Removal of aqueous Cr(VI) by a magnetic biochar derived from Melia azedarach wood

Bioresource Technology 256 (2018) 1–10

Contents lists available at ScienceDirect

Bioresource Technology journal homepage: www.elsevier.com/locate/biortech

Removal of aqueous Cr(VI) by a magnetic biochar derived from Melia azedarach wood Xin Zhanga,b, Lei Lvd, Yingzhi Qina, Min Xua, Xianbin Jiac, Zhihua Chena,

T

⁎

a

School of Environment, Henan Normal University, Key Laboratory for Yellow River and Huai River Water Environment and Pollution Control, Xinxiang 453007, China Guangdong Provincial Key Laboratory of Environmental Pollution Control and Remediation Technology, China c School of Chemistry and Chemical Engineering, Henan Normal University, Xinxiang 453007, China d Sinosteel Wuhan Safety & Environmental Protection Research Institute Co., Ltd., Wuhan No.1244, Heping Road, Qingshan District, Wuhan 430081, China b

A R T I C L E I N F O

A B S T R A C T

Keywords: Melia azedarach wood Hexavalent chromium Magnetization Adsorption mechanism

Magnetic biochar (MMABC) prepared from Melia azedarach wood was used for aqueous Cr(VI) removal. MMABC was a mesoporous material with SBET 5.219 m2/g and superparamagnetic magnetization 17.3 emu/g contributed by the contained Fe3O4. The MMABC showed higher removal efficiency (99.8%) than biochar under conditions of dosage 5 g/L, pH = 3.0, and Cr(VI) concentration 10 mg/L. The saturation magnetization (16.1emg/g) of MMABC still remained after adsorption. According to FTIR and Raman results, the benzene-ring adjacent carbonyl did not showed obvious positive effects on Cr(VI) removal. A potential mechanism and corresponding apparent kinetic model indicated the Cr(VI) removal process by MMABC followed adsorption-reduction-adsorption steps. Cr(VI) was firstly adsorbed on surface and subsequently reduced to Cr(III), which was further adsorbed on MMABC surface. Langmuir isotherm (with maximal adsorption capacity of 25.27 mg/g) and pseudo second-order kinetic model were suitable for adsorption step.

1. Introduction Hexavalent chromium (Cr(VI)) is a wastewater contaminant that mainly discharge from industrial procedures of chromium mining & smelting, metal processing, dyeing, manufactures of chromium salt, electroplating, and leather, etc. (Selvi et al., 2001). Cr(VI) is a high toxic, carcinogenic, mutagenic and teratogenic agent to living organisms (Selvi et al., 2001). For many decades, great efforts have been made to remove effluents’ Cr(VI) by means of adsorption, chemical precipitation, electrocoagulation, ion exchange, membrane separation, reduction and biosorption methods, etc. (Alvarado et al., 2013; Duan et al., 2017b; Owlad et al., 2009). Adsorption is considered as a simple, high-efficient (Demirbas, 2008), and low-cost (Mohan et al., 2014) method benefits from various adsorbents that can be derived from solid waste with “dealing with waste by waste” significances. Biochar is an activated-carbon like adsorbent (Chen et al., 2015a; Liu et al., 2015) commonly derived (or by-produced) from biomass pyrolysis that has poly-generated potentials on energy recoveries (biofuels) (Chen et al., 2014; Chen et al., 2016). Different kinds of biomass were pyrolyzed to biochar, which had been demonstrated appropriate to adsorb heavy metal ion and organic pollutant (Inyang et al., 2016; Mohan et al., 2014; Qiu et al., 2009). The functionalization (or

⁎

modification) of biochar is attractive to enhance adsorption capacities of contaminants. Many functionalization ways such as surface decoration (Liu et al., 2015; Yan et al., 2015), inorganics compound (Zhang et al., 2013a; Zhang et al., 2013b), molecules grafting (Shi et al., 2017), magnetization (Chen et al., 2011; Han et al., 2016; Wang et al., 2015a; Wang et al., 2015b; Yap et al., 2017) etc. had been reported. Among these functionalization ways, magnetization is fascinating since corresponding products (magnetic biochar, MBC) show strong paramagnetism and high saturation properties, which give advantages on high efficient solid-liquid separations (avoiding slurry production during water treatment plants running), as well as contaminant adsorb capacities improvements (Essandoh et al., 2017; Wang et al., 2015b; Zhang et al., 2013a). Many biochar derived from different biomass had been used for aqueous Cr(VI) adsorptions (Chen et al., 2015a; Deveci & Kar, 2013; Han et al., 2016; Zhou et al., 2016). It was believed that Cr(VI) adsorption was influenced by effluent’s pH value, which dominated the rate control steps of adsorption, Cr(VI) ionization, and Cr(VI) reduction to trivalent Cr(III) (Chen et al., 2015a; Zhou et al., 2016). High pH value reduced Cr(VI) adsorption efficiency since the abundant OH− led to Cr(OH)3 generation and hindered the chromium ions to transfer and interact with surface functional groups (Chen et al., 2015a). Inversely,

Corresponding author. E-mail address: [email protected] (Z. Chen).

https://doi.org/10.1016/j.biortech.2018.01.145 Received 17 December 2017; Received in revised form 27 January 2018; Accepted 31 January 2018 0960-8524/ © 2018 Elsevier Ltd. All rights reserved.

Bioresource Technology 256 (2018) 1–10

X. Zhang et al.

2.2. Batch adsorption experiment

lower pH value was preferred attributed to an increasing in protonation during Cr(VI) adsorption (Chen et al., 2015a; Deveci & Kar, 2013; Zhou et al., 2016). Cr(VI) adsorption capacity increased with effluent temperature since the favorable endothermic (thermodynamic) reactions (Inyang et al., 2016; Yang et al., 2017). Moreover, biochar surface is rich in active functional groups such as hydroxyl, carbonyl, phenol, quinone, and aliphatic, etc (Jindo et al., 2014). The distribution of these functional groups is highly depended on pyrolytic temperature, in addition to used biomass types. It was pointed out by Zhou et al. (Zhou et al., 2016) that surface carboxyl and hydroxyl groups were in favor of Cr(VI) adsorption. The surface richer carboxyl and hydroxyl groups were obtained from lower pyrolysis temperature (300 °C) rather than higher temperature (450 °C and 600 °C) (Zhou et al., 2016). Since the Cr (III) was retained in solution, Zhou et al. (Zhou et al., 2016) indicated that the mechanism of Cr(VI) removal included adsorption and reduction stages: (1) Cr(VI) ions were adsorbed to biochar surface via electrostatic attractions; (2) the adsorbed Cr(VI) was reduced to Cr(III) and then discharged into the solution. The MBC have been prepared and utilized for Cr(VI) removal by several studies. Yang et al. (Yang et al., 2017) prepared a pinewood derived MBC under 700 °C pyrolytic temperature, showing a maximal adsorption capacity of 52.1 mg/g at 40 °C. The MBC derived from eucalyptus leaf (pyrolysis under 400 °C) showed a 97.11% efficiency for Cr(VI) removal from electroplating wastewater (Wang et al., 2014). The post-adsorption MBC still retained its original magnetic separation performance (Wang et al., 2014). A MBC was prepared by Duan S et al. (Duan et al., 2017a) using co-pyrolysis of iron sludge and cotton stalk blends under 500 °C. After magnetization, the MBC showed excellent adsorption capacity (67.44 mg/g) compared with primitive biochar (40.91 mg/g) (Duan et al., 2017a). However, the potential mechanism of Cr(VI) removal using MBC, even the related adsorption-reduction process via kinetics have not been reported. Melia azedarach (MA) tree is an important source for paper pulp, logistics packing-case, furniture production since its fast-growing capacity. Annually, large amount of MA residues (wood, sawdust, bark, branch) is generated (such as in Guangxi Province, China), which is a potential feedstock for MBC preparation with environmental significances. Our study aims to use MA wood to prepare a MBC adsorbent for Cr(VI) removal via pyrolysis. The characterizations, Cr(VI) removal capacity, potential adsorption mechanism, kinetics were investigated.

