Facile synthesis of high performance porous magnetic chitosan - polyethylenimine polymer composite for Congo red removal

G Model

ARTICLE IN PRESS

BIOMAC-8320; No. of Pages 9

International Journal of Biological Macromolecules xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

International Journal of Biological Macromolecules journal homepage: www.elsevier.com/locate/ijbiomac

Facile synthesis of high performance porous magnetic chitosan polyethylenimine polymer composite for Congo red removal Lijun You ∗ , Ci Huang, Feifei Lu, Ao Wang, Xiaocui Liu, Qiqing Zhang ∗ Institute of Biomedical and Pharmaceutical Technology, College of Chemistry, Fuzhou University, Fuzhou, 350001, China

a r t i c l e

i n f o

Article history: Received 23 June 2017 Received in revised form 9 September 2017 Accepted 5 October 2017 Available online xxx Keywords: Polymer composites Chitosan Polyethylenimine Adsorption Dye removal

a b s t r a c t A new porous magnetic chitosan-polyethylenimine (Fe3 O4 /CS-PEI) polymer composite was synthesized by crosslinking chitosan (CS) with polyethylenimine (PEI) in the present of FeCl3 ·6H2 O and FeCl2 ·4H2 O in alkaline condition and applied to remove congo red (CoR) from aqueous solutions. The Fe3 O4 /CS-PEI composite was characterized by SEM, XRD, TGA and FT-IR analysis. The polymer composite owned high positive charge, large surface area, multi-level pore distribution and magnetic responsiveness. The porous magnetic Fe3 O4 /CS-PEI composite showed ultrahigh capacity (1876 mg/g) for CoR removal. It removed over 99.3% of CoR (100 mg/L) when the dosage was over 1.4 g/L. A higher temperature was benefit to CoR removal. The Fe3 O4 /CS-PEI composite was effective for CoR removal in a wide pH range (3–13). Kinetics studies suggested that the adsorption mechanism of CoR followed the pseudo-second model and it was also affected by the boundary layer diffusion. The adsorption process followed the RedlichPeterson isotherm equation. Thermodynamic studies also demonstrated that this adsorption process was spontaneous, favorable and endothermic. The activation energy (Ea ) of the adsorption process was 34.08 kJ/mol, indicating that chemisorption existed in the process. The results demonstrated that the porous magnetic Fe3 O4 /CS-PEI polymer composite is a promising adsorbent for the efficient removal of dye pollutants from aqueous solution. © 2017 Elsevier B.V. All rights reserved.

1. Introduction Dye is one of the most common pollutants in wastewater due to their strong toxicity, non-biodegradability and accumulation in plants, animals and human beings [1]. Because of the increasing diversity of industrial products, the component of dye wastewater turns increasing complicated and the treatment of which becomes an extremely difficult task. Congo red (CoR), one of the watersoluble anionic dyes, has been widely used in printing, textile, leather, cosmetic industries and biomedical laboratories [2,3]. It can severely affect the aquatic life and the food web even in a low concentration by weakening the penetration of light in water and inhibiting the photosynthesis capacity of aquatic organisms. It also can cause health problems such as difficulties in breathing, diarrhoea, vomiting and nausea to humans and animals. [4]. Therefore, it is significant and urgent to remove CoR from wastewater. A variety of methods including membrane separation [2], adsorption [5], flocculation [6], photocatalytic chemical decomposition [7], electrolysis [8], and biological treatments [9] have been

∗ Corresponding authors. E-mail addresses: [email protected] (L. You), [email protected] (Q. Zhang).

developed to remove CoR in wastewater. Among these approaches, adsorption process is regarded as an efficient alternative, due to its simplicity of design, wide adaptability, convenience and ease of operation, especially when the adsorbent is inexpensive and readily available [10,11]. Many adsorbents are applied to adsorb dyes in wastewater, including metallic oxide, activated carbon, natural materials, algae, bentonite, nanocomposite, etc. [12]. Nevertheless, these adsorbents show limitations such as low adsorption capacity, hazardous by-products, separation inconvenience, high cost or intensive energy requirements [12,13]. Chitosan (CS) is a biological polysaccharide and widespread in nature. It has desirable properties like hydrophilicity, low cost biocompatible and biodegradable [14,15]. It contains amino and hydroxyl groups that can serve as adsorptive sites for dyes. However, the low mechanical strength, poor water solubility, easily hydrolyzed under acidic conditions and other drawbacks limit its application in dyestuff wastewater treatment [16–18]. In order to overcome these limitations, the physical and chemical modifications need to be carried out on chitosan. Chemical cross-linking is an effective method to improve its mechanical strength and its chemical stability in acidic media. However, these improvements often lead to the loss of amino or hydroxyl groups and result in the decreased adsorption capacity [16]. Polyethylenimine (PEI) is an

https://doi.org/10.1016/j.ijbiomac.2017.10.025 0141-8130/© 2017 Elsevier B.V. All rights reserved.

Please cite this article in press as: L. You, et al., Int. J. Biol. Macromol. (2017), https://doi.org/10.1016/j.ijbiomac.2017.10.025

G Model BIOMAC-8320; No. of Pages 9

ARTICLE IN PRESS L. You et al. / International Journal of Biological Macromolecules xxx (2017) xxx–xxx

