Equilibrium and kinetic studies of ferricyanide adsorption from aqueous solution by activated red mud

Journal of Environmental Management 227 (2018) 277–285

Contents lists available at ScienceDirect

Journal of Environmental Management journal homepage: www.elsevier.com/locate/jenvman

Research article

Equilibrium and kinetic studies of ferricyanide adsorption from aqueous solution by activated red mud

T

Nazanin Deihimi, Mehdi Irannajad∗, Bahram Rezai Department of Mining & Metallurgical Eng., Amirkabir University of Technology, Tehran, Iran

A R T I C LE I N FO

A B S T R A C T

Keywords: Activated red mud Optimum conditions Ferricyanide Kinetic Equilibrium

In this study, activated red mud (ARM) was used as a new adsorbent for the removal of ferricyanide anions from aqueous solution. Based on the percentage of ferricyanide removal and ferricyanide adsorption capacity, optimum conditions were evaluated using the response surface method (RSM) and central composite design (CCD). In optimum conditions (pH = 5.6, adsorbent dosage of 2.59 g, ferricyanide concentration of 60 ppm and contact time of 60 min), the percentage of ferricyanide removal and ferricyanide adsorption capacity were obtained as 79.6% and 1.8 mg/g, respectively. The kinetics and equilibrium studies were evaluated by considering the effective parameters including pH and ferricyanide concentration. Kinetic studies were evaluated by kinetic models of pseudo first-order, pseudo-second-order (four different linearized forms), Elovich and intraparticle diffusion. The results of the kinetic study indicated that the mechanism of ferricyanide adsorption onto the ARM adsorbent is a chemisorption interaction by a fast ferricyanide adsorption onto ARM and subsequently the slow diffusion of ferricyanide ions into the ARM inner adsorption sites. The equilibrium studies showed that the adsorption process followed the Langmuir model in which ferricyanide adsorption onto ARM was homogeneous with monolayer adsorption. The results indicated that the activation process of red mud improved adsorbent efficiency and increased the adsorption capacity.

1. Introduction

human health and the environment. Thus, some standards have been legislated for the discharge of cyanide bearing wastewaters by environmental agencies in many countries (Mudder and Botz, 2004; Donato et al., 2007). In recent years, cyanide removal from groundwater and wastewater is carried out by different technologies such as electrodialysis, reverse osmosis, alkaline-chlorination-oxidation, electrowinning, flotation, solvent extraction, acidification–volatilization–reneutralisation, hydrolysis-distillation, hydrogen peroxide, caro's acid, iron cyanide precipitation, activated carbons, resin and adsorption methods with various minerals (Young and Jordan, 1995; Dash et al., 2009; Botz, 2001). Recently, different kinds of low-cost adsorbents were effectively used for removal ions such as soils, wastes and ore bearing minerals including ilmenite hematite bauxite (FeTiO3) , (Fe2 O3) , [AlO. OH / Al (OH )3] and pyrite (FeS2) (Dash et al., 2005; Young and Jordan, 1995). There are few studies about the iron-cyanide complex adsorptions onto soils and minerals. Noroozifar et al. (2009) used the modified natrolite zeolite–iron oxyhydroxide system as adsorbent for cyanide removal. They achieved a yield of 82% for removing cyanide at optimum conditions. The results indicated that the modified adsorbent can be effectively used to remove cyanide from industrial wastewaters

Wastewaters are produced from industries including petrochemical (nylon, fibers, resins, metal plating, coke-processing plants and gold/ silver extraction containing different amounts of both free and metalcomplexed cyanides (Shifrin et al., 1996). Cyanide is used as a solvent for extraction of gold and silver ores and a flotation reagent for base metal recovery such as copper, lead and zinc in the mining industry (Dzombak et al., 2005; Kuyucak and Akcil, 2013; Logsdon et al., 1999). Some cyanide complexes are in a low toxicity level such as: soluble prussian blue (KFe (III )[Fe (II )(CN )6]) , and insoluble prussian blue (Fe4 [Fe (CN )6]3 ) , ferricyanide [FeIII (CN )3 − 6] and ferrocyanide [FeII (CN ) 4 − 6]. They will decompose slowly in case of exposure to light and release of free cyanide. They are extremely toxic for humans, the environment and aquatic organisms (Meeussen et al., 1992). Overall, cyanide is divided into three categories including: free cyanide such as hydrogen cyanide, strong metal cyanide complexes such as iron cyanide and weak acid dissociable cyanide compounds such as nickel and copper cyanide (Nsimba, 2009). It is important to remove cyanide ions from wastewater before being discharged to the environment because of their harmful effect on

∗

Corresponding author. E-mail address: [email protected] (M. Irannajad).

https://doi.org/10.1016/j.jenvman.2018.08.089 Received 1 January 2018; Received in revised form 15 August 2018; Accepted 23 August 2018 0301-4797/ © 2018 Elsevier Ltd. All rights reserved.

