TEMPO-oxidized cellulose hydrogel for efficient adsorption of Cu 2+  and Pb 2+  modified by PEI


 In this study, hydrogel were prepared by dissolving and regenerating poplar-cellulose in NaOH/urea/water system. The TEMPO-oxidized cellulose hydrogels (TCH) were prepared using microwave-assisted accelerated TEMPO-oxidation system. Polyethyleneimine (PEI) was grafted onto TCH with glutaraldehyde as a cross-linking agent and the products named as TCP. The hydrogels were characterized by SEM, FTIR, XPS and elemental analyzer. The maximum adsorption capacities of Cu2+ and Pb2+ by TCP were 109.89 mg/g and 279.32 mg/g, respectively. TCP was a single molecule adsorption process with better fitting of Langmuir model. Adsorption kinetics showed that the Pb2+ adsorption rate of TCP was higher than that for Cu2+. The Cu2+ affinity of TCP was higher than the Pb2+. The adsorption capacity of TCP for Cu2+ and Pb2+ was 58.26 mg/g and 91.96 mg/g, respectively, after five cycles. This study provided a promising option of preparing an efficient and recyclable adsorbent in treating wastewater containing heavy metal, such as Cu2+ and Pb2+.


Introduction
Nowadays, heavy metals are one of the most important pollutants due to industrial and human activities (Shahnaz et al. 2016). Heavy metal contamination in industrial wastewater is a huge environmental problem because heavy metals cannot be metabolized and will constitute a great of threat to human health (Fu et al. 2011). Thus, appropriate techniques should be used to remove heavy metals from industrial wastewater before they are released into the environment. Several methods have been implemented to remove heavy metal, including membrane ltration, chemical precipitation, ion exchange, biological treatment, and adsorption. Among these methods, the adsorption method has the advantages of easy operation, high e ciency, good recyclability, and has been widely used (Bilal et al. 2013).
A variety of inorganic, organic, and organic/inorganic hybrid adsorbents have been developed to remove heavy metals from water treatment (Cui et al. 2015). However, many conventional adsorbents (e.g. activated carbon and clays) display inconvenient recyclability or expensive regeneration cost, and increase the expense for wastewater treatment. In recent years, biomass based materials have attracted extensive research and attention due to their advantages such as low cost, high absorption capacity, reproducibility and outstanding pollution control effect (Suhas et al. 2016;Kumar et al. 2017). As the main component of biomass based materials, cellulose has the advantages of being renewable (Nechyporchuk et al. 2016), biodegradaable and low cost (Denisov et al. 2017). Because it has a considerable number of hydroxyl groups, the oxygen in the hydroxyl group has unbonded electrons, which can cooperate with the vacant orbital of metal ions to form coordination bond adsorption, natural cellulose exhibits excellent adsorption properties for heavy metal (Wang et al. 2016).
However, the adsorption capacity of cellulose may not be as high as expected without any modi cation.
In order to prepare cellulose based heavy metal ion adsorbents, the hydroxyl in cellulose must be modi ed by esteri cation, etheri cation and other chemical modi cations (Gurgel et al. 2008). Although many previous studies have focused on the modi cation of cellulose adsorbents, most of studies introduced functional groups directly on bulk cellulose (Aoki et al. 1999;Zhong et al. 2014;Ge et al. 2018;Godiya et al. 2019). In addition, only functional groups in the outer layer of cellulose bers can be used, while a large number of hydroxyl groups are restricted by hydrogen bonds between cellulose micro bers.
The organic solvent systems were required by these modi cation methods, which will harm the environment in some extent.
Previous studies reported that the surface of cellulose hydrogel after being oxidized by TEMPO (2,2,6,6tetramethylpyperidine-1-oxy radical) tended to open holes Saito 2010;Rodionova et al. 2012). The further modi cation of cellulose hydrogel is necessary to make it more competitive in practical application. Grafting branched polyethyleneimine (PEI) using glutaraldehyde crosslinking method has been reported to be an easy and cheap method of introducing amino groups on various materials with hydroxyl, aldehyde or carboxyl groups (Ma et al. 2014;Sun et al. 2011). By using glutaraldehyde crosslinking method, the adsorption capacity of the adsorbent can be greatly improved by grafting PEI. Therefore, in this paper, a highly functional heavy metal ion adsorbent was prepared by modifying oxidized cellulose hydrogel with PEI. The adsorbents were characterized by SEM, FTIR, XPS and elemental analyzer. Moreover, the effects of pH, adsorption time, initial ion concentration, temperature, coexisting ions concentration and competitive adsorption on cellulose hydrogel adsorption of Cu 2+ and Pb 2+ and performance were studied.
Preparation of the adsorbents Poplar cellulose has poor solubility in NaOH/urea/water system with a high degree of polymerization. In this study, HCl/ethanol solution was used to pretreat the cellulose, and a lower degree of polymerization cellulose was obtained. At rst, 50 g cellulose was hydrolyzed with 1 L hydrochloric acid/ethanol (HCl: ethanol = 1:25, V/V) at 70 ℃ for 2 h. Then, 4 g pretreated cellulose was hydrolyzed in 96 g NaOH/urea/water system solvent and stirred for 5 min. After cooled to -20 ℃ for 2 h, the solvent was stirred vigorously for 10 min, and resulting in a transparent solution. The solution was subjected to centrifugation for 20 min to remove air bubbles. In the following, 5 mL cellulose solution was absorbed by a disposable needle and dropped into the coagulation bath (trichloromethane: ethyl acetate: glacial acetic acid = 3:3:1, V/V/V) to form hydrogel balls of the same size. After curing for 10 min, the hydrogel balls were placed in owing water and washed to neutral. After that, the hydrogel balls were soaked in pure water for 3 days, and changed water every 6 h. The cellulose hydrogel balls named as CH. 0.5 g CH were taken into 50 mL 0.05 M pH = 6.8 phosphate buffer solution, which also contained 0.048 g TEMPO, 0.8475g NaClO 2 and 0.75 mL NaClO. The mixture was subjected in the McR-3 microwave chemical reactor at 60 ℃ for 6 h and the products was named as TCH. 1 g TEMPO-oxidized cellulose hydrogel (TCH) was added into 100 mL water and then PEI with different qualities (1, 2, 3 g) was added into TCH solution and stirred at 20 ℃ for 1 h. After that, 20 mL 1% glutaraldehyde solution was added dropwisely and reaction at 60 ℃ for 2 h. After three times of displacement in ethanol and tertiary butyl alcohol, PEI-modi ed poplar cellulose hydrogel were obtained after freeze-drying, and named as TCP.