2

amorphous high performance well-known biocompatible macromolecule with abundant amino groups on its surface and widely used in drug carrier, genetic vector, flocculant etc. [19,20]. Thus, cross-linking chitosan with PEI can not only improve its mechanical strength and its chemical stability, but also increase amino groups and result in the increase of adsorption capacity. Porous materials are appealing because of their unique pore architecture, high surface area, and specific physical and chemical properties, which contribute to their excellent performance as adsorbents [13]. Compared to the long and tedious centrifugation separation process, the magnetic separation technology can be easily manipulated by an external magnetic, and can treat a large amount of wastewater in a short time. Thus, combining functionalities of magnetism and porous structures may show facile separation properties and highly efficient adsorption [21,22]. In light of these, we assumed the design of a porous magnetic chitosan-polyethylenimine crosslinking product would achieve the purpose of developing an adsorbent with superior adsorption capability and facile separation properties. Hence, in the present work, the porous magnetic chitosanpolyethylenimine (Fe3 O4 /CS-PEI) polymer composite was prepared using a facile one-pot synthesis approach through crosslinking chitosan (CS) with polyethylenimine (PEI) in the present of epichlorohydrin as crosslinker, and FeCl3 ·6H2 O and FeCl2 ·4H2 O as the magnetic source. The as synthesised adsorbent not only possessed good mechanical strength, but also exhibited a super high adsorption performance for CoR removal. The designed structure provided Fe3 O4 /CS-PEI distinctive characteristics including highly positive charged, high surface area, multi-level pore distribution and magnetic responsiveness. The BET surface area of the Fe3 O4 /CSPEI is 109.2 m2 /g with average pore width of 15.08 nm and total pore volume of 0.24 cm3 /g. The Fe3 O4 /CS-PEI showed enhanced high capacity (1876 mg/g at 40 ◦ C) for CoR removal in aqueous solutions. The effects of dosage of Fe3 O4 /CS-PEI, initial pH and initial CoR concentration on the adsorption were studied. The thermodynamics and kinetics of the adsorption process were investigated in detail. The results showed that Fe3 O4 /CS-PEI is a high performance adsorbent with facile magnetic separation properties. 2. Experimental 2.1. Materials Chitosan (CS) with a molecular weight of 5.0 × 105 and a deacetylation degree of 95% was purchased from Yuhuan Ocean Biochemical Ltd, China. Polyetherimide (PEI) with a molecular weight of 1.0 × 104 , epichlorohydrin and congo red (CoR, C32 H22 N6 Na2 O6 S2 ) were bought from Aladdin. The chemical structure of congo red is presented in Scheme S1 in the supporting information. Sodium hydroxide, ammonia aqueous solution (30%), hydrochloric acid (37%) and acetic acid were purchased from Shanghai Chemical Reagent Co. Ltd, China. Ferric chloride hexahydrate (FeCl3 ·6H2 O) and ferrous chlorid (FeCl2 ·4H2 O) were purchased from Shanghai Chemical Reagents Company.

epichlorohydrin and 1.0 g PEI were added, and the mixture was reacted for 2 h. After the reaction, the obtained product was collected by magnetic separation, washed with deionized water and dried by freeze dryer. 2.3. Characterization methods and instruments Fourier transform infrared spectra (FT-IR) were obtained on a Nicolet, Avatar 360 FT-IR spectrometer (USA). The spectrum widths were typically in the range of 4000–400 cm−1 . Scanning electron microscopy (SEM) was investigated using a scanning electron microscope (Nova Nano SEM 230, USA). Thermogravimetric analysis (TGA) measurements were performed in nitrogen by an STA449C thermal analyzer (Germany). Powder X-ray diffraction (XRD) patterns were collected on a Bruker D8 Advance X-ray diffraction spectrometer (Germany) with Cu K␣ radiation at ␭ = 0.154 nm operating at 40 kV and 40 mA. Magnetic characterization was carried out with a vibrating sample magnetometer on a Model 6000 physical property measurement system (Quantum, USA) at 300 K. Zeta potential was conducted with a zetasizer nano potential analyzer (Malvern, ZS) using He-Ne laser at a wavelength of 632.8 nm. The N2 adsorption–desorption isotherms was carried out using a surface area analyzer (Quanta Chrome Nova1200). The specific surface area was determined by Brunauer-Emmett-Teller equation (BET) and total pore volume was defined as the maximum amount of nitrogen adsorbed at relative pressure of P/P0 = 0.99. 2.4. Adsorption of CoR by Fe3 O4 /CS-PEI polymer composite Stock solutions were prepared by dissolving congo red in deionized water to concentrations in range of 30–420 mg/L. Adsorption experiments were carried out in a constant temperature shaker. The concentration of congo red was measured by a spectrophotometer at 488 nm (the maximum absorption wavelength) after the adsorbent-adsorbate complex separated by magnet. Solution without the adsorbent was served as the control. The adsorption capacity qe (mg/g) was determined as flowing equation: qe =

(C0 − Ce ) × V m

(1)

The removal efficiency of congo red was calculated by: Removal efficiency =

C0 − Ce × 100% C0

(2)

Where C0 (mg/L) is the initial concentration of congo red, Ce (mg/L) is the final or equilibrium concentration of congo red, V (L) is the total volume of the solution, m (g) is the dosage of Fe3 O4 /CS-PEI and qe (mg/g) is the amount of congo red adsorbed per unit weight of Fe3 O4 /CS-PEI. The effects of dosage of Fe3 O4 /CS-PEI, initial pH, and initial congo red concentration on removal efficiency were studied. The pH in the experiments was original pH of the congo red solutions unless mentioned specifically. 3. Results and discussion

2.2. Preparation of the porous magnetic chitosan-polyethylenimine (Fe3 O4 /CS-PEI) polymer composite

3.1. Synthesis and characterization of Fe3 O4 /CS-PEI composite

The Fe3 O4 /CS-PEI polymer composite was synthesized using a one-pot synthesis approach. Typically, 1.3525 g FeCl3 ·6H2 O and 0.4975 g FeCl2 ·4H2 O were dissolved in 50 mL deionized water to form a homogeneous medium. 2.0 g chitosan was dissolved in 40 mL 3% acetic and the as prepared chitosan-acetic acid solution was added in to the above homogeneous medium. Subsequently, N2 was bubbled in and the temperature was heated to 90 ◦ C followed by addition of 40 mL ammonia aqueous solution. Then, 1.0 mL

The schematic representation for the synthesis of the Fe3 O4 /CSPEI is displayed in Scheme 1 and the reactions happened in the process is represented in Scheme S2 in the supporting information. In the preparation process, FeCl3 ·6H2 O reacted with FeCl2 ·4H2 O and the magnetic Fe3 O4 nanoparticles were generated in the present of ammonium hydroxide at 90 ◦ C. Chitosan (CS) was crosslinked with polyethylenimine (PEI) in the present of epichlorohydrin as a crosslinker. In alkaline condition, the epoxy

Please cite this article in press as: L. You, et al., Int. J. Biol. Macromol. (2017), https://doi.org/10.1016/j.ijbiomac.2017.10.025

G Model BIOMAC-8320; No. of Pages 9

ARTICLE IN PRESS L. You et al. / International Journal of Biological Macromolecules xxx (2017) xxx–xxx

3

Scheme 1. Synthetic procedure of the Fe3 O4 /CS-PEI polymer composite.