Desirable concentration (5mg/L, 10 mg/L, 30 mg/L, 50 mg/L, 100 mg/L, 150 mg/L, 200 mg/L) of Cr(VI) solutions were prepared by diluting the stock solution (1000 mg/L), which was obtained by dissolving 2.8287 g K2Cr2O7 (analytical reagent) in 1000 ml deionized water. Certain mass of MMABC was added to 100 ml Cr(VI) solution and shaken mechanically at 120 rpm under room temperature (25 °C). The solid-liquid separations at different time were realized using injected Cellulose-Acetate membrane filter (< 0.45 μm). The Cr(VI) and Cr(III) concentrations were measured according to PRC National Standard (GB/T 7467-1987) using visible light spectrophotometry (V5800H, METASH) at λ = 540 nm after diphenylcarbazide (C13H14N4O) colorations. The removal efficiency (η, %) and adsorbing capacity (qt, mg/g) was respectively calculated by:

C0−Ct × 100 C0

(1)

η ·C0·V (C0−Ct )·V = m 100m

(2)

η (%) =

qt =

where C0 and Ct denoted initial and current concentration of Cr(VI), mg/L, respectively; m, the mass of adsorbent, g; V, the solution volume, L. The separated MMABC after adsorption were further air-cyclic dried under 100 °C for 12 h aimed to perform further characterizations.

2.3. Sample characterizations The proximate and ultimate analysis of MA was carried out via PRC National Standard (GB/T 212-2008), and CHNS/O analyzer (Vario Micro cube, Elementar), respectively. Inorganic elements contained in ash were detected using X-ray fluorescence (XRF) microprobe (EAGLE III, EDAX Inc.). Thermo-analysis (TG/DSC) were experimented for MA, MA120, Fe (NO3)3 9H2O crystal, MAFeN007 to contrastively study the weight loss behaviors during pyrolysis by using a TG/DSC analyzer (STA449C, NETZSCH). 5–10 mg samples was contained in an α-alumina crucible (another empty one was used for tare) and heated (10 °C/min) from room temperature to 700 °C under anoxic pyrolytic atmosphere maintained by nitrogen (100 ml/min flow). BET specific surface areas (SBET) was determined by nitrogen adsorption technique (−196 °C) in a Surface Area and Porosimetry system (BK112-T, JWGB). The functional groups were analyzed using FTIR (NEXUS 470) and Laser Confocal Raman Microspectroscopy system (RTS-EX, TEO) in wavenumber and Raman Shift range of 400–4000 cm−1 and 50–3000 cm−1, respectively. Chemical valences and semi-quantitative of C1s, O1s, Fe2p, Cr2p were analyzed by X-ray Photoelectron Spectroscopy (XPS) system (ESCALAB250Xi, Thermo Fisher Scientific) referencing binding energy of C1s at 285.0 eV. X-ray Diffraction (XRD) patterns were analyzed by using X’Pert PRO diffractometer (PANalytical B.V.) by scanning 2θ from 10° to 80° with a 2°/min rate. Magnetic hysteresis loops of materials (1–10 mg) were measured using Multi-functional Magnetic Measurement system (VersaLab, Quantum Design). pH dependent point zero charge (pHPZC) was measured according to pH shift method (Gatabi et al., 2016): 10 ml of 0.01 M NaCl solutions (NS) were firstly adjusted to desirable pH value (pHinitial, ranged from 3 to 10 in step-size of 1) by 0.1 M NaOH and 0.1 M HCl; then, 30 mg of MMABC were added to NS and shaken mechanically for 48 h at 250 rpm (25 °C). The final pH (pHfinal) was measured and differential to pHinitial obtained the ΔpH, which were further plotted against with pHinitial. The pHPZC value was obtained at ΔpH = 0 using cubic spline interpolation method.

2. Materials and methods 2.1. Magnetic biochar preparation Air-dried Melia azedarach (MA) chips was obtained from Guiping City, Guangxi province (China). Wood chips was further crushed and sieved into 0.177–0.25 mm (i.e. 80–60 mesh) size. Fe(NO3)3 9H2O (analytical reagent) was dissolved in 300 ml deionized water and impregnated with MA via Fe/C (mol/mol) = 0.07 ratio at 120 °C (oil-bath with Silicone oil) for 4 h. The MA feedstock after impregnation was subsequently dried in air- cyclic oven at 100 °C for 24 h to obtain magnetic biochar precursor (MAFeN007). The raw MA, and MA without Fe(NO3)3 9H2O impregnation but hot water treated (120 °C) for 4 h (denoted as MA120), were used for parallel experiments. The conditions (400 °C, 1h) for magnetic MA biochar (MMABC) preparation was referenced to Wang et al. (Wang et al., 2014). 10 g precursor was firstly contained in a quartz funnel and inserted in a stainless steel tube, followed by down-stream nitrogen (99.99% purity with 100 ml/min flow) aerification for 10 min to eliminate intra-tubal air. Afterwards, precursor was heated from room temperature to 400 °C in rate of 10 °C/min and maintained for 1 h. Once the intra-tubal temperature balanced to room temperature, MMABC was extracted and stored in vacuum (0.01 atm). Biochar derived from MA and MA120 (signed as MABC and MABC120, respectively) were obtained in a same programmed temperature. 2

Bioresource Technology 256 (2018) 1–10

X. Zhang et al.

equilibrium isotherm, which combines the Herry’s low, Langmuir, and Freundlich models by incorporating unknown parameters. In RP model, qe v.s. Ce is described as:

2.4. Adsorption kinetics Kinetic of Cr(VI) adsorption was studied by the following three common models that had been investigated widely for metal ions adsorption.

qt = qe ·exp(1−k1·t )…pseudo-first-order…

qt =

qe2 ·k2·t 1 + qe ·k2·t

qe =

(3)

…pseudo-second-order…

2.6. Calculation details

(5)

where, t, the adsorption time, min; k1, the first-order rate constant, min−1; k2, the second-order rate constant, g/(mg min); qe, the absorption capacity, mg/g; kid, the intra-particle diffusion rate constant, g/(mg min0.5); bL, a constant relates to boundary layer thickness, mg/g.