Characterization
The morphologies of the hydrogels, were observed with the scanning electron microscopy (SEM, Hitachi S-4800, Japan) at an accelerating voltage of 3.0 kV. The functional groups of the hydrogels were characterized using a fourier transform infrared (FTIR) spectrometer (Bruker Tensor , Germany) within the wavenumber range of 500 cm − 1 -4000 cm − 1 . The mass ratios of C, H and N of TCH and TCP were measured using elemental analyzer (Vario EL cube, Elementar Co., Germany). The C 1s , O 1s and N 1s spectra of adsorbents were characterized using a XPS instrument (ESCALAB 250Xi, Thermo-VG Scienti c Co., US). Thermal properties of each absorbents were measured by thermogravimetry (TG 209, Germany).
Heating was conducted under nitrogen with heating rate of 10 ℃/min from 35 ℃ to 700 ℃.

Adsorption experiments
Determination of carboxyl content Carboxyl content of TCP was determined by conductivity titration. The TCP was ground into powder, and 0.3g TCP powder was added into 55 mL pure water with 5 mL 0.01mol/L NaCl. The mixture was stirred well and the pH of mixture was adjusted to 2.5-3.0 by HCL (0.1 mol/L). Finally, the mixture was titrated with 0.1 mol/L NaOH until the pH was 11, record the pH value during the titration. The carboxyl content can be calculated from the following Eqs.: Where C is the concentration of NaOH (mol/L); V 1 is the volume of initial NaOH (L); V 2 is the volume of NaOH at the in ection point of the second derivative (L); m is the quality of sample.