Fig. 1. SEM images of the Fe3 O4 /CS-PEI composite.

group in epichlorohydrin reacted with the amino/hydroxyl groups in CS and amino group in PEI, and resulted in the generation of a cross-linked copolymer. Simultaneously, the generated Fe3 O4 nanoparticles were coated by the cross-linked copolymer through affinity between Fe-O bonds from Fe3 O4 and functional groups (NH2 and −OH) from crosslinked CS-PEI component. Finally, the product was separated by magnet, washed by deionized water and dried by vacuum freeze, and then the Fe3 O4 /CS-PEI adsorbent was obtained. The copolymer product coated the Fe3 O4 nanoparticles through the intertwining of polymer chains and affinity interaction between amino/hydroxyl groups and Fe-O bonds. The Fe3 O4 /CSPEI was characterized by SEM and the image is shown in Fig. 1. It is obvious that the Fe3 O4 /CS-PEI is porous. Its specific surface is rough and coarser. The Fe3 O4 nanoparticles was coated by the polymers and distributed throughout the material. The macromolecule chains of the components in Fe3 O4 /CS-PEI crosslinked assembled together and formed structure of networks, which could promote the adsorption and bridging performance as an adsorbent. The FT-IR spectroscopy of the Fe3 O4 nanoparticles, Chitosan (CS), polyethylenimine (PEI) and Fe3 O4 /CS-PEI were characterized (Fig. 2a–d, respectively). In the spectra of (Fig. 2a, the adsorption peaks at 582 cm−1 was the characteristic absorption of Fe-O bonds [23]. In the FT-IR spectra of CS (Fig. 2b), the broad strong peaks

Fig. 2. FT-IR spectrum of (a) Fe3 O4 , (b) Chitosan (CS), (c) polyethylenimine (PEI) and (d) Fe3 O4 /CS-PEI.

at 3452 cm−1 were due to the O H/–NH2 stretching vibration; absorption at 2905 cm−1 was due to the C H stretching vibration; the peaks at 1606 cm−1 were derived from scissoring vibrations of NH2 ; 1423 and 1381 cm−1 were derived from the scanty amount

Please cite this article in press as: L. You, et al., Int. J. Biol. Macromol. (2017), https://doi.org/10.1016/j.ijbiomac.2017.10.025

G Model BIOMAC-8320; No. of Pages 9 4

ARTICLE IN PRESS L. You et al. / International Journal of Biological Macromolecules xxx (2017) xxx–xxx

Fig. 4. Magnetic hysteresis curves of Fe3 O4 nanoparticles and Fe3 O4 /CS-PEI. Fig. 3. XRD patterns of (a) CS-PEI (chitosan-polyethylenimine), (b) Fe3 O4 particles, (c) Fe3 O4 /CS-PEI.

of O C NH2 and the bending vibration of C H, respectively. The absorption at 1084 and 891 cm−1 belonged to the C O bonds and pyranoid ring stretching, respectively [24]. In the FT-IR spectra of PEI (Fig. 2c), the broad strong peaks at 3337 cm−1 were derived from symmetric and asymmetric N H stretching bands; strong bands at 2908 and 2843 cm−1 were due to the C H stretching mode. In the region from 1500 to 1100 cm−1 , PEI shows several types of modes C N stretching at 1112 cm−1 , CH2 twisting around 1304 cm−1 , and NH bending parallel to the chain axis and N H bending mixed with CH2 scissors from 1590 to 1304 cm−1 [25,26]. These bands was weaken in the spectrum of Fe3 O4 /CS-PEI (Fig. 2d) indicated the crosslinking reaction happened and N H reduced. The enhanced C N stretching at 1152 cm−1 and the new aliphatic C O at 1021 cm−1 also suggested the CS and PEI was crosslinked by epichlorohydrin. The peaks observed at 3437, 2920, 2852, 1611, 1463, 1337, 1319 cm−1 were similar to the characteristic peaks of CS and PEI. Moreover, the band at 578 cm−1 was attributed to Fe-O bonds which confirmed the existence of Fe3 O4 nanoparticles. It suggests that CS, PEI and Fe3 O4 nanoparticles are the main components of the Fe3 O4 /CS-PEI. The crystalline structure and phase purity of Fe3 O4 /CS-PEI was characterized by powder X-ray diffraction (XRD) as shown in Fig. 3. It was obvious the typical diffraction peaks (220, 311, 400, 511 and 440) of the Fe3 O4 particles well all appeared in the XRD pattern of the Fe3 O4 /CS-PEI [23]. It suggested that the crystalline structure of Fe3 O4 nanoparticles was well retained under the whole fabrication process. Compared with the XRD pattern of the chitosan-polyethylenimine (CS-PEI) composites without Fe3 O4 , a broad peak presented at 2␪ = 8.7 and 20.3 ◦ came out of the amorphous phase of the PEI and CS component. The above results confirmed that magnetic Fe3 O4 nanoparticles were coated in the Fe3 O4 /CS-PEI. The magnetic hysteresis curves were shown in Fig. 4. The saturation magnetization of Fe3 O4 nanoparticles and Fe3 O4 /CS-PEI were 40.1 and 15.6 emu/g, respectively. The saturation magnetization value of Fe3 O4 /CS-PEI was lower than Fe3 O4 was attributed to the crosslinking coating of CS and PEI components. The magnetic hysteresis curve showed the Fe3 O4 /CS-PEI was a superparamagnetic material and presented well magnetic separation property with a high saturation magnetization of 15.6 emu/g. Nitrogen physisorption measurements were performed to analyze the textural characteristics of the adsorbent. The nitrogen adsorption/desorption isotherms and the pore size distribution of the Fe3 O4 /CS-PEI is showed in Fig. 5. The profiles showed that it was a type IV adsorption isotherm with hysteresis loops in the relative range of 0.45–1.0, which indicated the existence of abundant mesoporous structures [27]. The isotherm of the Fe3 O4 /CS-PEI dis-

Fig. 5. Nitrogen adsorption/desorption isotherms of Fe3 O4 /CS-PEI. The inserted image was the pore size distribution.