Journal of Environmental Management 227 (2018) 277–285

N. Deihimi et al.

2.2. Adsorbent preparation

(Noroozifar et al., 2009). Hanela et al. (2015) evaluated the removal of iron–cyanide complexes by combined UV–ozone and modified zeolite treatment (Hanela et al., 2015). They showed that this method achieved a significant iron–cyanide complex removal (66%). Gebresemati et al. (2017) investigated the optimization of cyanide sorption from aqueous medium by coffee husk using the response surface methodology. They obtained 90.6% cyanide adsorption in optimal conditions (initial concentration of 10 mg/L, contact time of 1 h, adsorbent dose of 1 g and pH of 8) (Gebresemati et al., 2017). Red mud (RM) is an industrial waste that is produced in large amounts as a consequence of caustic digestion of bauxite during alumina production (Brunori et al., 2005; Schwarz and Lalík, 2012). It has high alkalinity (pH of 10–12) that leads to serious environmental problems (Rai et al., 2012; Liu et al., 2011). The treatment of red mud due to the harmful environmental impacts of red mud and its enormous volume is necessary. This problem can be efficiently solved by conversion of the harmful red mud to useful products such as adsorbents. The high concentrations of oxides of aluminum, iron, titanium, silica and hydroxides in red mud make it appropriate as a cheap adsorbent for the removal of various ions from aqueous solution such as: dyes, phosphate, fluoride, chromium, arsenate, copper, zinc and nickel (Bhatnagar et al., 2011; Rai et al., 2012; Wang et al., 2005; Sutar et al., 2014). The RM is a heterogeneous mixture of various minerals with different active sites. The monolayer coverage of adsorbate does not occur in the heterogeneous surface of the absorbent. In this regard, the activation method of washing with HCl and then precipitating with ammoniac was applied to change and modify the heterogeneous surface of RM to the homogeneous surface. This facilitates the cost-effective removal of ferricyanide ions from the solution. In this study, the possibility of utilizing activated red mud as an adsorbent for cyanide removal from the solution was investigated. Initially, the effect of some operating parameters on the percentage of ferricyanide removal and ferricyanide adsorption capacity was evaluated. Then, the adsorption data for activated red mud was studied with different isotherm and kinetic equations in different pH and ferricyanide concentrations.

At first, RM was grinded and then sieved through a 100 mesh screen to obtain a grain size of less than 0.149 mm. In order to decrease the pH of RM (pH is usually greater than 13 for raw red mud), it was suspended in a solution of seawater with the weight ratio of liquid to solid of 3/1. This led to the precipitation of hydroxide, carbonate or hydroxy carbonate. Until reaching the equilibrium pH of 8–8.5, the solution was stirred for 1 h and then it took more than 1 h to sediment the mixture. After RM filtration by using Whatman filter paper No.42, it was kept in the oven overnight (100 °C) to completely dry up. Before using any activation method to obtain a uniform size of powder, the RM powder was sieved through a 0.149 mm screen. The sodalite compounds in the RM can block the available adsorption sites for ferricyanide adsorption. The acid treatment increases the adsorption capacity by leaching out of the sodalite compounds. Recently, activation of acid after neutralizing red mud has been efficacious to improve the physicochemical characters of red mud. In this regard, to increase the capacity of RM adsorption, the activation method of washing with HCl and then precipitating with ammoniac was examined. In the preparation step, 5gr of red mud was refluxed in %20 wt HCl solution for 30 min. Then ammonia was added to complete precipitation until obtaining a pH range of 8–8.5. It is optimal to neutralize red mud until a pH of around 8–8.5 is obtained for a number of reasons including releasing chemically adsorbed Na, neutralizing alkaline buffer minerals and insolubility of toxic metals in this pH value. The BET results show that the BET of RM increased from 26.75 to 101.99 m2/g after the activation method. The absorbent special surface of ARM increased, creating high pores on its surface, and making it appropriate to adsorb more cyanide ions from the solution. 3. Results and discussion 3.1. Experimental design and optimization of parameters In this study, the effect of the main variables including pH, adsorbent dosage, ferricyanide concentration and time (as independent variables) on the percentage of ferricyanide removal and ferricyanide adsorption capacity (as dependent variables) were investigated. In this regard, the response surface methodology based on a central composite design (CCD) was used by Design expert 10 (DX10) statistical software. Nineteen experiments were designed and randomly carried out to evaluate a relation between dependent variables and independent variables by the linear, two factor interaction, quadratic and cubic equations. Then, the analysis of variance (ANOVA) was used to determine the effective variables and their interaction on the output variables and the fitted model was evaluated (Gebresemati et al., 2017; Hoseinian et al., 2015; Sadeghalvad et al., 2016). The input and output variable ranges are presented in Table 1. The ferricyanide removal percentage (R (%)) and ferricyanide adsorption capacity (qe (mg/g)) are expressed as Eqs. (1) and (2), respectively:

2. Experimental methods 2.1. Material and methods Red mud samples of Jajarm mine, in north Khorasan province of Iran, were applied as adsorbent base material. The XRF studies show that the main chemical components of red mud are oxides of aluminum, calcium, iron, titanium and silicon. The XRD results showed that the primary red mud samples mainly contain hematite, sodalite, rutile, calcite, kaolinite, anatase, katoite, bohemit, cancrinite, hydrotalcite and nushadir salt. In order to improve the reactivity of RM as adsorbent for removal of ions in the solution, activated red mud by ammonia was used (Deihimi et al., 2018). Potassium ferricyanide (K4Fe(CN)6· 3H2O) and ammoniac (NH3) were used in the study. The solution ionic strength was adjusted using KCl (1 M). Hydrogen chloride (HCl) and sodium hydroxide (NaOH) were used for adjusting the pH of the solution. All reagents were supplied from Merck. Experiments were carried out in the batch system in a 250 ml Erlenmeyer flask with a constant stirring rate of 160 rpm at a temperature of 298.15 K. The Erlenmeyer flask was covered by aluminum foil and sealed with rubber stoppers to prevent ferrocyanide decomposed by light. After equilibrium, the samples were centrifuged at 8000 rpm for 10 min and filtered to completely separate the liquid from solid phases. The ferricyanide percentage in the solution was determined by a UV–visible spectrophotometer at 420 nm (model HITACHI U-2000). Fourier transform infrared (FTIR) spectrometer at a spectral resolution of +4 cm−1 (Nicolet 6700) and scanning electron microscopy coupled with energy dispersive X-ray (SEM/EDX) spectroscopy (FEI Quanta 200 electron microscopy (Holland)) were used to determine the nature of interactions and sublate characterization, respectively.