Effect of adsorption time and pH on Cu 2+ and Pb 2+ adsorption
The adsorption kinetics of the TCP was evaluated by dosing 50 mg TCP into 50 mL 100 mg/L Cu 2+ and Pb 2+ solution. The pH of the Cu 2+ solution and Pb 2+ solution were controlled at 5.0. The mixture of the TCP and two ion solutions were shaken in a thermostatic shaker at 180 rpm and 30 ℃. At a predetermined time intervals, the solution was collected and centrifuged. The Cu 2+ and Pb 2+ concentrations in the samples were examined using an ICP-AES (IRIS Intrepid IIXSP, Thermo Electron Corporation, USA). The in uence of adsorption times (varying from 1 to 200 min) and pH (2-5) on the adsorption was examined. Because the high pH would cause the precipitation of Cu 2+ and Pb 2+ , the pH in this study was only set in the range 2-5.
In order to better perception of adsorption process, the kinetic predictions in the adsorption mechanism, as well as the controlling mechanism of adsorption process are remarkable and fundamental for the designation of adsorption equilibrium time and adsorption rate. Pseudo-rst-order kinetics model, pseudo-second-order kinetics model and intraparticle diffusion ) were used to examine the adsorption kinetics, which can be expressed by Eqs. as following: Pseudo-rst-order kinetics model: pseudo-second-order kinetics model: Where Q e is the equilibrium adsorption capacity (mg/g); Q t is adsorption capacity (mg/g) at time t; t is adsorption time (min); K 1 is the pseudo rst-order reaction rate constant (min − 1 ); K 2 is the pseudosecond-order reaction rate constant (g·mg − 1 ·min − 1 ); K id is the intraparticle diffusion rate constant (mg·g − 1 ·min − 0.5 ). For pseudo-second-order kinetics model, when t → 0, the initial adsorption rate h could be de ned as h = k 2 q e 2 Effect of initial Cu 2+ and Pb 2+ concentrations on adsorption The adsorption isotherms of TCP was tested by dosing 30 mg TCP into asks containing 50 mL Cu 2+ and Pb 2+ solution of different concentrations (50,100,150,200,250,300,350 and 400 mg/L). The mixture of adsorbent and two ion solutions were shaken in a thermostatic shaker at 180 rpm and 30 ℃ for 24 h to reach adsorption equilibrium.
Langmuir and Freundlich models were widely used to describe the adsorption process. The former wasvalid for monolayer sorption on the adsorbent surface with nite number of similar active sites, while the later was an empirical model which was valid for the multilayer adsorption. In this study, the two equilibrium models were used, which can be expressed by Eqs. as following (Charpentier et al. 2016): Langmuir model: C e /Q e = C e /Q max + 1/(K L Q max ) Freundlich model: log Q e = log K F + 1/n (log C e ) Where C e is the concentration of Cu 2+ at equilibrium (mg/L); Q e is the adsorption of Cu 2+ and Pb 2+ at equilibrium (mg/g); Q max is the maximum adsorption capacity (mg/g) at equilibrium; K L is Langmuir adsorption constant (L/mg); K F is Freundlich adsorption constant (mg/g); 1/n is the adsorption strength.The best-t equilibrium model was determined based on the non-linear regression correlation coe cient (R 2 ).
Effect of temperature on Cu 2+ and Pb 2+ adsorption The further study on the adsorption mechanism of TCP was tested by dosing 30 mg TCP into asks Where R, T, and K c represent the gas constant (8.314 J mol − 1 K − 1 ), Kelvin temperature and adsorption equilibrium constant, respectively; Q e is the adsorption amount at equilibrium (mg/g); C e is the concentration of Cu 2 + and Pb 2+ at adsorption equilibrium (mg/L).
Effect of coexisting ions on Cu 2+ and Pb 2+ adsorption 30 mg TCP was dosed into asks containing 50 mL of 100 mg/L Cu 2+ and Pb 2+ solution with different ionic strengths. The pH of two ion solutions were controlled at 5.0. The mixture of the TCP and ion solutions were shaken in a thermostatic shaker at 180 rpm and 30 ℃ for 24 h to reach adsorption equilibrium. The concentration of Cu 2+ and Pb 2+ in the supernatant were determined after centrifuging to obtain the in uence of different ionic strength on TCP adsorption. NaCl, KCl and CaCl 2 were selected as coexisting ions with 1.0, 2.0, 3.0, 4.0, 5.0 and 6.0 mmol.
The competitive adsorption of Cu 2+ and Pb 2+ solution. The pH of ion solution was controlled at 5.0. The mixture of the TCP and ion solution was shaken in a thermostatic shaker at 180 rpm and 30 ℃ for 24h to reach adsorption equilibrium. The concentration of Cu 2+ and Pb 2+ in the supernatant was determined after centrifuging to obtain the in uence of competitive adsorption on TCP adsorption.