Nonlinear least square was used to estimate the unknown parameters in kinetic and adsorption isotherm equations by minimizing variance (s2) that related to residual sum of squares (RSS) and statistical F-test (for RSS significant testing). The Genetic Algorithm (ga global optimization tool in MATLAB®) was used without initial values guess, using a 3000 PopulationSize and 20,000 Generations. Once difference value of objective function between penultimate and last iteration was smaller than 1e-8 (TolFun = 1e-8), the optimization was considered stable and stopped. The objective functions for optimization calculations were given as:

2.5. Adsorption isotherms Adsorption isotherm can evaluate the adsorption capacity and understand corresponding physiochemical essence being in practical significances. The Langmuir isotherm model is a monolayer adsorption model, in which active sites have a uniform energy. Equilibrium absorption capacity according to Langmuir isotherm is described by:

qe =

qm ·K ·Ce 1 + KL·Ce

⎧ s2 ⎪ 2 ⎪ = ∑Nd [qt,Exp,i − qt,Cal,i ] i=1 ⎪ Nd − Np O. F1 = min ⎨ for pseudo first, second order kinetic ⎪ 2 ⎪ 2 NC [qe,Exp,j − qe,Cal,j] for adsorption isotherms ⎪ s = ∑ j=1 NC − Np ⎩

(6)

where qm, the maximal adsorption capacity, mg/g; Ce, the Cr(VI) concentration at absorption equilibrium, mg/L; KL, the adsorption rate dependent constants, L/mg. Freundlich model considers the surface energy of adsorbent is in heterogeneous. Such heterogeneous surface energy makes the stronger binding sites are occupied firstly and then leads to binding strength decreases as the occupied sites increasing. Freundlich adsorption isotherm is expressed by:

qe = KF ·Ce1/ n

(10) where i denoted i-th adsorption time point with total of Nd; j denoted jth adsorption under different initial concentrations with total of NC; Np denoted the number of unknown parameters in model; subscript Exp and Cal represented experimental and calculated data, respectively. The lower the variance (s2), the better fitting goodness of models to experimental data.

(7)

where n, the surface heterogeneity factor, n > 1; KF, the adsorption capacity of the adsorbent, L1/n mg(1−1/n)/g. Temkin model states that the adsorptive heat for all molecules linearly decrease with coverage degree. The decrease of heat attributes to adsorbent-adsorbate interactions, which is characterized by a factor of binding energy uniform distribution. The Temkin model is expressed as:

qe =

RT RT ·lnαT + Ce βT βT

(9)

where KRP (L/g) and aRP are constants; θ is a constant that range from 0 to 1, whose borders relates to Herry’s low (θ = 0) and Langmuir model (θ = 1), respectively.

(4)

qt = kid· t + bL…Intra-particle diffusion…

KRP ·Ce 1 + aRP ·Ceθ

3. Result and discussion 3.1. Physi-chemical characteristics of MA feedstock Table 1 indicated that the carbon content (48.3 wt%) of Melia azedarach (MA) wood was lower than coal (62.9–86.9 wt%)(Vassilev et al., 2015). Little H, N, S was observed that located in the common range H∈(3.2,10.2), N∈(0.1,12.2), and S∈(0.01,1.69) in biomass(Vassilev et al., 2015), respectively. Oxygen content (43.8 wt%) was quite higher than coal (4.4–29.9 wt%) and led to a higher volatiles content (76.7 wt %) and lower fixed carbon content (13.4 wt%) compared with coal (volatiles 12.2–44.5 wt%, and fixed carbon 43.9 wt%)(Vassilev et al., 2015). The ash content also was in 0.1–34.3 wt% range of common

(8)

where R, the universal gas constant, 8.314 J/(mol K); T, the adsorption temperature, K; αT, the equilibrium binding constant decided by maximum binding energies, L/g; βT, the Temkin isotherm constant. Adsorption heat (J/mol) can be calculated as Qa = R T/βT . An empirical equation is expressed by Redlich-Peterson (RP) Table 1 Ultimate analysis, proximate analysis, and ash’s XRF result of Melia azedarach wood. Ultimate analysis (wt%)b

Calorific value (MJ/kg)a

Proximate analysis (wt%)a

C

H

N

S

Oc

HHV

Moisture (M)

Volatiles (V)

Ash (A)

Fixed Carbon (FC)d

48.3 ± 0.21

5.5 ± 0.1

0.21 ± 0.08

0.1 ± 0.02

43.8

19.02 ± 0.82

7.8 ± 0.89

76.7 ± 1.32

2.1 ± 0.14

13.4

P[O] 4.05

K[O] 43.40

Sc[O] 0.26

Mn[O] 0.27

Fe[O] 1.44

Zn[O] 0.24

S[O] 5.71

XRF result(wt%) Mg[O] Si[O] 4.29 25.26 a b c d

Air-dried base. Dry ash-free basis (daf). O ≈ 100-C-H-N-S-A. FC = 100-M-V-A.

3

Bioresource Technology 256 (2018) 1–10

X. Zhang et al.

Fig. 1. TG/DSC results of different biochar precursors and Ferric nitrate crystal: (a) Weight loss curve and conversion rate (dα/dT, °C−1) curves; (b) Reference heat flow versus temperature (°C) dependences.

biomass, in which Fe, Si, Mn, Mg, P etc. were often detected. K (43.40 wt%) and Si (25.26 wt%) were the two richest inorganic elements in MA feedstock, and were similar to the reported results (Vassilev et al., 2015).