played a H3-type hysteresis loop, which indicated that mesopores have cylindrical pore geometries and the materials also comprised of aggregates forming slit like pores [28,29]. The BET surface area of the Fe3 O4 /CS-PEI was 109.2 m2 /g. The adsorption average pore width was 15.08 nm and the single point adsorption total pore volume at P/Po = 0.99 was 0.29 cm3 /g. The t-Plot micropore volume was 0.004 cm3 /g and the Barrett−Joyner−Halenda desorption cumulative volume of pores was 0.24 cm3 /g. It showed the existence of different pore sizes in the hybrid material. The Fe3 O4 /CS-PEI possessed multi-level pore distribution in range of 1 nm to 65 nm (the inserted image in Fig. 5). Moreover, the almost vertical tails near to P/P0 = 0.99 suggested the presence of macropore. The high specific surface areas, mesoporous channels and unique pore distribution of Fe3 O4 /CS-PEI could benefit the contact between adsorption sites and adsorbate molecules, thus could enhance the CoR removal. 3.2. Removal of CoR 3.2.1. Effect of Fe3 O4 /CS-PEI dosage on CoR removal The schematic representation for the CoR removal process by Fe3 O4 /CS-PEI is displayed in Scheme 2. The removal of CoR by different dosages of Fe3 O4 /CS-PEI was studied. The results in Fig. 6 showed that the removal efficiency of CoR increased sharply from 38.8% to 99.4% as the dosage of Fe3 O4 /CS-PEI increased, due to the increase of active adsorption sites and larger surface area at higher amount of the adsorbent [30]. It reached a peak value of 99.3% when the dosage was over 1.4 g/L. The adsorption capacity (qe ) of the Fe3 O4 /CS-PEI was over 779 mg/g when the dosage was below 0.1 g/L and it reached 971 mg/g when the dosage of the absorbent

Please cite this article in press as: L. You, et al., Int. J. Biol. Macromol. (2017), https://doi.org/10.1016/j.ijbiomac.2017.10.025

G Model BIOMAC-8320; No. of Pages 9

ARTICLE IN PRESS L. You et al. / International Journal of Biological Macromolecules xxx (2017) xxx–xxx

5

Scheme 2. Schematic representation of CoR removal process by Fe3 O4 /CS-PEI.

Fig. 6. Effect of the dosage of Fe3 O4 /CS-PEI on the removal efficiency (CoR concentration 100 mg/L, temperature 20 ◦ C, pH = 7 and reaction time 160 min). Fig. 8. Zeta potential of Fe3 O4 /CS-PEI in various pH conditions (aqueous solution).

Fig. 7. Effect of initial pH on removal efficiency. (CoR concentration 100 mg/L, dosage of Fe3 O4 /CS-PEI 0.25 g/L, temperature 20 ◦ C and adsorption time 160 min).

was 0.04 g/L. The results showed that the Fe3 O4 /CS-PEI removed CoR from aqueous solutions efficiently. 3.2.2. Effect of initial pH on decolorization efficiency The pH determines the adsorption because it has significant impact on the surface properties of adsorbates and adsorbents. To investigate the effect of initial pH on CoR removal, adsorption experiments at different initial pH in the range of 3–13 were carried out. The adsorption capacity of CoR was decreased from 418.7 to 302.6 mg/g when the pH value increased from 3 to 13 (Fig. 7). CoR was an anionic dye with the sulphonate (SO3 ) groups and the Fe3 O4 /CS-PEI is positive with amino groups, thus the Fe3 O4 /CS-PEI could removal CoR by electrostatic adsorption in the solution. At lower pH, the H+ ion concentration in the system was high and surface of the Fe3 O4 /CS-PEI composite trended to become more

positive charged (Fig. 8) due to adsorption of H+ ions and protonation [3,30]. Therefore, at lower pH, higher electrostatic attraction existed between Fe3 O4 /CS-PEI and CoR and resulted in higher CoR adsorption. Meanwhile, with an increase of pH the Fe3 O4 /CS-PEI become less positive charged and even turned negative charged, which led to the reduce in electrostatic attraction and decrease of adsorption capacity. Moreover, the existence of excess OH− could competition with CoR for the adsorption sites, thus caused the reducing of the adsorption capacity. It found the negative charged surface sites of the Fe3 O4 /CS-PEI at high pH conditions still presented effective removal performance of the anionic CoR, it was caused by the high specific surface areas and unique multi-pore structure of Fe3 O4 /CS-PEI. Moreover, dye–polymer–dye bridges adsorption also happened through affinity interaction between the functional groups such as OH, −SO3 , NH2 and NH in the dye molecules and polymer chains of Fe3 O4 /CS-PEI, which also contributed to the adsorption capacity. 3.2.3. Effect of initial CoR concentration The removal of different concentration CoR by Fe3 O4 /CS-PEI is presented in Fig. 9. The results show that the percentage removal of CoR decreased with the increase in initial CoR concentration, but the amount of CoR adsorbed on per mass of Fe3 O4 /CS-PEI (qe ) increased. Though the increase in initial concentration led to an increase in adsorption capacity, it was negligible compared with the sharply increase of the concentration of CoR, thus the removal percentage decreased [31,32]. Meanwhile, The increase in initial concentration caused the augment in driving force of the concentration gradient, accelerated the diffusion of CoR molecules into the adsorbent and decreased the resistance of adsorption, thus increased the uptake of CoR [33]. As the initial concentration increased from 30 to 420 mg/L, the amounts of CoR adsorbed on

Please cite this article in press as: L. You, et al., Int. J. Biol. Macromol. (2017), https://doi.org/10.1016/j.ijbiomac.2017.10.025

G Model BIOMAC-8320; No. of Pages 9

ARTICLE IN PRESS L. You et al. / International Journal of Biological Macromolecules xxx (2017) xxx–xxx

6

Fig. 9. Effect of the initial dye concentration (temperature 20 ◦ C, dosage of Fe3 O4 /CSPEI 0.2 g/L, pH = 7, adsorption time 160 min).

at any time (mg/g) and k1 , k2 and kid are rate constants of pseudofirst, pseudo-second and intra-particle diffusion; C is a constant describing the thick-ness of boundary layer. The simulated kinetic curves were presented in Fig. S1- S3 (see the supporting information) and the kinetic parameters for the adsorption of CoR onto Fe3 O4 /CS-PEI are summarized in Table 1. The R2 values (0.7693 < R2 < 0.9542) showed that the pseudo-firstorder kinetics did not adequately fit the experimental values. The pseudo second-order kinetic model, described all the adsorption processes very well according to the correlation coefficients (0.9865 < R2 < 0.9968). Additionally, the qe values calculated from the pseudo-second-order model were very close to the experimental values (qe ,exp ), which also confirmed that the adsorption process followed the pseudo-second-order kinetics. The change of adsorption rate constants at various temperatures suggested there was not only chemical but also physical interaction in the process of CoR removal. The well-fitting of the pseudo-second-order equation suggested the adsorption process was interaction controlled with chemisorption, which involved valance forces through sharing or exchange of electrons between the Fe3 O4 /CS-PEI and the CoR [37]. The Fe3 O4 /CS-PEI was not only positive charged with lots of amino groups, but also possessed high specific surface area and unique porous structure with multi-level pore distribution. Thus, in order to investigate if the intra-particle diffusion mechanism played an important role in CoR removal, the kinetic curves were also fitted by intra-particle diffusion model. The results indicated that the intraparticle diffusion equation fit the experimental data with relatively high R2 values (0.9073–0.9758) and the Cid values (312.9-512.1) were all larger than zero, which suggested the CoR adsorption can be affected by the boundary layer diffusion [38]. In order to compare the validity of the kinetics equations, a normalized standard deviation qe (%) is calculated by Eq. (6):

 qe (%) = 100 Fig. 10. Kinetic curves of the CoR removal by Fe3 O4 /CS-PEI (CoR concentration 100 mg/L, dosage of Fe3 O4 /CS-PEI 0.15 g/L).