Desorption and regeneration experiments
In order to evaluate the stability of the adsorbent, the adsorption-desorption cyclic experiments were carried out. 30 mg freeze-dried TCP was dosed to 50 mL 100 mg/L Cu 2+ and Pb 2+ solution in a 250 mL ask, and were shaken in a thermostatic shaker at 180 rpm for 24 h. The pH of two ion solutions were controlled at 5.0. After the adsorption procedure completed, the TCP with Cu 2+ or Pb 2+ adsorbed was obtained by ltration and then regenerated with 100 mmol/L Na 2 EDTA for 24 h to free the occupied adsorption sites. Then the regenerated adsorbents were rinsed several times with deionized water and dried at ambient temperature. The air-dried adsorbent was added in 100 mg/L Cu 2+ and Pb 2+ solution again for the next adsorption experiment. The adsorption-desorption experiment was carried out in 5 cycles in total and the adsorption capacity of TCP in each cycle was measured.

Results And Discussion
Characterization of TCP absorbent The surface and cross-section morphologies of CH, TCH and TCP were shown in Fig. 1. It can be seen that the surface of the CH was at. After TEMPO oxidization, the typical network structure was occurred. After grafted with PEI, a PEI layer covered uniformly around the TCP which forming quantities of pores with small size between the TCP branches. The speci c surface area of TCP was increased by the grafting of PEI and thus the accessibility of metal ions to the TCP would be promoted. The FTIR spectra of CH (a), TCH (b) and TCP (c) were presented in Fig. 2. The peaks at 3409 cm − 1 and 2900 cm − 1 were attributed to the tensile vibration of O-H and C-H, respectively (Zhang et al. 2016). The band at 1112 cm − 1 was assigned to the stretching vibration of C-O, and the 1638 cm − 1 was the characteristic peak of cellulose water absorption (Zhang et al. 2016). New absorption peaks at 1614 cm − 1 and 1415 cm − 1 were appeared after TEMPO oxidization, which belonged to -COO- . This results showed that carboxyl were successfully introduced into CH after TEMPO oxidization. Due to the superposition of O-H and N-H absorption peaks, the absorption peak of TCH (Fig. 2b) at 3422 cm − 1 became broad in Fig. 2c.
The stretching vibrations of N-H at 1613 cm − 1 and C-N stretching vibrations at 1456 cm − 1 in TCP spectrum also proved the successfully introduction of PEI to oxidized cellulose hydrogel (Zhang et al. 2016;Guo et al. 2017).
The contents of C, N and H in different cellulose-based adsorbents were listed in Table 1. The N content of oxidized cellulose was increased signi cantly after the reaction with PEI, indicating that PEI was successfully introduced into TCH. Due to the high content of N element in PEI, the N content of TCP was increased signi cantly from 0.016-11.574% (TCH: PEI = 1:1) compared to that of oxidized cellulose. The N content of TCP tended to be stable with the addition of PEI, indicating that the amount of PEI was overused during the reaction. The contents of each element in TCP was presented in Fig. 3, TCP was mainly composed of C (67.04 %), N (13.69 %), O (19.27 %). The peak tting of C 1s , O 1s and N 1s showed that C-C (284.7eV), C-N (285.4 eV), C-O (285.9 eV), and C = O (287.4 eV) bonds were existed in TCP. The three tting peaks of N 1s in TCP were primary amine (398.51eV), secondary amine (399.18eV), and tertiary amine (400.7eV), respectively (Zhao et al. 2017). The results of elemental analysis data, the XPS and FTIR spectra con rmed that PEI was grafted successfully on TCH. The thermal stabilities of CH, TCH and TCP were determined by thermogravimetric analysis in nitrogen from 35 ℃ to 700 ℃ (Fig. 4). Due to the carbonization and pyrolysis of cellulose, the main pyrolysis interval of poplar cellulose appeared at 338 ℃-375 ℃, and the mass loss at this stage was 77.46%. Compared with poplar cellulose, the initial pyrolysis temperature of oxidized cellulose was at 222.8 ℃.
This phenomenon was mainly due to the decrease of crystallinity of oxidized cellulose (Yang 2011), and the initial decomposition temperature could be advanced by the presence of carboxyl (Zhao et al. 2017).
The thermal decomposition peaks of 173-195 ℃ and 343-385 ℃ appeared in TCP (Zhao et al. 2017;Li et al. 2018) was due to the aminolysis of PEI and the broke of PEI chain. In addition, the nal residual masses of oxidized cellulose and TCP were 28.36% and 15.78%, indicating that the thermal stability of TCP was reduced.
Effect of oxidation time on the carboxyl content and adsorption amount of TCP were shown in Fig. 5. The oxidation process of CH can be accelerated by microwave. The carboxyl content of CH increased rapidly with the increase of oxidation time, which was consistent with Lin's ) study. The cuticle of CH was peeled off, and the carboxyl content was increased by the process of oxidation, which provided a large number of adsorption sites for Cu 2+ . The adsorption capacity of TCP on Cu 2+ was also increased with the increase of oxidation time, and the carboxyl content and adsorption capacity reached the maximum at oxidation time of 6 h. Therefore, the hydrogels with oxidation time of 6 h were selected for further experiments. The adsorption amount of TCH was increased signi cantly after grafted with PEI, while the adsorption amount was limited increased with the increase of PEI. Finally, TCH: PEI = 1:1 was selected as the research object in this study.