α=

w0−wT w0−w∞

(11)

Here, w0, wT and w∞ referred to normalized (%) initial (30 °C), current, and final (700 °C) weight of samples, respectively. The conversion rates (first-order derivative of α v.s. temperature, dα/dT, °C−1) were consecutively obtained by numerical differential and three-order SavitzkyGolay smoothing, which were given in Fig. 1(a). Most weight loss of Fe (NO3)3 9H2O occurred before 400 °C, in which sharp peaks of dα/dT v.s. T profile appeared at ∼123 °C and ∼157 °C, respectively corresponded to the formation of FeOOH and γ-Fe2O3 (Wieczorek-Ciurowa and Kozak, 1999). After 400 °C, the formative γ-Fe2O3 transferred to α-

3.2. Pyrolytic behaviors Fig. 1 showed the weight loss of MA, MA120, Fe(NO3)3 9H2O (crystal), and MAFeN007 samples during TG/DSC experiments. To comparatively investigate the reaction rates, sample weights were normalized to conversion (α, α∈[0,1]) as: 4

Bioresource Technology 256 (2018) 1–10

X. Zhang et al.

Fe2O3 via exothermic reaction as demonstrated by endothermic peak G in Fig. 1(b1) (Wieczorek-Ciurowa and Kozak, 1999). The weight v.s. T and dα/dT v.s. T profiles of MA were similar to MA120, indicating 120 °C water-pretreatment almost did not affect the weight loss during pyrolysis. Endothermic peaks A and B appeared on DSC profile (for MA and MA120) was assigned to cellulose degradation, corresponded to the peaks on dα/dT v.s. T curves (Chen et al., 2015b). An endothermic peak C of MA120 in Fig. 1(b2) ranged from ∼500 to 700 °C appeared earlier than MA, whose partial endothermic peak began at ∼630 °C. The dα/ dT v.s. T profiles of MA and MA120 showed very slow rate after 400 °C, which was attributed to lignin component pyrolysis to generate the oxygen-complexes (Chen et al., 2015b). Regarding the precursor MAFeN007, its dα/dT v.s. T profile depicted more complex than the other samples, characterized with several peaks and shoulders (Fig. 1(a)). The maximum conversion rate of MAFeN007 happened at ∼320 °C, which was neither caused by Fe (NO3)3 9H2O decomposition, nor the cellulose pyrolysis that matched with endothermic peaks K (Fig. 1(b4)), A (Fig. 1(b1)), and B (Fig. 1(b2)). The maximum conversion rate was responsible for endothermic reactions ranged from ∼200 to 340 °C(peak J). Additionally, peaks D, E, F, G for Fe(NO3)3 9H2O decomposition disappeared or shift with respected to MAFeN007. Endothermic shoulder I for MAFeN007 was possibly overlapped by peak F. And peak G for α-Fe2O3 formation did not reflected in MAFeN007. The pyrolysis of precursor in ∼500–660 °C range showed ∼10 wt% weight loss corresponded to endothermic peak L, which was responsible for lignin decomposition similar to MA120. 3.3. Adsorption optimization and adsorption isotherm 3.3.1. Effect of contact time and initial concentration It was widely considered that lower pH value was favorable to Cr (VI) adsorption (Chen et al., 2015a; Selvi et al., 2001). The effects of contact time and Cr(VI) initial concentration on Cr(VI) removal were studied under acidic condition pH = 3.0. Fig. 2(a) showed the residual Cr(VI) concentration (Rc = Ct/C0) as a function of contact time (t, min) experimented from different initial Cr(VI) concentrations by using the same MMABC dosage (0.5 g/100 ml). Rc almost remained unchanged after 540 min for different concentration levels. Thus, 540 min could be considered as the optimum contact time that the adsorptions reached equilibrium state. Fig. 2(a) also suggested the Cr(VI) removal efficiency decreased with the initial concentration increase. However, highest removal efficiency was obtained for concentration of 10 mg/l, which was even higher than lower concentration of 5 mg/l. Reason possibly lied in ions’ diffusion resistance, in which Cr(VI) ions took more time to reach MMABC surface under dilute concentration level. The decrease in removal efficiency v.s. initial concentration was attributed to the limited active sites on specific MMABC, i.e. increase adsorbent dosage could improve active sites for Cr(VI) adsorption. 3.3.2. Adsorption isotherm The removal efficiency (η) related to active sites (initial concentration dependent) could be characterized by adsorption isotherm once the system achieved adsorption equilibrium state with equilibrium adsorption (qe, mg/g) and concentration (Ce, mg/L). Fig. 2(b) compared the fittings of experiments to different isotherm models, whose adsorption isotherm parameters were summarized in Table 2. It could be concluded that the Langmuir model gave best explanation to Cr(VI) adsorption, with a highest correlation coefficient (r2 = 0.9909) and lowest variance (s2) value. This illustrated the Cr(VI) adsorbed on MMABC followed monolayer adsorption mechanism with maximal

Fig. 2. Cr(VI) adsorption results for: (a) Removal efficiencies versus time at different initial concentration levels; (b) Comparisons between different calculated adsorption isotherms with experimental equilibrium absorption capacity; (c) Effect of MMABC dosage on Cr(VI) removal at pH = 3.0, initial concentration 10 mg/L, adsorption time 540 min.

5

Bioresource Technology 256 (2018) 1–10

X. Zhang et al.

illustrated the Cr(VI) adsorption on MMABC was followed pseudo second-order model, which were also reported by other magnetic adsorbents (Sun et al., 2014; Yang et al., 2017). The adsorption capacity (qe, Table 3) of MMABC was higher than MABC and MABC120, indicating the magnetization could enhanced Cr(VI) adsorption. The calculated equilibrium absorption capacities (qe) as well as maximal adsorption capacities (qm) of MMABC were close to experimental value according Langmuir isotherm. Regarding the multiple-steps kinetic, intra-particle diffusion model (Eq. (5)) considers the decrease process of concentration consecutively contains three steps: (1) molecules transmits from aqueous phase (boundary layer) to solid surface (includes instantaneous adsorption on adsorbent surface); (2) pore diffusion which is slow and considered as rate control step; (3) adsorbate adsorbes on surface active sites. By plotting ln(qt) v.s. t1/2 of Eq. (5), the steps could be detailed and corresponding rate constants could be estimated by linear regression. The linear regressions of intra-particle diffusion model were showed in Fig. 3(d). As could be seen in Fig. 3(d) and Table 3, the three stages were detailed well following a rate constant order of: kid,1(1st stage) > kid,2(2nd stage) > kid,3(3rd stage), which illustrated a rate decrease occurred in Cr(VI) adsorption process. The first stage represented external diffusion and instantaneous adsorption with highest rate constants. The 2nd and 3rd stages were attributed to intraparticle diffusion with lower rate constant. Especially, rate constant of 3rd stage was quite lower than the previous two stages, which was assigned to adsorption equilibrium and mesoporous/micropore diffusion. On the other hand, MMABC showed preferable adsorption on Cr(VI) adsorption than MABC and MABC120, evidenced by the higher rate constants of 1st and 2nd stages. Furthermore, kid,3 value of MMABC was lower, suggesting the Cr(VI) adsorption on MMABC reached equilibrium state earlier and further evidenced MMABC were more preferred than the two biochar without magnetizations.