the adsorbent increased from 144.7 to 1141.3 mg/g, which indicated that the Fe3 O4 /CS-PEI possessed enhanced high adsorption efficiency for CoR removal. 3.3. Adsorption kinetics The kinetics of adsorption describes the rate of uptake of adsorbate onto adsorbent and the rate controls the equilibrium time. According to Fig. 10, higher temperature led to faster rate and all the curves finally reached equilibrium in 160 min. To determine the kinetics of CoR adsorbed by Fe3 O4 /CS-PEI, the kinetic curves were simulated using the pseudo-first-equation, the pseudo-secondorder equation and intra-particle diffusion models. The pseudo-first-order rate model is based on solid capacity and as described by Eq. (3) [34]: ln(qe − qt ) = lnqe − k1 t

(3)

The pseudo-second-order equation can be represented as Eq. (4) [35]: t 1 t = + qt qe k2 q2e

(4)

The intra-particle diffusion model is shown as Eq. (5) [36]: qt = kid t 0.5 + C

(5)

Where qe is the amount of CoR adsorbed by per unit weight of the adsorbent (mg/g) at equilibrium, qt is the amount of CoR adsorbed

[(qe − qe,cat )/qe ]

2

N−1

(6)

Where N is the number of data, qe (mg/g) is the experimental value and qe,cat (mg/g) is the calculated value by the kinetics models. The lowest value of qe in Table 1 also proved that the adsorption kinetics of CoR by Fe3 O4 /CS-PEI was best fitted by the pseudosecond-order model. Accordingly, the rate constants (k2 ) of the pseudo second-order model are adopted to calculate the activation energy of the adsorption process using the Arrhenius equation. lnk2 =

−Ea + ln A RT

(7)

where, k2 is the rate constant of the pseudo second-order model, Ea is activation energy, R is the gas constant (8.314 J/(mol·K)), A is Arrhenius factor and T is the temperature in Kelvin. The activation energy could be determined from the slope of the plot of lnk2 versus 1/T. It was found lnk2 had a good linear relation with 1/T and the correlation coefficients of the plot was as high as 0.995 (see the plot in Fig. S4). The activation energy (Ea ) in this study was 34.08 kJ/mol, indicating that chemisorption existed in the process.

3.4. Adsorption isotherms The adsorption isotherms of CoR at various temperatures are shown in Fig. 11. The adsorption data was analyzed by fitting isotherms models including the Langmuir, Freundlich, Tempkin, Dubinin–Radushkevich and Redlich–Peterson. The fitting curves are listed in Fig. S5-S9 in the supporting information. The Langmuir adsorption isotherm assumes that adsorption takes place at homo-

Please cite this article in press as: L. You, et al., Int. J. Biol. Macromol. (2017), https://doi.org/10.1016/j.ijbiomac.2017.10.025

G Model

ARTICLE IN PRESS

BIOMAC-8320; No. of Pages 9

L. You et al. / International Journal of Biological Macromolecules xxx (2017) xxx–xxx

7

Table 1 Kinetic parameters for adsorption of CoR by Fe3 O4 /CS-PEI. Kinetic models

Parameters

Pseudo first order

qe ,exp (mg/g) qe (mg/g) k1( min−1 ) R2 qe (%) qe (mg/g) k2 (g/mg min) R2 qe (%) kid (mg/g min0.5 ) Cid R2 qe (%)

Pseudo second order

Intra-particle diffusion

Temperature (◦ C) 20

25

30

35

40

598.5 357.2 0.0199 0.7693 57.44 588.2 1.7 × 10−4 0.9865 19.58 18.10 312.9 0.9073 21.03

745.1 285.9 0.0206 0.8603 60.01 769.2 2.3 × 10−4 0.9965 12.10 16.85 512.5 0.9758 13.66

863.7 370.6 0.0201 0.8890 48.43 833.3 2.0 × 10−4 0.9968 11.86 24.64 526.1 0.9682 15.30

1018.7 709.3 0.0286 0.9542 21.96 1111.1 0.8 × 10−4 0.9904 17.31 42.99 477.9 0.9522 24.94

1156.4 993.9 0.0338 0.8224 43.89 1250 0.7 × 10−4 0.9932 14.74 52.22 508.3 0.9629 26.73

absence of steric hindrance between sorbed and incoming particles [42]. Its linear form equation is described as following: ln qe = ln QD − ˇε2

(11)

Where ˇ is a constant related to the mean free energy of adsorption. QD is the theoretical saturation capacity and ε is the Polanyi potential, which is equal to ε = − RTln(1 + 1/Ce ), R (R = 8.314 J/mol/K) is the gas constant and T (K) is the absolute temperature. Redlich-Peterson model [43] is a three parameter adsorption isotherm equation, which amends inaccuracies of two parameter Langmuir and Freundlich isotherm equations in some adsorption systems. The Redlich-Peterson isotherm equation is expressed as: ln(QRP bRp

Fig. 11. Adsorption isotherms of CoR at various temperatures by Fe3 O4 /CS-PEI (CoR concentration 30–420 mg/L, dosage of Fe3 O4 /CS-PEI 0.2 g/L and adsorption time 160 min.).

geneous sites within the adsorbent [39]. The Langmuir isotherm equation is represented by the following equation: 1 1 1 1 = +( ) qe Qm Qm KL Ce

(8)

where qe is the amount of CoR adsorbed by per unit weight of Fe3 O4 /CS-PEI, Ce is the equilibrium concentration of the CoR, Qm (mg/L) is the maximum monolayer adsorption capability and KL is a constant related to the free energy of adsorption. The Freundlich isotherm model [40] is usually used for a heterogeneous surface and a nonuniform distribution of adsorption heat over surface without a saturation of adsorption sites. The Freundlich model is listed as Eq. (6): 1nqe = lnKF +

1 lnCe n

(9)