Adsorption performance
Effect of pH on Cu 2+ and Pb 2+ adsorption The effect of pH value on the adsorption capacity of heavy metals is one of important parameters for sorption process. The pH dependence of metal adsorption is closely related to the surface functional groups of the adsorbent (Dehghani et al. 2016 Besides, electrostatic forces between adsorbent and heavy metal ions were also increased, resulting in the sharply increasing of adsorption capacity of TCP. Therefore, the pH of the solution was adjusted to 5 in the subsequent experiments.

Effect of adsorption time on Cu 2+ and Pb 2+ adsorption
The effect of different adsorption time on Cu 2+ and Pb 2+ adsorption was shown in Fig. 7. As shown in and Pb 2+ were decreased, and the active adsorption sites were occupied by Cu 2+ and Pb 2+ . The concentration difference between the metal ions and TCP decreased, resulting the decrease of adsorption rate until the adsorption equilibrium (Gao et al. 2016).
Pseudo-rst-order kinetics model, pseudo-second-order kinetics model, and intraparticle diffusion were used to investigate adsorption kinetic mechanisms . The simulated curves of Cu 2+ and Pb 2+ adsorption by TCP were shown in Fig. 7, and the tting parameters were shown in Table 2. It can be seen that the adsorption process of Cu 2+ and Pb 2+ tted well by pseudo-second-order kinetics, and the adsorption amounts obtained by tting were close to the actual measured results. Pseudo-second-order kinetics results showed that TCP adsorption was controlled by chemisorption. As shown in Fig. 7b, the adsorption processes of Cu 2+ and Pb 2+ by TCP were divided into two stages. Firstly, the amino and carboxyl groups on TCP rapidly chelating to Cu 2+ and Pb 2+ , which were controlled by chemical adsorption. Then, Cu 2+ and Pb 2+ were adsorbed slowly into the TCP and controlled by intraparticle diffusion. Although TCP had similar adsorption processes for Cu 2+ and Pb 2+ , the Pb 2+ adsorption rate of TCP was higher than that of Cu 2+ . Initial concentrations of the heavy metal ions play a major role in the adsorption capacity. Generally, the in uence of initial metal ions concentration rely on the relative between the existing sites on an adsorbent surface and concentration of the heavy metals (Geng et al. 2017). The relationship between adsorbents and adsorbates can be revealed by the adsorption isotherm (Guo et al. 2017). The effects of different initial Cu 2+ and Pb 2+ concentrations on adsorption were shown in Fig. 8a. With the increase of initial Cu 2+ and Pb 2+ concentrations, the adsorption of Cu 2+ and Pb 2+ by TCP showed gradually increasing trend, and the adsorption capacity of Pb 2+ was higher than that of Cu 2+ . When the initial concentrations were greater than 300 mg/L, adsorption of Cu 2+ and Pb 2+ by TCP reached saturation, and the adsorption capacity was stable. Langmuir and Freundlich models were used to investigate the adsorption mechanism. The Langmuir model assumed that the adsorbent surface had a nite number of binding sites with the same energy, and each binding site absorbed a single ion (Charpentier et al. 2016). The Freundlich model assumed that different binding sites on the adsorbent surface was based on multimolecular layer adsorption . The adsorption isotherms of the Langmuir and Freundlich models were shown in Fig. 8b and Fig. 8c. Adsorption isotherm parameters for the adsorption Cu 2+ and Pb 2+ by TCP were shown in Table 3. As shown in Table 3, the adsorption processes of TCP could be well tted by the Langmuir model, and the tting coe cients were all greater than 0.999. The actual adsorption capacities of TCP were consistent with the tted maximum adsorption capacity. The isothermal adsorption conformed to the Langmuir model, indicating that the adsorptions of Cu 2+ and Pb 2+ by TCP were belonged to the single molecular layer adsorption. According to the Langmuir model, the maximum adsorption capacity of Cu 2+ and Pb 2+ by TCP was 109.89 mg/g and 279.33 mg/g, respectively, which