Table 2 Calculated results of adsorption isotherms. Models

Parameters

Langmuir

qm 25.27

KL 0.047

R-P

αR 0.699

βR 0.684

Freundlich

nF 2.47 BT 1209.87

KF 3.38 AT 34.55

Temkin

KR 3.731

s2

RSS

R_square

1.1774

5.8870

0.9909

3.1693

12.6773

0.9704

2.75401

13.7701

0.9639

17.0933

85.4666

0.7742

adsorption capacity (qm) of 25.27 mg/g. According to Langmuir model, a separation parameter (dimensionless) RL (Eq. (12)) could be used to describe the adsorption characteristics: unfavorable (RL ≥ 1), favorable (0 < RL < 1), and irreversible (RL = 0). The RL values of different C0 levels ranged from 0.811 to 0.097, which further illustrated Cr(VI) adsorption was Langmuir isotherm favorable.

RL = (1 + KL·C0)−1

(12)

3.3.3. Effects of MMABC dosage Adsorbent dosage is an important parameter relates to the economical efficiency of adsorbent and even adsorbent-adsorbate equilibrium, since it affects adsorb capacity. The effect of MMABC dosage on Cr(VI) adsorption was investigated under conditions of: pH = 3, concentration 10 mg/l, contact time 540 min. Fig. 2(c) showed the η increased with MMABC. The effect of MMABC dosage on Cr(VI) adsorption mainly depended on surface active sites (Chen et al., 2015a; Yang et al., 2017): in case of a certain unit mass adsorbent, adsorption activity sites number was proportional to mass. However, after dosage of 0.5 g/100 ml, removal efficiency remained constant of 99.8% and indicated 5 g/L was optimal dosage.

3.5. Characterizations and potential mechanisms 3.5.1. Characterizations The nitrogen adsorption analysis result of MMABC was given in Table 4, indicating the MMABC was a mesoporous material (rp = 12.698 nm) with specific surface area (SBET) 5.219 m2/g. Magnetic loading curves nearly coincided to hysteresis loops, demonstrating a nearly superpara-magnetism performance of MMABCs. After adsorption, MMABC could be separated (60 s) from solution with a calculated saturation magnetization 16.1 emg/g, which was closed to the pristine one (17.3 emu/g). The reserved saturation magnetization was responsible for the contained ferric oxide (Fe3O4) evidenced by XRD patterns, in which the broad diffraction peak for amorphous carbon (MABC120) was weaken. FTIR results of pristine and post-adsorption MMABCs showed a infrared spectrum absorption at 578 cm−1, which were accounted for Fe3O4 (Sun et al., 2012). At the same time, XPS result of pristine MMABC revealed two characteristic valences could be peaked at binding energy (B.E) of 711.4 eV (Fe2p3/2) and 725.3 eV (Fe2p1/2) that assigned to Fe3O4 (Fe2p) (Wang et al., 2013). After adsorption, the assigned B.E.(Fe3O4) was increased to 712.1 eV and 726.2 eV respectively attributed to the shielding effects. Thus, the ferric oxide being in XRD pattern was assigned to Fe3O4 (ICOD number: 00003-0863) that has superpara-magnetism performance. FTIR spectrum of pristine MMABC showed several characteristic peaks presented in wavenumber range of 800–1800 cm−1, in addition to assigned hydroxy functional group (OeH, 3440 cm−1) and Fe3O4 (578 cm−1). The functional groups in such range led to D-band and Gband overlapped in Raman spectra, which was respectively caused by graphitic structure (D-band) and defected or disordered of carbon structure (G-band) that common found in activated carbons(Liu et al., 2017; Rahman et al., 2016). The peak at 1578 cm−1 (FTIR) belonged to C]C in aromatic ring stretching coupled to highly conjugated, which was corresponded to G-band in Raman spectra; and strong peak at

3.3.4. Effect of pH value The effect of pH value on Cr(VI) removal was investigated under conditions of: dosage 5 g/L, initial concentration 10 mg/L, contact time 540 min. η v.s pH value dependency ranged from 1 to 13 was depicted in Fig. 3(a). It was obvious that Cr(VI) removal efficiencies were quite depended on initial pH value: the elevating of pH value decreased the η from 99.8% to 45.5%. The highest removal efficiency was found at pH = 3 and decreased sharply as pH value increase. The higher removal efficiency at lower pH value (pH = 1, 2) were similar and justified the Cr(VI) removal by MMABC was suitable for acid condition. Fig. 3(a) also showed the final pH value manifested a reverse sigmoid shape versus initial pH value: the final and initial at strong acidity (pH = ∼1) and alkaline (pH = ∼13) almost unchanged. This illustrated the H+ and OH− was respectively consumed in acid and base condition. However, the experimental pHpzc value was observed as 7.16 (Fig. 3(a)). In other words, it evidenced that MMABC surface was neutral maintained by nearly equivalent amount of acid and base functional groups. In general, the solutus pH value affects adsorption capacity via changing adsorbent’s surface charge, ionization degree, and adsorbate species distribution. The downtrend of η v.s. pH values dependencies were widely found in many studies using different adsorbents (Chen et al., 2015a; Selvi et al., 2001; Yang et al., 2017). 3.4. Kinetics and comparison to biochar The adsorption and single-step kinetics (i.e. pseudo first and secondorder model) of MMABC, MABC and MABC120 were compared in Fig. 3(c). From statistical results (Table 3), pseudo-second order model was more suitable for Cr(VI) adsorption on three adsorbents with lower variances (s2) and higher r2 than the pseudo first-order model. It 6

Bioresource Technology 256 (2018) 1–10

X. Zhang et al.

Fig. 3. (a) effect of initial pH value on Cr(VI) removal; (b) pHPZC determination curve; (c) pseudo second-order kinetic model; (d) intra-particle diffusion kinetic model.

1384 cm−1 was assigned to CeC contributed to D-band peak (Peng and Zou, 2017). The infrared absorption at 1440 cm−1 was caused by CeH bond deformation vibrations in plan (Kaul and Kampfrath, 1986). And 1252 cm−1 along with 1002 cm−1 were assigned to CeO stretching vibrations being in ether and carboxyl groups, respectively. It was interesting that the carbonyl group (C]O, 1703 cm−1) of MMABC was not found in FTIR spectra but sharply peaked at Raman spectra. Such carbonyl group was in weak in high Raman scattering rather than infrared absorption activity, since it adjacent substituted to benzene ring via CeC group (Lee et al., 2003). Meanwhile, this benzene ring adjacent C]O was not found in biochar MABC120. It should be pointed out that

Table 4 SBET, pore volume parameters (Vtol), N2 adsorption mean pore size (rp) of MMABC. Identification

Value

SBET (m2/g)a Vtol (cm3/g) rp (nm)

5.219 0.017 12.698

a Calculated from Brunauer-Emmett-Teller (BET) equation.