R = ((Ci − Ce )/ Ci ) × 100

(1)

Table 1 The input and output variables range. Factor

Name

Low Actual

High Actual

Mean

Standard deviation

A B C

pH Adsorbent dosage (g) Ferricyanide concentration (ppm) Time (min) R (%) q (mg/g)

4.50 1.20 60

7.50 4.50 160

6 2.850 110

1.188 1.307 39.603

15 21.396 0.470

60 84.189 6.134

40.533 52.298 2.828

24.387 16.387 1.210

D Y1 Y2

278

Journal of Environmental Management 227 (2018) 277–285

N. Deihimi et al.

Table 2 The analysis of variance values for ferricyanide removal percentage and amount of ferricyanide adsorbed. Source

Sum of squares

df

Ferricyanide Model A B C D AB AC AD CD B2 D2 Residual Lack of Fit Pure Error Cor Total Ferricyanide Model A B C D AB AC AD BC BD CD D2 Residual Lack of Fit Pure Error Cor Total

removal percentage 0.014518226 10 0.000273429 1 0.003516443 1 0.00205632 1 0.000144091 1 0.000227197 1 0.000846474 1 9.72644E-05 1 0.001296966 1 0.001643382 1 9.54495E-05 1 0.000211578 8 0.000159852 6 5.17266E-05 2 0.014729804 18 adsorption capacity 27.40366 11 0.41011 1 5.765895 1 7.939996 1 0.325122 1 0.248769 1 3.934454 1 1.207995 1 1.768383 1 0.305521 1 4.964618 1 0.271174 1 0.408529 7 0.324098 5 0.084431 2 27.81219 18

Mean square

F value

Table 3 Validation of proposed models by various statistical parameters. Model

0.001451823 0.000273429 0.003516443 0.00205632 0.000144091 0.000227197 0.000846474 9.72644E-05 0.001296966 0.001643382 9.54495E-05 2.64473E-05 2.66419E-05 2.58633E-05

54.89499 10.33866 132.9605 77.75169 5.448245 8.590569 32.00611 3.677673 49.03968 62.13806 3.60905

< 0.0001 0.0123 < 0.0001 < 0.0001 0.0479 0.019 0.0005 0.0914 0.0001 < 0.0001 0.094

1.030104

0.5687

2.491242 0.41011 5.765895 7.939996 0.325122 0.248769 3.934454 1.207995 1.768383 0.305521 4.964618 0.271174 0.058361 0.06482 0.042215

42.68658 7.027101 98.79665 136.0491 5.570854 4.262571 67.41552 20.69859 30.30063 5.234995 85.06704 4.646473

< 0.0001 0.0329 < 0.0001 < 0.0001 0.0503 0.0778 < 0.0001 0.0026 0.0009 0.056 < 0.0001 0.0681

1.535444

0.4394

qe (mg / g ) = ((Ci − Ce )/ M ) × V

R (%) qe (mg/g)

(2)

1/ Sqrt (R) = (+2147.6 + 86.615A − 0191.79B − 9.846C − 8.503D − 48.419AD + 1.372AC − 1.802AD + 0.113179CD + 52.769B2 + 0.041D 2) × 10−4

Std. dev.