were far higher than other cellulose-based adsorbents (Table 4). This results might due to that TCP had large number of active chelate functional groups (amino, carboxyl, hydroxyl) and pore structures, it could provided many adsorption sites for heavy metal ions. Temperature is another major parameter to in uence the adsorption property. The adsorption capacity was effected by temperature which could change the molecular interactions and solubility. The effect of different temperatures on Cu 2+ and Pb 2+ adsorption by TCP were shown in Fig. 9. With the increase of temperature, the adsorptions of Cu 2+ and Pb 2+ by TCP were both gradually increased, indicating that the adsorptions were endothermic process (Dehghani et al. 2016). The adsorption mechanism was analyzed by linear tting of thermodynamic parameters (ΔG o , ΔH o , and ΔS o ), and the results were shown in Table 5. Negative value of ΔG o indicated that the adsorption process was spontaneous, and temperature was advantageous to the adsorption process. Positive value of ΔH o and ΔS o indicated that the adsorption process was endothermic, and the randomness of the solid-liquid interface was increased during the adsorption process (Demiral et al. 2016).  (Guo et al. 2018).
However, it is precisely the existence of coexisting ions that can cause an inhibition for the adsorption process of heavy metal ions. The morphology of heavy metals and the adsorption sites of adsorbents could be affected by coexisting ions (Wang et al. 2019). The effect of coexisting ions on Cu 2+ adsorption were discussed in this study, and the results were shown in Fig. 10. TCP had a small adsorption capacity for Cu 2+ at low NaNO 3 concentration (below 1 M). While NaNO 3 concentration was 1 M, the adsorption of Cu 2+ by TCP decreased slightly from 103 mg/g to 99 mg/g. Because of the activity coe cient of Cu 2+ could be affected by high ion concentration, TCP adsorption capacity for Cu 2+ was declined at high ion concentration. The results showed that the coexisting ions had little effect on the adsorption of Cu 2+ by TCP.
The competitive adsorption of Cu 2+ and Pb 2+ In the natural environment, Cu 2+ and Pb 2+ often coexist in contaminated areas. The in uence of single system and coexistence system on TCP adsorption were shown in Fig. 11. The adsorption capacities of Cu 2+ and Pb 2+ by TCP in the single system were 65.23 mg/g and 78.81 mg/g, respectively. In the coexistence system (Cu 2+ and Pb 2+ ), the adsorption capacity of Cu 2+ was slightly decreased to 64.42 mg/g, while that of Pb 2+ signi cantly decreased to 52.03 mg/g. Because TCP had a higher a nity for Cu 2+ than Pb 2+ , Cu 2+ could compete with Pb 2+ for adsorption sites.

Recycling performance of TCP
Recycling stability is also a signi cant factor for an adsorbent in the practical application. In order to evaluate the regeneration ability of TCP, 100 mmol/L Na 2 EDTA was used to desorb Cu 2+ and Pb 2+ from the adsorbent. Figure 12 showed the performance of TCP for Cu 2+ and Pb 2+ adsorption during 5 adsorption-desorption cycles. The results showed that the adsorption capacities of Cu 2+ and Pb 2+ by TCP were decreased with the increase of cycle times. In the second cycle, the Cu 2+ and Pb 2+ adsorption quantity was 77.28 mg/g and 116.94 mg/g, respectively. The decrease of adsorption capacity in the second cycle might be due to that some active sites of the absorbent combined with metal ions in an irreversible way, reducing the density of ions binding sites in the second cycle. Obviously, TCP still had a high adsorption capacity for Cu 2+ and Pb 2+ after 5 cycles with an adsorption capacity of 58.26 mg/g for Cu 2+ and 91.96 mg/g for Pb 2+ . Although the adsorption capacity of the regenerated TCP was far lower than that of the original TCP, it still had a good adsorption effect. The stability in the adsorptiondesorption experiment implied that the good regeneration ability of TCP, making it a practical adsorbent in the real wastewater treatment.