Table 3 Calculated results of kinetic models for MABC, MABC120 and MMABC. Kinetic model

Sample

qe/mg g−1

k1/min−1

r2

s2

Pseudo first-order

MMABC MABC MABC120

1.93 1.45 1.44 qe/mg g−1

0.109 0.070 0.096 k2/g mg−1 min−1

0.9605 0.9337 0.9544 r2

0.0158 0.0162 0.0103

Pseudo second-order

MMABC MABC MABC120

2.00 1.53 1.50

0.092 0.071 0.104

0.9957 0.9917 0.9955

0.0034 0.0040 0.0020

r2

kid,2/ g mg−1 min−0.5 0.064 0.044 0.044

r2

MMABC MABC MABC120

kid,1 g mg−1 min−0.5 0.43 0.28 0.31

Intra-particle diffusion

1 1 1

7

0.9934 0.9863 0.9869

kid,3 g mg−1 min−0.5 3.99E-4 2.63E-3 3.47E-3

r2 0.6831 0.8019 0.9114

Bioresource Technology 256 (2018) 1–10

X. Zhang et al.

spectrum of post-adsorption MMABC, the oxygen (O1s) contained in Cr (VI) and Cr(III) was found at 530.1 eV. (Paparazzo, 1987). The adsorbed chromium (Cr2p) on MMABC was clearly analyzed on MMABC surface in atomic level of 2.8 at.% next to Fe2p (3.43 at.%). The valences of Cr2p were separated at 578.0 eV(Cr2p1/2) and 587.6 eV (Cr2p3/2), which was assigned to Cr(III) and Cr(VI), respectively (Zhu et al., 2017). The atomic content ratio of Cr(III)/Cr(VI) was calculated by integral as AtCr(III)/ AtCr(VI) = 1.13, indicated the adsorbed Cr(III) content was higher than Cr(VI). And it illustrated the Cr(VI) was reduced as Cr(III) during adsorption process, and then the generated Cr (III) was adsorbed on MMABC surface again. The disappearance of ether group (1252 cm−1) as well as carboxyl (1002 cm−1) possibly were caused by Cr(VI) oxidizing under acidic condition, i.e. providing electronic (e−) for Cr(VI) reduction. Such oxidizing reactions led to the generations of hydroxy (OeH), carbonyl (C]O), alkenyl (C]C), and CeC (1384 cm−1) groups that still remained in FTIR and Raman spectrograms. Thus, the 816 cm−1 and 735 cm−1 infrared absorptions were possibly caused by Cr(VI) and Cr(III) adsorptions (foreign substitutions) on MMABC surface. k1

k2

k3

Cr(VI) → Cr∗VI → CrIII → Cr∗III

Fig. 4. Calculated result of apparent kinetic model corresponding to potential mechanism.

(R-1)

According to the adsorption-reduction-adsorption process of chromium removal, a potential mechanism was described in Eq. (R-1): the 1st step denoted Cr(VI) adsorbed on MMABC surface with rate constant (min−1) k1; in 2nd step, a part of adsorbed Cr(VI) was seen as intermediate Cr∗VI which was further reduced by surface electron e− into Cr (III) with rate constant k2; then, partial Cr(III) was adsorbed on MMABC surface in Cr∗III form at 3rd step with rate constant k3. By assuming the three steps was accorded to collision theory with different reaction orders n1, n2, n3, the apparent kinetic of Cr(VI) removal was combined with four ordinary differential equations (ODEs), as:

the Fe3O4 was hard to be found in Raman spectra because of the weak scattering in the full Raman spectrum. As reported by Muraliganth et al. (Muraliganth et al., 2009), Fe3O4 characterized three peaks at 331 cm−1, 524 cm−1, and 670 cm−1 which were weakened after carbon coated. However, the strongest one (Muraliganth et al., 2009) appeared at 670 cm−1 still could be found both for two MMABCs (Fig. 4(d)), further indicated the magnetization intensity was still remained after adsorption. The narrow spectrum of C1s and O1s matched the FTIR an Raman results well. C1s could be separated into three valences at 284.4 eV, 285.1 eV, and 287.9 eV, respectively. Peak at 284.4 eV was the strongest one was caused by sp2 hybridized orbital that could be assigned to graphite (Merelet al., 1998) or CeC bond (Smith et al., 2016). And the peaks at 285.1 eV and 287.9 eV was assigned to carboxylic group and aromatic rings (C]C), respectively (Merel et al., 1998; Rahman et al., 2016). Regarding the O1s, 529.9 eV, and 531.1 eV was assigned to Fe3O4 and OeC (531.1 eV), respectively (McIntyre and Zetaruk, 1977; Smith et al., 2016). The higher O1s B.E. (532.7 eV) could be assigned to oxygen in phenol, alcohol, aliphatic ether and ketone structures (Wu et al., 2016).

d[Cr

]

⎧ dtVI = −k1·[CrVI]n1 ⎪ d[Cr∗ ] VI ⎪ = k1·[CrVI]n1 −k2·[Cr∗VI]n2 dt ⎨ d[Cr III] = k2·[Cr∗VI]n2 −k3·[Cr∗III]n3 ⎪ dt ⎪ d[Cr ast III ] = k3·[CrIII]n3 ⎩ dt

(13)

To solved the ODEs, total chromium concentration [Crtot] of was normalized as:

[Crtot] = Ct ,tot / C0

(14)

And the expression of [Crtot] during removal process was given as:

[Crtot] = [CrVI] + [CrIII] 3.5.2. Potential mechanism study After adsorption, vC-O (1002 cm−1 and 1252 cm−1), CeH (1440 cm−1), and CeC (1384 cm−1) groups disappeared from FTIR after adsorption, indicating these groups were favorable for Cr(VI) removal. However, benzene ring adjacent carbonyl was still remained in Raman spectra, indicated such functional groups possible did not showed positive effect on Cr(VI) removal. Particularly, two new peaks were generated in fingerprint region of FTIR spectra at 816 cm−1 and 735 cm−1 but were not found in Raman spectra. It indicated the corresponding functional groups were in high infrared absorption but in low Raman scattering activities. The infrared absorption at 816 cm−1 and 735 cm−1 (Near Infrared region) could be assigned to CeH caused by foreign groups or molecules substitutions on C]C of benzene ring or alkenyl group (Duranoglu et al., 2012; Kweon et al., 2003). In XPS

(15) [CrVI∗] = [CrIII]

= [CrIII∗] =

0. Since At t = 0, [Crtot] = [CrVI] = 1, the Cr(VI) adsorption step included three evens being in intra-particle diffusion kinetic, constant k1 was obtained as:

k1 = kid,1 + kid,2 + kid,3

(16)

By further ignoring the intra-particle outward diffusion of Cr(III), kinetic parameters of Eq. (13) was estimated via Least Squares by minimizing objective function of: 2

min O. F2 =

Nd

∑∑

([Crtot]e,i (z )−[Crtot]c,i (z ))2 (17)

z=1 i=1

where subscript z = 1, 2 respectively denoted [CrVI] and [CrIII], respectively; e, and c referred to experiment and calculation, respectively.