Mean

CV%

R-Squared

Pred. Rsquared

Adeq. precision

0.005143 0.24

0.14 2.83

3.55 8.54

0.9856 0.9853

0.9197 0.897

27.87 28.912

respectively. The adequate precision value for ferricyanide removal percentage and the ferricyanide adsorption capacity models were obtained as 27.87 and 28.912, respectively, in which the desirable value is more than 4. The results indicate a satisfactory adjustment of the inverse sqrt and power model to the experimental data for the ferricyanide removal percentage and ferricyanide adsorption capacity. The perturbation and interaction plots are shown in Figs. 1–3, respectively. Perturbation plots of ferricyanide removal percentage and ferricyanide adsorption capacity (Fig. 1) show the comparative and simultaneous effects of pH, adsorbent dosage, ferricyanide concentration and time on the ferricyanide removal percentage and ferricyanide adsorption capacity. As can be seen from Fig. 1 (a) and 1 (b), the ferricyanide removal percentage and ferricyanide adsorption capacity are very sensitive to all variables. The variables effecting the adsorbent dosage, ferricyanide concentration and time are higher than pH on the ferricyanide removal percentage and ferricyanide adsorption capacity. The adsorbent dosage (B) and time (D) have a positive effect, while the pH (A) and ferricyanide concentration (c) have a negative effect on the ferricyanide removal percentage (Fig. 1 (a)). This is while the ferricyanide concentration (c) and time (D) variables have a positive effect and pH and the adsorbent dosage have a negative effect on the ferricyanide adsorption capacity (Fig. 1 (b)). The simultaneous effect of pH and adsorbent dosage on the adsorption percentage is shown in Fig. 2 (a), in which pH and adsorbent dosage changed in the experimental ranges while the ferricyanide concentration and time were constant. Fig. 2 (a) shows that the effect of adsorbent dosage on the adsorption percentage was more than pH. The adsorption percentage was increased with an increase in adsorbent dosage and pH due to a more available surface area with more functional groups at a higher dosage of adsorbent. Fig. 2 (b) shows the simultaneous effect of pH and ferricyanide concentration. It shows that the adsorption percentage increased with an increase in pH and a decrease in ferricyanide concentration. The remaining cyanide concentration in the solution increased with an increase in ferricyanide concentration due to the certain capacity of functional groups on the surface of ARM. Although the percentage of removal value usually reaches an asymptotic value or decreases above a certain value, it leads to a decrease of the adsorption percentage. Simultaneous effects of pH-adsorbent dosage and pH-ferricyanide concentration on ferricyanide adsorption capacity were shown in Fig. 3 (a) and 3 (b), respectively. The same results were obtained with ferricyanide removal percentage. The ferricyanide adsorption capacity was increased with increasing the adsorbent dosage and decreased with increasing the pH and ferricyanide concentration. After evaluating the effect of the main factors in the process, process optimization was performed using the design expert software. According to the obtained models, the optimum conditions were predicted as follows: pH = 5.6, adsorbent dosage of 2.59 g, ferricyanide concentration of 60 ppm and contact time of 60 min, for 84.18% ferricyanide removal percentage and 1.95 mg/g ferricyanide adsorption onto ARM (with desirability function value of 0.72). The predicted optimum conditions were evaluated, in which ferricyanide removal percentage and ferricyanide adsorption reached 79.6% and 1.8 mg/g, respectively. The results showed good agreement with the value predicted by the model.

where Ci , Ce , V and M are the initial ferricyanide concentration (ppm), final ferricyanide concentrations (ppm), the volume of the solution (ml) and the mass of the dry adsorbent (g), respectively (Behnamfard and Salarirad, 2009). The results of ANOVA and the indicative parameters of the proposed model validities are presented in Table 2 (a) and 2 (b), respectively. Eqs. (3) and (4) show the proposed models of ferricyanide removal percentage and ferricyanide adsorption capacity. These models agreed to inverse the sqrt model and power model for ferricyanide removal percentage and ferricyanide adsorption capacity, respectively. In these models, A is pH, B is adsorbent dosage, C is ferricyanide concentration and D is time.

(3)

qe = −10.090 + 1.109A − 0.725B + 0.115C + 0.258D + 0.160AB − 0.009AC − 0.020AD − 0.006BC − 0.009BD − 0.0007CD − 0.0002D 2

Statistical parameters

p-value prob > F

(4)

The amounts of F-value and p-value (Table 2) for ferricyanide removal percentage and ferricyanide adsorption capacity models are obtained at 54.89499, < 0.0001 and 42.68658, < 0.0001, respectively. The results indicated that the models are significant based on a confidence level of 95%. Validation of the proposed models was also evaluated by various statistical parameters (R-squared, coefficient of variance and the adequate precision value). The results are presented in Table 3. The R-squared of ferricyanide removal percentage and ferricyanide adsorption capacity models were 0.9856 and 0.9853, 279

Journal of Environmental Management 227 (2018) 277–285

N. Deihimi et al.

Fig. 1. Perturbation plot for ferricyanide adsorption onto ARM (a) the percentage of ferricyanide removal (R (%)), (b) ferricyanide adsorption capacity (qe (mg/g)).

Fig. 2. The 3D surface response for ferricyanide removal percentage: (a) pH-adsorbent dosage, (b) and pH-ferricyanide concentration.

Fig. 3. The 3D surface response for ferricyanide adsorption capacity: (a) pH-adsorbent dosage, (b) and pH-ferricyanide concentration.

equilibrium in all adsorption experiments. The results show that the cyanide sorption capacity by ARM increases with an increase in initial cyanide concentration. Fig. 4 also shows that the equilibrium time increases with an increase in the initial cyanide concentration. More readily available sorption sites are available to adsorb the cyanide ions at low initial cyanide concentration, while the cyanide ions diffuse to the ARM inner sites at higher cyanide concentrations that lead to an increase of equilibrium time. The increase of the initial ferricyanide concentration increased the ferricyanide adsorption capacity and the remaining ferricyanide concentration in the solution. Evaluation of adsorption kinetics of ferricyanide onto ARM is evaluated by pseudofirst order, pseudo-second order, Elovich and intraparticle diffusion kinetic models (Table 4) to study the mechanism and rate controlling step of adsorption. The four different types of linear expression of the pseudo-second order kinetic model were used as shown in Table 4. The kinetic models parameters were determined using the linear regression method of least squares. In order to determine the reaction mechanism (chemical or physical), the kinetic models of the pseudo-first and