Adsorption mechanism
The functional groups change for the adsorption of Cu 2+ and Pb 2+ by TCP were shown in Fig. 13. The absorption peaks were belonged to amino, and the peak of hydroxyl at 3422 cm − 1 became weaker after adsorption. This peak was shifted to 3432 cm − 1 and 3440 cm − 1 after adsorption of Cu 2+ and Pb 2+ , respectively, indicating that the amino and hydroxyl in TCP were involved in Cu 2+ and Pb 2+ adsorption. In addition, the absorption peaks belonged to N-H and C-N at 1613 cm − 1 and 1456 cm − 1 were also changed after adsorption. The results of XPS spectrum further con rmed the adsorption mechanism (Fig. 14).
After adsorption, the absorption peaks of Cu 2p at 933.86 eV and Pb 4f at 138.16 eV were appeared , indicating the successful adsorption of Cu 2+ and Pb 2+ by the TCP. Meanwhile, the absorption peaks belonged to -NH 2 were shifted from 398.51eV to 399.22 eV and 398.82 eV, respectively, and the peaks belonged to -NH-were shifted from 399.15 eV to 399.96 eV and 399.85 eV, respectively.
Furthermore, the peaks belonged to -N-were shifted from 400.7 eV to 400.98 eV and 401.33 eV, respectively. All of peaks were shifted to higher binding energies, indicating that primary, secondary and tertiary amines were played an important role in the adsorption. Moreover, a new absorption peak appeared at 406.49 eV in N 1s after the adsorption of Pb 2+ , probably due to the formation of compound between N and Pb 2+ on the primary and secondary amino groups. This result was consistent with the study of Zhao (Zhao et al. 2017). The above results con rmed that the hydroxyl, amino and carboxyl groups on the surface of TCP played an important role in the adsorption.

Conclusion
In this study, a poplar cellulose based adsorbent for Cu 2+ and Pb 2+ uptake was prepared. Further characterization con rmed that large amounts of carboxyl groups and amino groups had been successfully introduced to the adsorbent. The maximum adsorption capacity of Cu 2+ and Pb 2+ by TCP was 109.89 mg/g and 279.32 mg/g, respectively, which were signi cantly higher than cellulose-based adsorbents in other studies. Langmuir model tted the adsorption process well and showed that TCP was a single molecule adsorption process. Adsorption kinetics showed that the Pb 2+ adsorption rate of TCP was higher than that for Cu 2+ . The coexisting ion contents had little in uence on the adsorption of Cu 2+ by TCP. Because TCP had a higher a nity for Cu 2+ than Pb 2+ , Cu 2+ could compete with Pb 2+ for adsorption sites. After 5 cycles, the adsorption capacity of TCP for Cu 2+ and Pb 2+ was 58.26 mg/g and 91.96 mg/g, respectively. Considering the ultra-porosity structure, green synthetic way, recyclable and supreme adsorption, the TCP as a native adsorbent can be compete as appropriate candidate for the treatment of e uent polluted with heavy metal ions. FTIR spectra of (a) CH, (b) TCH, and (c) TCP.

Figure 3
The wide XPS spectra and high-resolution XPS spectra of C1s, O1s , and N1s in TCP.

Figure 5
Effect of oxidation time on the carboxyl content and adsorption amount of TCP (a), effect of PEI content on the adsorption amount (b).

Figure 6
The adsorptions of Cu2+ and Pb2+ by TCP at various pH values.

Figure 7
The adsorption kinetics of Cu2+ and Pb2+ adsorption by TCP (a: linear plot of pseudo-rst-order and pseudo-second-order; b: Intraparticle diffusion model).

Figure 8
The effects of different initial Cu2+ and Pb2+ concentrations on adsorption (a) and the adsorption isotherms of the Langmuir (b) and Freundlich (c) models.

Figure 9
The effect of different temperatures on Cu2+ and Pb2+ adsorption by TCP. Figure 10 The effect of the ionic strength (NaNO3) on the adsorption Cu2+.

Figure 11
Adsorption of Cu2+ and Pb2+ from single and binary mixtures.

Figure 12
Recycling stability of TCP for the adsorption of Cu2+ and Pb2+.

Figure 13
FTIR spectra of TCP before and after adsorption of Cu2+ and Pb2+ (a: before adsorption, b: after adsorption of Cu2+, c: after adsorption of Pb2+).