Table 5 Calculated results of apparent kinetic model for potential mechanism. Parameters

kid,1 (min−1)

kid,2 (min−1)

kid,3 (min−1)

k2 (min−1)

k3 (min−1)

n1

n2

n3

Value

0.17

0.052

4.05E-4

1.71E-3

0.31

2.20

1.17

2.31

8

Bioresource Technology 256 (2018) 1–10

X. Zhang et al.

The optimization of Eq. (16) was coded in MATLAB® by using Genetic Algorithm toolbox (ga) fourth-order Runge-Kutta method (ode45 solver). Fig. 4 showed the calculated species fractions of chromium as a function of time (min). The calculated result gave agreement to experiments well with correlation coefficient (Pearson’s r) of 0.9965 (for [CrVI]) and 0.9676 (for [CrIII]). The fraction of Cr∗VI was sharply increased until ∼100 min, which was in accordance with instantaneous adsorption of Cr(VI). The rate constants of 1st step (Table 5) followed an order of kid,1 > kid,2 > kid,3 also gave agreement with intra-particle diffusion kinetic that the adsorption rate decreased as time extended. The calculated reaction order (n1 = 2.2) for adsorption step was also in accordance with pseudo second-order model (n = 2). Further more, the rate constant of 3rd step was higher than 2nd step, indicating Cr(III) adsorption of was faster than Cr∗VI reduction, which was the rate control step and led to a little detected Cr(III) decrease after ∼200 min.

bio-chars obtained during biomass pyrolysis. J. Ind. Eng. Chem. 19, 190–196. Duan, S.B., Ma, W., Pan, Y.Z., et al., 2017a. Synthesis of magnetic biochar from iron sludge for the enhancement of Cr (VI) removal from solution. J. Taiwan Inst. Chem. E 80, 835–841. Duan, W.Y., Chen, G.D., Chen, C.X., et al., 2017b. Electrochemical removal of hexavalent chromium using electrically conducting carbon nanotube/polymer composite ultrafiltration membranes. J. Membr. Sci 531, 160–171. Duranoglu, D., Kaya, I.G.B., Beker, U., et al., 2012. Synthesis and adsorption properties of polymeric and polymer-based hybrid adsorbent for hexavalent chromium removal. Chem. Eng. J. 181, 103–112. Lee, M., Lee, J.P., Rhee, H., et al., 2003. Applicability of laser-induced Raman microscopy for in situ monitoring of imine formation in a glass microfluidic chip. J. Raman Spectrosc. 34, 737–742. Paparazzo, E., 1987. XPS and auger spectroscopy studies on mixtures of the oxides SiO2, Al2O3, Fe2O3 and Cr2O3. J. Electron. Spectrosc. 34, 97–112. Essandoh, M., Wolgemuth, D., Pittman, C.U., et al., 2017. Adsorption of metribuzin from aqueous solution using magnetic and nonmagnetic sustainable low-cost biochar adsorbents. Environ. Sci. Pollut. R 24, 4577–4590. Gatabi, M.P., Moghaddam, H.M., Ghorbani, M., 2016. Point of zero charge of maghemite decorated multiwalled carbon nanotubes fabricated by chemical precipitation method. J. Mol. Liq. 216, 117–125. Han, Y., Cao, X., Ouyang, X., et al., 2016. Adsorption kinetics of magnetic biochar derived from peanut hull on removal of Cr (VI) from aqueous solution: effects of production conditions and particle size. Chemosphere 145, 336–341. Inyang, M.I., Gao, B., Yao, Y., et al., 2016. A review of biochar as a low-cost adsorbent for aqueous heavy metal removal. Crit. Rev. Environ. Sci. Technol. 46, 406–433. Jindo, K., Mizumoto, H., Sawada, Y., et al., 2014. Physical and chemical characterization of biochars derived from different agricultural residues. Biogeosciences 11, 6613–6621. Wieczorek-Ciurowa, K., Kozak, A.J., 1999. The thermal decomposition of Fe (NO3) 3⋅9H2O. J. Therm. Anal. Calorim. 58, 647–651. Zhu, K., Duan, Y., Wang, F., et al., 2017. Silane-modified halloysite/Fe3O4 nanocomposites: simultaneous removal of Cr (VI) and Sb (V) and positive effects of Cr (VI) on Sb (V) adsorption. Chem. Eng. J. 311, 236–246. Kweon, H., Yoo, M.K., Park, I.K., et al., 2003. A novel degradable polycaprolactone networks for tissue engineering. Biomaterials 24, 801–808. Wang, L., Huang, Y., Ding, X., et al., 2013. Supraparamagnetic quaternary nanocomposites of graphene@ Fe3O4@ SiO2@ SnO2: synthesis and enhanced electromagnetic absorption properties. Mater. Lett. 109, 146–150. Liu, W.J., Jiang, H., Yu, H.Q., 2015. Development of biochar-based functional materials: toward a sustainable platform carbon material. Chem. Rev. 115, 12251–12285. Liu, Y., Liu, X., Dong, W., Zhang, L., et al., 2017. Efficient adsorption of sulfamethazine onto modified activated carbon: a plausible adsorption mechanism. Sci. Rep. 7, 12437. Kaul, M., Kampfrath, G., 1986. Investigations on the doping of acetonitrile glow discharge polymer films: an inelastic electron tunnelling spectroscopy study. Thin Solid Films 137, 293–296. Mohan, D., Sarswat, A., Ok, Y.S., et al., 2014. Organic and inorganic contaminants removal from water with biochar, a renewable, low cost and sustainable adsorbent–a critical review. Bioresour. Technol. 160, 191–202. Muraliganth, T., Murugan, A.V., Manthiram, A., 2009. Facile synthesis of carbon-decorated single-crystalline Fe3O4 nanowires and their application as high performance anode in lithium ion batteries. Chem. Commun. 47, 7360–7362. Owlad, M., Aroua, M.K., Daud, W.A.W., et al., 2009. Removal of hexavalent chromiumcontaminated water and wastewater: a Review. Water Air Soil Pollut. 200, 59–77. Merel, P., Tabbal, M., Chaker, M., et al., 1998. Direct evaluation of the sp 3 content in diamond-like-carbon films by XPS. Appl. Surf. Sci. 136, 105–110. Qiu, Y., Zheng, Z., Zhou, Z., et al., 2009. Effectiveness and mechanisms of dye adsorption on a straw-based biochar. Bioresour. Technol. 100, 5348–5351. Rahman, M.Z., Ran, J.R., Tang, Y.H., et al., 2016. Surface activated carbon nitride nanosheets with optimized electro-optical properties for highly efficient photocatalytic hydrogen production. J. Mater. Chem. A 4, 2445–2452. McIntyre, N.S., Zetaruk, D.G., 1977. X-ray photoelectron spectroscopic studies of iron oxides. Anal. Chem. 49, 1521–1529. Selvi, K., Pattabhi, S., Kadirvelu, K., 2001. Removal of Cr(VI) from aqueous solution by adsorption onto activated carbon. Bioresour. Technol. 80, 87–89. Shi, Y., Zhang, T., Ren, H., et al., 2017. Polyethylene imine modified hydrochar adsorption for chromium (VI) and nickel (II) removal from aqueous solution. Bioresour. Technol. 247, 370–379. Smith, M., Scudiero, L., Espinal, J., et al., 2016. Improving the deconvolution and interpretation of XPS spectra from chars by ab initio calculations. Carbon 110, 155–171. Sun, X.T., Yang, L.R., Li, Q., et al., 2014. Amino-functionalized magnetic cellulose nanocomposite as adsorbent for removal of Cr(VI): synthesis and adsorption studies. Chem. Eng. J. 241, 175–183. Vassilev, S.V., Vassileva, C.G., Vassilev, V.S., 2015. Advantages and disadvantages of composition and properties of biomass in comparison with coal: an overview. Fuel 158, 330–350. Wang, S., Gao, B., Li, Y., et al., 2015a. Manganese oxide-modified biochars: preparation, characterization, and sorption of arsenate and lead. Bioresour. Technol. 181, 13–17. Wang, S., Gao, B., Zimmerman, A.R., et al., 2015b. Removal of arsenic by magnetic biochar prepared from pinewood and natural hematite. Bioresour. Technol. 175, 391–395. Wang, S.Y., Tang, Y.K., Li, K., et al., 2014. Combined performance of biochar sorption and magnetic separation processes for treatment of chromium-contained electroplating wastewater. Bioresour. Technol. 174, 67–73.