Where KF (L/mg) and n (dimensionless) are constants related to adsorption capacity and intensity. By ignoring the extremely low and large value of concentrations, the Tempkin model assumes that heat of adsorption of all molecules in the layer would decrease linearly rather than logarithmic with coverage [41]. The equation is presented as: qe = B1 ln Kt + B1 ln Ce

(10)

Where Kt is the equilibrium binding constant corresponding to the maximum binding energy and constant B1 is related to the heat of adsorption. A plot of qe versus lnCe enables the determination of the isotherm constants Kt and B1 . The D-R isotherm equation is more general because it does not assume a homogeneous surface or a constant sorption potential or

Ce − 1) = lnbRp + ˛ ln Ce qe

(12)

Where QRp and bRP is are parameters of the R–P isotherm equation. ␣ is the exponent which lies between 0 and 1. Because there are three unknowns, QRp , bRp and ␣, thus the parameters of the equations were determined by minimizing the distance between the experimental data and the theoretical model predictions through trial and error. In order to compare the validity of the isotherm equations, chisquare (␹2 ) test is used, which is used to calculate the error in the experimental data and the calculated data by using particular isotherm model. The chi-square (␹2 ) statistic is the sum of the squares of the differences between the experimental data and calculated data by using various models, with each squared difference divided by the corresponding calculated data [44]. 2 =

 (qe,exp − qe,cal )2 qe,cal

(13)

Where qe,cal is the equilibrium capacity obtained by calculating from the model (mg/g) and qe,exp is the experimental data on the equilibrium capacity (mg/g). The isotherm parameters calculated from plots of these isotherm equations are given in Table 2. It is evident from these data that the surface of Fe3 O4 /CS-PEI was made up of homogeneous and heterogeneous adsorption patches. The adsorption of CoR by Fe3 O4 /CS-PEI followed the Redlich-Peterson isotherm best when the correlation coefficients (R2 ) were compared, of which the R2 values at various temperatures were all larger than 0.9936. All ␣ values (between 0.49 and 0.7604) are less than 1, which was caused by the existence of a solid impediment between pores and adsorbate [45]. Meanwhile, error analysis results showed the ␹2 values (Table 2) calculated by Redlich-Peterson model were lowest and narrowly distributed ranging in 2.589–23.7, which also showed the reaction followed Redlich-Peterson isotherm excellently. It was found that Freundlich isotherm equation described the experimental data well with high coefficients (0.9406–0.9869) at various

Please cite this article in press as: L. You, et al., Int. J. Biol. Macromol. (2017), https://doi.org/10.1016/j.ijbiomac.2017.10.025

G Model

ARTICLE IN PRESS

BIOMAC-8320; No. of Pages 9

L. You et al. / International Journal of Biological Macromolecules xxx (2017) xxx–xxx

8

Table 2 Fitting isotherms parameters for the removal of CoR by Fe3 O4 /CS-PEI at various temperatures. Isotherm models

Isotherm parameters

Langmuir

Qm (mg/g) KL (L/mg) R2 ␹2 KF (mg1−1/n L1/n /g) n R2 ␹2 Kt (L/mg) B1 R2 ␹2 QD (mg/g) ␤(mg2 /kJ2 ) R2 ␹2 QRP (mg/g) bRP (L/mg) ␣ R2 ␹2

Freundlich

Tempkin

Dubinin–Radushkevich

Redlich–Peterson

Temperature (◦ C) 20

25

30

35

40

769.2 0.2321 0.9427 460.4 178.36 2.951 0.9514 58.71 1.699 168.8 0.9009 175.6 721.9 0.6374 0.7533 657.8 200.3 2.396 0.683 0.9936 33.77

1111 0.0332 0.9564 385.6 69.88 1.822 0.9869 10.85 0.1926 330.4 0.9366 792.1 900.1 10.55 0.6527 882.6 88.16 2.155 0.49 0.9946 9.966

1000 0.1449 0.8965 729.8 135.21 2.175 0.9794 45.61 0.5618 296.4 0.8513 242.9 1008 1.089 0.6232 1200 122.9 0.9762 0.496 0.9948 7.315

1667 0.0368 0.9811 127.9 115.72 1.944 0.9869 37.85 0.2980 391.9 0.9701 371.9 1183 5.425 0.71 980.8 98.37 0.8641 0.6941 0.9954 9.893

2000 0.0323 0.956 59.57 163.69 2.128 0.9406 130.11 0.3938 408.5 0.9835 18.61 1437 7.839 0.898 431.9 537.3 0.3071 0.7604 0.9996 2.589

Table 3 Thermodynamic parameters for the removal of CoR by Fe3 O4 /CS-PEI. T(K)

lnKc

G (kJ/mol)

H (kJ/mol)

S (J/(mol·K))

293.15 298.15 303.15 308.15 313.15

3.24 3.42 4.26 4.14 4.59

−7.89 −8.49 −10.73 −10.61 −11.96

52.36

205.51

potential resistance. The positive value of H (52.36 kJ/mol) indicated that the adsorption process was endothermic in nature, further confirming that higher temperature was favorable for the CoR adsorption. The entropy change (S) in adsorption process was 205.51 J/(mol·K), which reflected the increased randomness at the solid/solution interface. 3.6. Comparison with other adsorbents

temperatures, but with relative high ␹2 values (20.85–130.1). The value of n indicates the type of isotherm to be favorable (n > 1), unfavorable (n < 1), linear (n = 1). The high values of n (n > 1.8) at equilibrium at all the temperature conditions indicated that the CoR removal process was a favorable. It was observed that the Langmuir (0.896 < R2 < 0.9811), Tempkin (0.851 < R2 < 0.9835) and Dubinin-Radushkevich (0.623 < R2 < 0.898) isotherms did not adequately fit the experimental values. The low coefficients indicated that the removal of CoR could not be described by the three models. Meanwhile, the high and widely distributed chi-square values (59.57–729.8, 18.61–792.1 and 431.9–1200, respectively) in Table 2 also showed the adsorption process didn’t follow the Langmuir, Tempkin or Dubinin-Radushkevich isotherms. 3.5. Thermodynamic parameters To further investigate the effect of the temperature on the adsorption and explore the mechanism involved in the adsorption process, the thermodynamic behavior was evaluated using the following equations: G = −RT lnKc

(14)

S H lnKc = − R RT

(15)