3.2. Kinetic studies In order to determine the adsorption kinetics, sampling was performed from 2 to 120 min. The cyanide adsorption values at different initial cyanide concentrations and pH as a function of equilibrium time were shown in Fig. 4. The adsorption capacity increases with increasing time and then remains unchanged in a constant value at a certain time. It can be observed from Fig. 4 that the adsorption kinetics of ferricyanide onto ARM includes two steps of a fast initial adsorption and subsequently becomes slow and gradual. A considerable portion of the ferricyanide adsorption is performed in the initial rapid phase (first 1.59 min for a pH of 7.5 and first 10 min for a pH of 4.5), which can be ascribed to the ferricyanide ion rapid diffusion from the solution to the ARM external surface (Sadeghalvad et al., 2016). The adsorption process reached equilibrium at a pH of 7.5 faster than that at a pH of 4.5. The adsorption approximately reached equilibrium in 15 min, however, a 120 min period for a pH of 7.5 and a 180 min period for a pH of 4.5 were considered according to the initial experiments to ensure complete 280

Journal of Environmental Management 227 (2018) 277–285

N. Deihimi et al.

Fig. 4. The ferricyanide adsorption capacity (qe) as a function of time at different initial ferricyanide concentrations (a) pH = 7.5 and (b) pH = 4.5.

the Pseudo-second order Type (I) model that are in a good agreement. The ferricyanide concentration and the active sites number of the ARM surface have a main effect on the mechanism and rate of the adsorption process. The increase in the initial ferricyanide concentration increases the value of qt. The k2p was decreased by decreasing the pH, while the qt increased. The ferricyanide adsorption capacity onto the ARM increased as the pH decreased being equal to 8.56 mg/g at a pH of 4.5 and 2.88 mg/g at a pH of 7.5. This may be due to performing two different mechanisms during the adsorption process in these pH that gave rise to a chemical adsorption at a pH of 4.5 on the ARM phases, and to the formation of metal ferricyanide precipitates at a pH of 7.5 (Castaldi et al., 2010). Furthermore, the surface of ARM is positively charged at a pH of 4.5, which is below the pHpzc of ARM (pHPZC = 5.5–7). At this pH, most of the iron as exchangeable trivalent iron species and part of aluminum species have positive charge. This condition is favorable for the adsorption of ferricyanide anions due to the Coulombic attraction as follows:

pseudo-second order were investigated (Doğan et al., 2004; Sparks, 1986; Fulazzaky et al., 2013). The parameters of kinetic models and the parameters of validity of the kinetic model including the values of the correlation coefficients (R2), the normalized standard deviation (NSD) (Eq. (5)), average relative error (ARE) (Eq. (6)) and the root mean square error (RMSE) have been presented in Table 5.

NSD = 100

100 N

ARE =

1 N−1 N

∑

2

qtexp − qtcal ⎤ ⎥ qtexp ⎦i ⎣

i=1

(5)

qeexp − qecal qeexp

i=1

qtexp

N

∑ ⎡⎢

(6)

i

qtcal

where and (mg/g) are the experimental and calculated ferricyanide adsorption capacity on ARM at time t, respectively, and N is the number of measurements made. The smaller values of NSD and ARE represent more accurate qt estimation (Kumar et al., 2008; Hameed and Rahman, 2008). The results show that the pseudo-first order, linearized form of pseudo-second order of type (II, III & IV), Elovich and the intraparticle diffusion kinetic model have low R2, high NSD, ARE and RMSE, which indicated that these models do not give a good regression. The values of R2, NSD, ARE and RMSE for the Pseudo-second order Type (I) model are obtained better than other kinetic models for different initial cyanide concentrations at both a pH of 7.5 and 4.5. These results strongly suggest that the experimental data have appropriately represented the Pseudo-second order Type (I) model, which conformed that the ratecontrolling step in the ferricyanide adsorption process onto ARM for all different ferricyanide concentrations was chemisorption interaction. According to this model, two reactions (series or in parallel) are proposed including the fast reaction that quickly reaches equilibrium and slow reaction that can continues for a long period of time (Behnamfard and Salarirad, 2009). Fig. 5 compares the real qt and calculated qt using

3 ≡ S − OH + [Fe (CN )6]3 − → (≡S )3 − Fe (CN )6 + 3OH−

(7)

where S represents the surface of ARM. In this condition, ferricyanide anions were more active for chemical adsorption and surface hydrolysis reactions on ARM, so the ferricyanide anion removal capacity increased (LIU et al., 2007). The ferricyanide adsorption onto ARM can be carried out by forming inner-sphere surface complexes (Altundoğan et al., 2002). This is while the surface of the ARM is negatively charged at a pH of 7.5, which is above the pHpzc and the competition adsorption of OH against ferricyanide anion ions was much stronger. Also, ferricyanide anion adsorption must compete with Coulombic repulsion in this condition which leads to a decrease in the amount of ferricyanide absorbtion onto the ARM (Castaldi et al., 2010). The diffusion mechanism cannot be evaluated using the kinetic model of the pseudo-second order. Thus, it was evaluated by the theory proposed by Weber and Morris about the intraparticle diffusion model.

