 2+ , Zn 2+ and Pb 2+ from AMD (Table 4).

Conclusion
1. The best adsorption conditions for Cu 2+ , Zn 2+ and Pb 2+ by lignite and MML were pH = 4, adsorbent dosage 4 g/L and temperature 25℃. Under the same metal concentration conditions, the adsorption capacity and removal rates of Cu 2+ , Zn 2+ and Pb 2+ by MML were higher than that of lignite. When the concentration of heavy metal ions in AMD solution was higher, MML had greater adsorption potential than lignite. 2. The isothermal adsorption of Cu 2+ , Zn 2+ and Pb 2+ by lignite and MML was consistent with the Langmuir model, indicating that the adsorption was consistent with the monolayer layer adsorption process. The adsorption processes of Cu 2+ , Zn 2+ and Pb 2+ by lignite obeyed the Quasi first-order kinetic model, indicating that the adsorption process was dominated by physisorption, and the fitted equations were: y = 9.2602*(1 − e −0.00607x ), y = 10.2839*(1 − e −0.00468x ), and y = 11.8456*(1 − e −0.01265x ), respectively. The adsorption process of MML obeyed the Quasi second-order kinetic model, which indicates that the adsorption process was dominated by chemisorption and the adsorption rate was affected by the coordination between the surface active site of adsorbent and the heavy metal ions, and the fitted equations are: y = 0.09599x + 8.81861, y = 0.09333x + 10.01582, y = 0.05103x + 5.02836. The fitting results of Intra-particle diffusion model showed that the adsorption processes of lignite and MML was jointly controlled by multiple adsorption stages. The Elovich model and desorption experiments confirmed that the adsorption process of Cu 2+ , Zn 2+ and Pb 2+ by MML was mainly chemisorption. The adsorption thermodynamics showed that the adsorption of Cu 2+ and Zn 2+ by lignite and MML was spontaneous and endothermic, while the adsorption of Pb 2+ was exothermic. 3. The dynamic experimental results showed that the removal effect of heavy metal ions by lignite was significantly better than that of lignite. The average removal rates of Cu 2+ , Zn 2+ and Pb 2+ by lignite were 78.00, 76.97 and 78.65%, respectively, and the average removal rates of the studied heavy metal ions by MML were 82.83, 81.57 and 83.50%, respectively. In addition, the adsorption capacity of the three heavy metal ions in AMD by lignite and MML is as follows: Pb 2+ > Cu 2+ > Zn 2+ . 4. From SEM, XRD and FTIR tests, it showed that Fe 3 O 4 was successfully loaded onto the lignite surface during the magnetic modification process. SEM test showed that the surface morphology of lignite was rougher after magnetic modification, and more sediment was generated on MML surface after the reaction. XRD results showed that Lignite and MML removed Cu 2+ , Zn 2+ and Pb 2+ from AMD in different forms. FTIR results showed that the adsorption process of Cu 2+ , Zn 2+ and Pb 2+ by MML was related to the O-H stretching vibration of carboxylic acid ions and Fe-O stretching vibration of Fe 3 O 4 particles. 5. The raw lignite has the characteristics of low cost and wide source. Using magnetic modification method to modify lignite can not only improve the adsorption capacity of the lignite, but also solve the problem that the lignite is difficult to separate from the solution. In this paper, a new modification method of lignite was proposed, which verified the feasibility of MML in the treatment of AMD, and provided a basis for the adsorption and use of MML. Adsorbent preparation. Lignite: Lignite was pulverized with a high-speed mill and screened out with a diameter of 250 mesh (58 μm). The lignite was immersed in deionized water for 2-3 times to remove impurities, and then dried at 60 °C for 24 h as the raw material. MML: The chemical co-precipitation method is used to magnetically modify the lignite, that is, to load Fe 3 O 4 magnetic particles on the surface of the lignite 57 . The formation reaction formula of Fe 3 O 4 is as follows.