4. Conclusions Magnetic biochar derived from Melia azedarach wood (MMABC) was suitable for Cr(VI) removal. The removal capacity was enhanced by MMABC, in which benzene-ring adjacent carbonyl did not showed obvious positive effects on Cr(VI) removal. Cr(VI) removal followed threesteps process: Cr(VI) was adsorbed and subsequently reduced to Cr(III), which was further adsorbed on MMABC surface. Apparent kinetic model of three-steps process fitted experiment well. The Cr(VI) adsorption step was accounted for Langmuir isotherm (with maximal adsorption capacity of 25.27 mg/g) and pseudo second-order kinetic. The superparamagnetism (17.3 emu/g) was attributed to Fe3O4 that remained in MMABC after adsorption (16.1 emu/g). Acknowledgements Authors would like to acknowledge the financial supports from: Youth Science Foundation (No. 2016QK21), Scientific Research Startup Funding (No. qd16118) of Henan Normal University; National Natural Science Foundation of China (No. 51604099), Research Fund Program of Guangdong Provincial Key Laboratory of Environmental Pollution Control and Remediation Technology (2018K17); Basic Scientific and Technological Frontier Project of Henan Province (No. 132300410287). The corresponding author also appreciates Junming Tang (XRD), Dong Sun (XPS), and Dejun Chen (TG/DSC) for their characterizations in School of Chemistry and Chemical Engineering, Henan Normal University. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.biortech.2018.01.145. References Alvarado, L., Torres, I.R., Chen, A.C., 2013. Integration of ion exchange and electrodeionization as a new approach for the continuous treatment of hexavalent chromium wastewater. Sep. Purif. Technol. 105, 55–62. Chen, B., Chen, Z., Lv, S., 2011. A novel magnetic biochar efficiently sorbs organic pollutants and phosphate. Bioresour. Technol. 102, 716–723. Chen, T., Zhou, Z., Xu, S., et al., 2015a. Adsorption behavior comparison of trivalent and hexavalent chromium on biochar derived from municipal sludge. Bioresour. Technol. 190, 388–394. Chen, Y., Yang, H., Yang, Q., Hao, H., et al., 2014. Torrefaction of agriculture straws and its application on biomass pyrolysis poly-generation. Bioresour. Technol. 156, 70–77. Chen, Y.Q., Yang, H.P., Wang, X.H., et al., 2016. Biomass pyrolytic polygeneration system: adaptability for different feedstocks. Energy Fuels 30, 414–422. Chen, Z., Hu, M., Zhu, X., et al., 2015b. Characteristics and kinetic study on pyrolysis of five lignocellulosic biomass via thermogravimetric analysis. Bioresour. Technol. 192, 441–450. Demirbas, A., 2008. Heavy metal adsorption onto agro-based waste materials: a review. J. Hazard. Mater. 157, 220–229. Deveci, H., Kar, Y., 2013. Adsorption of hexavalent chromium from aqueous solutions by

9

Bioresource Technology 256 (2018) 1–10

X. Zhang et al.

Technol. 38, 1665–1674. Yap, M.W., Mubarak, N.M., Sahu, J.N., et al., 2017. Microwave induced synthesis of magnetic biochar from agricultural biomass for removal of lead and cadmium from wastewater. J. Ind. Eng. Chem. 45, 287–295. Zhang, M., Gao, B., Varnoosfaderani, S., et al., 2013a. Preparation and characterization of a novel magnetic biochar for arsenic removal. Bioresour. Technol. 130, 457–462. Zhang, M., Gao, B., Yao, Y., et al., 2013b. Phosphate removal ability of biochar/MgAlLDH ultra-fine composites prepared by liquid-phase deposition. Chemosphere 92, 1042–1047. Zhou, L., Liu, Y., Liu, S., et al., 2016. Investigation of the adsorption-reduction mechanisms of hexavalent chromium by ramie biochars of different pyrolytic temperatures. Bioresour. Technol. 218, 351–359.

Wu, T.T., Wang, G., Zhan, F., et al., 2016. Surface-treated carbon electrodes with modified potential of zero charge for capacitive deionization. Water Res. 93, 30–37. Peng, Y., Zou, Q., 2017. The Synthesis and Characterization of Tetrakis [(p-amino phenoxy) methyl] Methane. IOP Conference Series: Materials Science and Engineering. IOP Publishing, pp. 012–025. Sun, Y., Tian, Y., He, M., et al., 2012. Controlled Synthesis of Fe3O4/Ag Core-Shell Composite Nanoparticles with High Electrical Conductivity. J. Electron. Mater. 41, 519–523. Yan, L.L., Kong, L., Qu, Z., et al., 2015. Magnetic Biochar Decorated with ZnS Nanocrytals for Pb (II) Removal. ACS Sustain. Chem. Eng. 3, 125–132. Yang, P., Guo, D.B., Chen, Z.H., et al., 2017. Removal of Cr (VI) from aqueous solution using magnetic biochar synthesized by a single step method. J. Dispersion Sci.

10