Table 4 Kinetic models (Li et al., 2009; Olgun and Atar, 2009; Tan et al., 2008; Nandi et al., 2009; Khambhaty et al., 2009; Donia et al., 2009; Kumar and Sivanesan, 2006). Kinetic models

Equations

Linear expression

Parameters

Pseudo-first order Pseudo-second order Type (I)

qt = qe [1 − exp (−k1p t )]

ln (qe − qt ) = lnqe − k1p t

qe = exp(intercept), k1p = −(slope)

qt = k2p qe2 t /(1 + qe k2p t )

t / qt = 1/ k2p qe2 + t / qe

qe = slope−1, k2p = (slope2)/intercept

Pseudo-second order Type (II)

qt =

Pseudo-second order Type (III)

qt =

Pseudo-second order Type (VI)

qt =

k2p qe2 t /(1 k2p qe2 t /(1 k2p qe2 t /(1

(1/ k2p qe2)(1/ t )

+ (1/ qe )

qe = intercept−1, k2p = (intercept2)/slope

+ qe k2p t )

1/ qt =

+ qe k2p t )

qt = qe − (1/ k2p qe ) qt / t

qe = intercept

+ qe k2p t )

qt / t = k2p qe2 − k2p qe qt

qe = −intercept/slope, k2p = (slope2)/intercept

−1

, k2p = −1/(slope × intercept)

Elovich

qt = βln (αβt )

qt = βln (αβ ) + βlnt

β = slope, α = (slope)−1exp(intercept/slope)

Intraparticle diffusion

qt = kp t 0.5

qt = kp t 0.5

kp = slope

281

Journal of Environmental Management 227 (2018) 277–285

N. Deihimi et al.

Table 5 Kinetic model parameters for ferricyanide adsorption onto ARM. Model

Parameter

Initial cyanide concentration (ppm) 300

Pseudo-first order

Pseudo-second order Type(I)

Pseudo-second order Type(II)

Pseudo-second order Type(III)

Pseudo-second order Type(IV)

Elovich

R2 NSD ARE RMSE k1p qe R2 NSD ARE RMSE k2p qe R2 NSD ARE RMSE k2p qe R2 NSD ARE RMSE k2p qe R2 NSD ARE RMSE k2p qe R2 NSD ARE RMSE α β

150

100

75

50

pH = 7.5

pH = 4.5

pH = 7.5

pH = 4.5

pH = 7.5

pH = 4.5

pH = 7.5

pH = 4.5

pH = 7.5

pH = 4.5

0.49 88.81 82.17 2.46 0.003 4.6 0.99 1.29 1.89 0.16 0.34 2.88 0.9 2.37 1.9 0.09 0.69 2.82 0.87 2.66 2.03 0.07 0.69 2.82 0.87 2.53 1.94 0.07 0.6 2.84 0.78 3.61 2.65 0.09 2.6E+07 18.25

0.73 97.23 89.79 7.997 0.006 2.69 0.99 7.96 4.49 0.38 0.02 8.56 0.74 10.04 7.023 1.01 0.02 8.53 0.71 12.36 10.02 1.02 0.02 8.51 0.71 11.09 9.24 0.9 0.02 8.67 0.92 4.08 2.78 0.30 316.258 1.30

0.6 94.43 88.94 3.46 0.007 1.71 0.99 4.42 3.27 0.10 0.24 2.86 0.9 4.5 4.1 0.11 0.32 2.77 0.87 7.25 6.15 0.11 0.3 2.8 0.87 7.99 5.25 0.10 0.26 2.84 0.89 4.38 3.43 0.12 386 4.08

0.87 90.83 83.66 6.764 0.01 2.74 0.99 7.96 4.49 0.38 0.01 8.20 0.9 5.55 3.90 1.01 0.032 7.74 0.87 6.875 5.558 0.46 0.03 7.77 0.87 7.005 4.76 0.46 0.02 8.11 0.92 11.63 9.68 0.96 426203 2.822

0.47 101.51 96.21 3.67 0.002 1.27 1 2.01 1.52 0.05 0.69 2.58 0.93 2.04 1.64 0.05 0.73 2.56 0.91 2.03 1.6 0.05 0.72 2.57 0.91 2.1 1.5 0.05 0.65 2.58 0.81 3.15 2.37 0.08 7397551.3 8.58

0.68 92.95 85.65 6.637 0.007 2.89 0.99 6.08 4.49 0.38 0.01 7.80 0.93 5.245 4.16 0.37 0.031 7.49 0.91 4.041 4.95 0.36 0.03 7.35 0.91 8.61 5.57 0.52 0.02 7.60 0.75 4.08 2.78 0.30 79.19 1.28

0.75 96.97 91.6 2.62 0.003 1.84 0.99 4.99 1.8 0.15 0.22 2.15 0.85 5.14 4.11 0.10 0.54 2 0.81 5.08 3.94 0.05 0.5 2.02 0.81 5.33 3.75 0.09 0.4 2.06 0.95 2.82 2 0.05 545.64 5.95

0.73 96.97 91.6 6.043 0.006 2.672 0.99 4.86 3.15 0.27 0.02 6.84 0.93 3.33 2.7 0.20 0.03 6.64 0.91 3.33 2.73 0.20 0.03 6.66 0.91 3.39 2.83 0.20 0.03 2.06 0.93 2.69 1.77 0.21 105.86 1.51

0.71 96.22 90.81 2.33 0.003 1.52 0.99 11.5 10.88 0.19 0.25 1.95 0.89 4.54 4.07 0.08 0.45 1.84 0.81 7.45 5.69 0.12 0.41 1.86 0.89 4.49 3.63 0.07 0.36 1.89 0.95 3.69 3.12 0.06 110.58 5.61

0.87 94.31 86.91 4.52 0.004 2.67 0.99 4.38 3.12 0.36 0.02 5.22 0.85 4.54 4.07 0.24 0.45 4.92 0.8 4.82 3.17 0.23 0.05 4.9 0.8 5.2 4.06 0.24 0.36 1.89 0.92 3.29 2.4 0.16 108.54 2.09