Where R (8.314 J/(mol·K)) denotes the universal gas constant, T (K) represents the absolute temperature, Kc is the thermodynamic equilibrium constant, G is free energy change (kJ/mol), H is enthalpy change (kJ/mol) and S (J/(mol·K)) is entropy change. lnKc was determined by plotting ln(qe /Ce ) versus qe and extrapolating zero qe. H and S were calculated from the slop and intercept of van’t Hoff plot of ln Kc versus 1/T. The results are given in Table 3. The negative values of G at various temperatures indicated that the adsorption force were strong enough to overcome the

The experimental maximum adsorption capacity of Fe3 O4 /CSPEI polymer composite for the removal of CoR has been compared with the other reported adsorbents is given in Table 4. It is clear that the Fe3 O4 /CS-PEI composite is very promising compared with other adsorbents for CoR adsorption described in Table 4. This indicates that Fe3 O4 /CS-PEI has an ultrahigh adsorption capacity for CoR removal. 4. Conclusions A new porous magnetic chitosan-polyethylenimine (Fe3 O4 /CSPEI) polymer composite was synthesized by crosslinking chitosan with polyethylenimine using a facile one-pot synthesis approach and applied to remove CoR from aqueous solutions. The Fe3 O4 /CSPEI polymer composite owned high positive charge, large surface area, multi-level pore distribution and magnetic responsiveness. The BET surface area of the Fe3 O4 /CS-PEI is 109.2 m2 /g with average pore width of 15.08 nm and total pore volume of 0.24 cm3 /g. The Fe3 O4 /CS-PEI showed enhanced high capacity (1876 mg/g at 40◦ C) for CoR removal in aqueous solutions. It removed over 99.3% of CoR (100 mg/L) when the dosage was over 1.4 g/L. A high temperature was benefit to CoR adsorption. The adsorption capacity decreased when the pH value increased, but at high pH (over 10) conditions it still presented effective performance for CoR removal. Kinetics studies suggested that the adsorption mechanism of CoR followed the pseudo-second model and it was also affected by the boundary layer diffusion. The thermodynamics study indicated the adsorption of CoR by Fe3 O4 /CS-PEI followed the Redlich-Peterson isotherm equation with high correlation coefficients values over 0.9936 and low chi-square ranging in 2.589-23.7. The overall negative free energy change (G) showed that it was a spontaneous process. The positive value of H (52.36 kJ/mol) indicated that the adsorption process was endothermic. The entropy change (S)

Please cite this article in press as: L. You, et al., Int. J. Biol. Macromol. (2017), https://doi.org/10.1016/j.ijbiomac.2017.10.025

G Model BIOMAC-8320; No. of Pages 9

ARTICLE IN PRESS L. You et al. / International Journal of Biological Macromolecules xxx (2017) xxx–xxx

9

Table 4 Comparison of CoR maximum adsorption capacity (qm ) of various adsorbents. Adsorbents

qm (mg/g)

References

Schiff base-CS grafted 1-monoguluronic acid Fe3 O4 @nSiO2 @mSiO2 core-shell microspheres Fe3 O4 /Bi2 S3 microspheres Ag-doped hydroxyapatite MWCNTs/Calcined eggshell Magnetic CS/carrageenan ampholytic microspheres Mg–Al-layered double hydroxide Calcium alginate beads impregnated with nano-goethite Zinc peroxide nanomaterial MIL-68 (In) nano-rods Hierarchical porous ZnO Cetyltrimethyl ammonium bromide (CTAB) modified pumice TEP A-modified sugarcane bagasse Fe3 O4 /CS-PEI

459.75 1428 92.24 554.54 139.99 212.7 37.16 181.1 208 1273 334 27.32

[46] [16] [31] [47] [48] [49] [50] [51] [52] [53] [5] [54]

790.7 1876

[55] This work

in adsorption process was 205.51 J/(mol·K), which reflected the increased randomness at the solid/solution interface. The activation energy (Ea ) of the adsorption process was 34.08 kJ/mol, indicating that chemisorption existed in the process. The results demonstrate that the Fe3 O4 /CS-PEI is a promising adsorbent with high adsorption performance for the efficient removal of dye pollutants from wastewater. Acknowledgements We gratefully acknowledge the financial support of the National Natural Science Foundation of China (Grant No. 51503040 and Grant No. 31401609). References [1] L. Hu, Z. Yang, L. Cui, Y. Li, Y. H. H. Ngo, Q. Wang, H. Wei, L. Ma, B. Yan, Chem. Eng. J. 287 (2016) 545–556. [2] J. Shu, Z. Wang, Y. Huang, N. Huang, C. Ren, W. Zhang, J. Alloy. Compd. 633 (2015) 338–346. [3] M.O. Ansari, R. Kumar, S.A. Ansari, S.P. Ansari, M.A. Barakate, A. Alshahrie, M.H. Cho, J. Colloid Interf. Sci. 496 (2017) 407–415. [4] V. Munagapati, D. Kim, Ecotoxicol. Environ. Saf. 141 (2017) 226–234. [5] C. Lei, M. Pi, C. Jiang, B. Cheng, J. Yu, J. Colloid Interf. Sci. 490 (2017) 242–251. [6] U. Habiba, T.A. Siddique, T.C. Joo, A. Salleh, B.C. Ang, A.M. Afif, Carbohyd. Polym. 157 (2017) 1568–1576. [7] C. Ma, F. Wang, C. Zhang, Z. Yu, J. Wei, Z. Yang, Y. Li, Z. Li, M. Zhu, L. Shen, G.G. Zeng, Chemosphere 168 (2017) 80–90. [8] H. Jalife-Jacobo, R. Feria-Reyes, O. Serrano-Torres, S. Gutiérrez-Granados, Juan M. Peralta-Hernández, J. Hazard Mater 319 (2016) 78–83. [9] F. Bosco, C. Mollea, B. Ruggeri, Environ. Technol. 1 (2016) 1–8. [10] S. Thakur, O. Pandey, Int. J. Biol. Macromol. (2017), http://dx.doi.org/10.1016/ j.ijbiomac.2017.05.087. [11] A. Saad, I. Bakas, J.-Y. Piquemal, S. Nowak, M. Abderrabba, M.M. Chehimi, Appl. Surf. Sci. 367 (2016) 181–189. [12] Y. Zhang, L. Bai, W. Zhou, R. Lu, H. Gao, S. Zhang, J. Mol. Liq. 219 (2016) 88–94. [13] H. Vijwani, M.N. Nadagoud, V. Namboodiri, S.M. Mukhopadhyay, Chem. Eng. J. 268 (2015) 197–207. [14] L. You, L. Song, A. Wang, F. Lu, Q. Zhang, Colloids and surfaces A: physicochem, Eng. Aspects 520 (2017) 841–849. [15] S.J. Boardman, R. Lad, D.C. Green, P.D. Thornton, J. Appl. Polym. Sci. 134 (2017) 44846. [16] R. Huang, L. Zhang, P. Hu, J. Wang, Int. J. Biol. Macromol. 86 (2016) 496–504. [17] A. Anitha, S. Sowmya, P.T.S. Kumar, S. Deepthi, K.P. Chennazhi, H. Ehrlich, M. Tsurkan, R. Jayakumar, Prog. Polym. Sci. 39 (2014) 1644–1667. [18] J. Wang, Y. Chen, Y. Wang, S. Yuan, G. Sheng, H. Yu, RSC Adv. 2 (2012) 494–500. [19] S. Liu, D. Zhou, J. Yang, H. Zhou, J. Chen, T. Guo, J. Am. Chem. Soc. 139 (2017) 5102–5109. [20] Y. Ma, B. Zhang, H. Ma, M. Yu, L. Li, J. Li, Sci. Chin. Mater. 59 (2016) 38–50.