The molar ratio of Fe 3+ to Fe 2+ substance was set to 2:1 (i.e. 1.31 g FeSO 4 ·7H 2 O and 1.88 g Fe 2 (SO 4 ) 3 ). 200 mL and 0.7 mol/L iron ion solution was prepared and placed in a thermostatic water bath at 60 °C. Weigh 10 g lignite and add it to iron solution, stir it for 1 h under the action of an electric stirrer with speed regulation at 350r/ min, then add concentrated ammonia water with mass fraction of 25% drop by drop to pH value of 9, continue to stir for 1 h, and stand for 2 h for aging. The resulting precipitate was repeatedly cleaned with deionized water to make the supernatant neutral and then separated by magnets to obtain the magnetic material. The magnetic material was dried in a vacuum drying oven for 12 h to obtain MML. The comparison of solid-liquid separation between lignite and MML in Fig. 9.
Heavy metal ions solutions preparation and analysis. CuSO  were used to prepare Cu 2+ , Zn 2+ and Pb 2+ standard solutions, respectively. Metal concentrations were measured thorough an atomic absorption spectrometer (AAS) with an air-acetylene flame (Hitachi-Z2000, Japan). The wavelengths used for the analysis of the Cu 2+ , Zn 2+ , and Pb 2+ were 324.8, 213.9, and 283.3 nm, respectively. The pH values of the solutions were adjusted by using 3% nitric acid or sodium hydroxide. Experimental methods. Adsorption condition optimization experiment. Prepared a Cu 2+ standard solution with a concentration of 30 mg/L. Added a certain amount of lignite and MML into 250 mL conical flasks containing 250 mL Cu 2+ standard solution. Placed the conical flasks in a constant tremors shaking at 150 r/ min, adsorb for 180 min, and then sample with a pipette gun. The samples were filtered through a 0.45 μm microporous membrane before analysis with AAS. Calculated the removal rates and adsorption amounts of heavy metal ions by Eqs. (12) and (13) 58 . All the experiments were repeated three times, and the average values were taken as the final measured values. The research was carried out by changing the pH value of the solution (2-4), adsorbent dose (2-6 g/L) and temperature (298.15, 308.15 and 318.15 K). The test conditions and process of the Zn 2+ standard solution with a concentration of 30 mg/L and the Pb 2+ standard solution with a concentration of 50 mg/L were exactly the same as the Cu 2+ adsorption test.  14) and (15) 59 . The test conditions and process of the Zn 2+ standard solution with a concentration of 30 mg/L and the Pb 2+ standard solution with a concentration of 50 mg/L were exactly the same as the Cu 2+ adsorption test.
All the experiments were repeated three times and the average values were taken as the final measured values.
where q e is the adsorption capacity at equilibrium (mg/g), Q d is desorption amount (mg/g), C is the concentration of heavy metal ions in ethanol solution at 180 min (mg/L), V is the volume of solution (L), M is the mass of adsorption material (g), and TD is desorption rate (%). If TD > 50%, the adsorption is physical adsorption. If TD < 50%, chemisorption.
Dynamic experiment. Two Plexiglas's tubes with an inner diameter of 40 mm and a height of 250 mm were filled with lignite and MML respectively. Glass beads with a height of 25 mm and a diameter of 3-5 mm were arranged at the top and botton of the dynamic column, and 100 mm high lignite was filled in the middle of #1 Plexiglas tube, and 100 mm high MML was filled in the middle of #2 Plexiglas tube. According to the results of the static beaker experiments, prepared the AMD with Cu 2+ and Zn 2+ concentration of 30 mg/L and Pb 2+ concentration of 50 mg/L, and adjusted the pH value to 4. The overall operation mode adopted "bottom in and top out" continuous operation, and the inlet water flow rate was adjusted to 0.556 mL/min by peristaltic pump and flowmeter. The experimental device was shown in Fig. 10. The two groups of dynamic columns were operated at room temperature for 22 days, and samples were taken every 12 h. After the samples were filtered by 0.45 μm