In this model, the plot of the adsorption amount is drawn versus t0.5. If the plot is linear, it indicated that the intraparticle diffusion is dominated. Otherwise, the intraparticle diffusion in the adsorption process is not the rate-limiting step (Behnamfard and Salarirad, 2009; Tan et al., 2008; Zhu et al., 2008). The plots of qt versus t0.5 for the initial cyanide concentrations of 50, 75, 100, 150 and 200 mg/L for a pH of 7.5 and 4.5 are presented in Fig. 6. As can be seen from Fig. 6, the plots are

nonlinear which indicates that the intraparticle diffusion for the whole reaction is not the rate-limiting step. The data can be divided into two linear steps with different slopes, in which the first step is fast due to ferricyanide adsorption onto ARM and the second step is slow due to the ferricyanide ions diffusion into the ARM inner adsorption sites (Behnamfard and Salarirad, 2009).

Fig. 5. Calculated qe by Pseudo-second order Type (I) model versus real qe at different ferricyanide concentrations (a) pH = 7.5 and (b) pH = 4.5. 282

Journal of Environmental Management 227 (2018) 277–285

N. Deihimi et al.

Fig. 6. The plots of Weber–Morris for ferricyanide adsorption onto ARM at different initial ferricyanide concentrations (a) pH = 7.5 and (b) pH = 4.5.

3.3. Equilibrium studies

Table 7 Isotherm model parameters for ferricyanide adsorption onto ARM.

To study the equilibrium isotherms, 100 mL of ferricyanide solution with different initial concentrations of 50, 75, 100, 150 and 300 mg/L were stirred by 3 g of ARM at 180 rpm for 120 min in two pHs of 7.5 and 4.5. Equilibrium isotherms of ferricyanide adsorption onto ARM are evaluated by two-parameter models (Langmuir, Freundlich, Temkin and Dubinin–Radushkevich) and three-parameter models (Redlich–Peterson) (Table 6) to study the removal amount of ferricyanide at equilibrium by unit mass of ARM from the solution at constant temperature. The isotherm model parameters were determined using the linear regression method of least squares by the solver add-in function, in Microsoft Excel (Behnamfard and Salarirad, 2009; Sadeghalvad et al., 2016). The Langmuir adsorption model indicates that the adsorbent is homogeneous and the adsorption onto the adsorbent surface is monolayer. Moreover, all sites of the adsorbent are identical and energetically equivalent. In this model, the favorable range of the separation factor (RL) is 0 < RL < 1. The Freundlich adsorption model indicates that the adsorbent surfaces are heterogeneous and the adsorption onto the adsorbent surface is multilayer. Also, adsorption is reversible. In this model, the favorable range of adsorption intensity (n) is 0 < 1/n < 1. The Temkin model investigates the heat of adsorption and some indirect interactions of adsorbate/adsorbate in the adsorption process. The fall in the heat of adsorption due to adsorbate/adsorbate interactions is linear rather than logarithmic. The Dubinin–Radushkevich model investigates the energy of adsorption and determines the type of adsorption process (physical adsorption (E < 8), chemical adsorption or ion exchange (E > 16) and particle diffusion that governs the reaction (E > 16)). In this model, it is assumed that the amount adsorbed for any adsorbate concentration follows a Gaussian function of the Polanyi potential. The three-parameter model of Dubinin–Radushkevich is used for the homogeneous or heterogeneous adsorption process. It includes the features of two isotherm

Model Langmuir

Two parameters

Freundlich

Temkin

Dubinin–Radushkevich

Three parameter models

Redlich–Peterson

Parameter −1

qm(mgg ) RL KL R2 RMSE ARE kf(mg1−1/n L1/n g−1) n R2 RMSE ARE BT (KJ/mol) AT(Lm/g) R2 RMSE ARE qm(mg/g) E(KJ/mol) R2 RMSE ARE BRP ARP g R2 RMSE ARE

pH = 7.5

pH = 4.5

2.73 0.03–0.005 0.64 0.999 0.245 6.757 1.58

8.81 0.08–0.27 0.052 0.969 0.694 0.957 2.82

8.65 0.83 0.256202 7.6481 9938.115 506.659 0.467 0.227 6.892 4.028 22617.03 0.871 0.2291 6.892 1.53 3.26 0.95 0.975 0.903 5.87

4.89 0.776 0.640587 6.882751 1848.383 2.311 0.764 0.638 6.610 3.446 26411.01 0.781 0.556 6.419 0.349 1.445 0.866 0.976 0.634 6.584

models of Langmuir and Freundlich (Gimbert et al., 2008; Sadeghalvad et al., 2016; Castaldi et al., 2010; Khambhaty et al., 2009). The results of isotherm studies using two-parameter models (Table 7) showed that the process of ferricyanide adsorption onto ARM is effectively performed (Langmuir: 0 < RL < 1 and Freundlich:

Table 6 Isotherms models (Olgun and Atar, 2009; Khambhaty et al., 2009; Sari et al., 2009; El Nemr, 2009; Altundoğan et al., 2002; Han et al., 2009; Başar, 2006). Isotherms models

Equations

Linear expression

Parameters

Langmuir

qe = (qm KL Ce )/(1 + KL Ce )

Ce / qe = (1/ KL qm) + (Ce / qm)

qm : (slope )−1 KL: slope /intercept

Freundlich

qe = KF (Ce )1/ n

lnqe = lnKF + n−1lnCe

KF : exp (intercept ),n = (slope )−1

Temkin

qe = qmln(KT Ce )

qe = qm lnKT + qm lnCe

qm : slope, KT : exp (intercept / slope )

Dubinin–Radushkevich

qe = qmexp(−Dε 2) ,

lnqe = lnqm − Dε 2

qm : exp (intercept ), D: −slope

ln[(APR Ce / qe ) − 1] = glnCe + ln BRP

D: slope, BRP : exp (intercept ), APR : Optimized using a trial and error method

ε = RTln (1 + Ce−1) Redlich–Peterson

qe = (APR Ce )/(1 + BRP Ceg )

283

Journal of Environmental Management 227 (2018) 277–285

N. Deihimi et al.

Fig. 7. FTIR of ARM: (a) before ferricyanide adsorption and (b) after ferricyanide adsorption.