[21] C. Ding, Y. Sun, Y. Wang, J. Li, Y. Lin, W. Sun, C. Luo, Int. J. Biol. Macromol. 99 (2017) 578–585. [22] K. Li, Z. Zeng, J. Xiong, L. Yan, H. Guo, S. Liu, Y. Dai, T. Chen, Colloids Surf. A 465 (2015) 113–123. [23] P. Wang, X. Wang, S. Yu, Y. Zou, J. Wang, Z. Chen, N. Alharbi, A. Alsaedi, T. Hayat, Y. Chen, X. Wang, Chem. Eng. J. 306 (2016) 280–288. [24] S. Wang, L. Zhang, B. Yan, H. Xu, Q. Liu, H. Zeng, J. Phys. Chem. C 119 (2015) 7327–7339. [25] H.S. Choi, T. Ooya, S.C. Lee, S. Sasaki, M. Kurisawa, H. Uyama, N. Yui, Macromolecules 37 (2004) 6705–6710. [26] C.S. Harris, M.A. Ratner, D.F. Shriver, Macromolecules 20 (1987) 1778–1781. [27] R. Jiang, H.Y. Zhu, J.B. Li, F.Q. Fu, J. Yao, S.T. Jiang, G.M. Zeng, Appl. Surf. Sci. 364 (2015) 604–612. [28] H. Zhu, R. Jiang, J. Li, Y. Fu, S. Jiang, J. Yao, Sep. Purif. Technol. 179 (2017) 184–193. [29] M. Nguyen-Le, B. Lee, Chem. Eng. J. 281 (2015) 20–33. [30] V.K. Gupta, D. Pathania, S. Sharma, S. Agarwal, P. Singh, J. Mol. Liq. 177 (2013) 325–334. [31] M. H. Isa, L. S. Lang, F. A. H. Asaari, H. A. Aziz, N. A. Ramli and J. P. A. Dhas, Dyes Pigments 74(2007) 446–453. [32] L. You, L. Song, F. Lu, Q. Zhang, Polym. Eng. Sci. 56 (2016) 1213–1220. [33] K. Singh, D.H. Lataye, K.L. Wasewar, J. Fluorine Chem. 194 (2017) 23–32. [34] L. Zhou, S. Pan, X. Chen, Y. Zhao, B. Zou, M. Jin, Chem. Eng. J. 257 (2014) 10–19. [35] J. Gong, B. Wang, G. Zeng, C. Yang, C. Niu, Q. Niu, W. Zhou, Y. Liang, J. Hazard, Mater. 164 (2009) 1517–1522. [36] G.A. Say˘gılı, F. Güzel, J. Disper, Sci. Technol. 38 (2017) 381–392. [37] Q. Li, Q. Yue, H. Sun, Y. Su, B. Gao, J. Environ. Manage. 91 (2010) 1601–1611. [38] M. Wawrzkiewicz, P. Bartczak, T. Jesionowski, Int. J. Biol. Macromol. 99 (2017) 754–764. [39] I. Langmuir, J. Am. Chem. Soc. 40 (1918) 1361–1403. [40] H.M.F. Freundlich, J. Phys. Chem. 57 (1906) 385–470. [41] M.I. Temkin, V.M. Pyzhev, Acta Physiochim. URSS 12 (1940) 327–356. [42] M.M. Dubinin, L.V. Radushkevich, Sci. USSR 55 (1947) 331–333. [43] O. Redlich, D.L. Peterson, J. Phys. Chem. 63 (1959) 1024–1026. [44] V.S. Munagapati, D.S. Kim, Ecotoxicol. Environ. Saf. 141 (2017) 226–234. [45] F. Wu, B. Liu, K. Wu, R. Tseng, Chem. Eng. J. 162 (2010) 21–27. [46] B. Yuan, L.G. Qiu, H. Su, C. Cao, J. Jiang, Int. J. Biol. Macromol. 82 (2016) 355–360. [47] C. Srilakshmi, R. Saraf, Microporous Mesoporous Mater. 219 (2016) 134–144. [48] E.N. Seyahmazegi, R. Mohammad-Rezaei, H. Razmi, Chem. Eng. Res. Des. 109 (2016) 824–834. [49] X. Liang, J. Duan, Q. Xua, X. Wei, A. Lu, L. Zhang, Chem. Eng. J. 317 (2017) 766–776. [50] R. Shan, L. Yan, Y. Yang, K. Yang, S. Yu, H. Yu, B. Zhu, B. Du, J. Ind. Eng. Chem. 21 (2015) 561–568. [51] B. Baheri, R. Ghahremani, M. Peydayesh, M. Shahverdi, T. Mohammadi, Res Chem. Intermed. 42 (2016) 5309–5328. [52] S. Chawla, H. Uppal, M. Yadav, N. Bahadur, N. Singh, Ecotoxicol. Environ. Saf. 135 (2017) 68–74. [53] L. Jin, X. Qian, J. Wang, H. Aslan, M. Dong, J. Colloid Interf. Sci. 453 (2015) 270–275. [54] H. Shayesteh, A. Rahbar-Kelishami, R. Norouzbeigi, J. Mol, J. Mol. Liq. 221 (2016) 1–11. [55] J. Yu, W. Xiong, J. Zhu, J. Chen, R. Chi, Clean Technol. Environ. Policy 19 (2017) 517–525.

Please cite this article in press as: L. You, et al., Int. J. Biol. Macromol. (2017), https://doi.org/10.1016/j.ijbiomac.2017.10.025