Fig. 8. SEM micrograph of ARM: (a) before ferricyanide adsorption and (b) after ferricyanide adsorption.

0 < 1/n < 1) with particle diffusion that governs the reaction (Dubinin–Radushkevich model: E > 16 kJ/mol) between ferricyanide ions and the ARM molecule by 2.73 mg/g maximum ferricyanide adsorption capacity (qm of Langmuir model) at a pH of 7.5 and 8.81 mg/g at a pH of 4.5. The order of the best model based on the R2 values, ARE and RMSE at a pH of 7.5 (Table 7) are as follow: Langmuir > Redlich– Peterson > Dubinin– Radushkevich > Freundlich > Temkin. The order of the best models at a pH of 4.5 are as follow: Redlich– Peterson > Langmuir > Dubinin– Radushkevich > Freundlich > Temkin. The R2 value of the Langmuir model is more than the other models at a pH of 7.5 that indicated the adsorption process is homogeneous with monolayer adsorption in which all sites for adsorption in the ARM are identical and energetically equivalent. This is while the R2 value of the Redlich– Peterson model is more than the other models at a pH of 4.5. According to the assumption of the three-parameter models of Redlich–Peterson (Table 6), the parameter value of g is close to unity (Table 7), which indicated this model converges with the Langmuir model. It confirms that the ferricyanide adsorption process onto ARM is homogeneous. The results show that maximum ferricyanide adsorption capacity was increased with a decrease in pH. The raw red mud is a heterogeneous mixture of several minerals with various active sites which do not occur in the monolayer coverage of adsorbate onto the heterogeneous surface of the absorbent. The heterogeneous surface of raw red mud is modified and changed to the homogeneous surface (according to the Langmuir isotherm) with the activation method of washing with HCl and then precipitating with ammoniac. The results show that the activation process improves the adsorbent efficiency and increases the adsorption capacity of red mud.

almost similar. The broad bands at the 3600-3100 cm−1 region and weak bands at 1628 cm−1 were attributed to OeH stretching vibration and OeH bending vibration, respectively (Ye et al., 2016). The peak at 1410 cm−1 corresponds to the existence of CaCO3 in RM (Cao et al., 2014; Gök et al., 2007). The bands at 1009 cm−1 can be corresponded to Si–O–Si and Si–O–Al vibrations and symmetric bending vibrations of Si–O–Si and O–Si–O (Noroozifar et al., 2009). The bands in the region of 400–500 cm−1 (the bands of 424 cm−1 and 453 cm−1) indicated the T-O bond (where T = Si or Al). Moreover, the bands in the region of 460–500 cm−1 are due to stretching vibrations of the Fe-O bond (Castaldi et al., 2008). The stretching band at 565 cm−1 can be attributed to the Si-O-Al bonds, and aluminosilicates existing in RM. The obvious new band at 2052 cm−1 is shown after adsorption, which is assigned to ferricyanide adsorption on the ARM. This indicates that ferricyanide adsorption has been successfully performed.

3.4. FTIR studies

The modification of red mud with the activation method of washing with HCl and then precipitating with ammoniac was evaluated to improve ferricyanide absorbency so that the maximum ferricyanide adsorption capacity of 8.81 mg/g was achieved at a pH of 4.5. Given the

3.5. SEM studies The scanning electron microscope was used to investigate the surface and morphology of the ARM before and after ferricyanide adsorption. Fig. 8 (a) shows the SEM of ARM before ferricyanide adsorption, which clearly shows the shape of aggregate particles with amorphous forms due to the precipitation of metal oxides and soluble material in hydrochloric acid during the modification process of red mud. SEM images of ARM after ferricyanide adsorption clearly indicate the presence of round shaped aggregate particles, which confirmed the adsorption of ferricyanide in the ARM. 4. Conclusion

The FT-IR spectrum of ARM before and after adsorption was shown in Fig. 7. The peaks of adsorption bands in the FT-IR spectrums are 284

Journal of Environmental Management 227 (2018) 277–285

N. Deihimi et al.

ANOVA analysis, pH, adsorbent dosage, ferricyanide concentration and time had a significant effect on the process of ferricyanide adsorption onto the ARM. It was concluded from experimental results that the optimum condition for the ferricyanide removal was: pH = 5.6, adsorbent dosage = 2.59 g, ferricyanide concentration = 60 ppm and time = 60 min. The kinetic studies showed that the pseudo-second order Type (I) model controls chemical kinetics. They confirmed that the rate-controlling step in the ferricyanide adsorption process onto ARM was the chemisorption interaction which depended on the concentration of ferricyanide ions and the number of active sites of ARM. Based on the intraparticle model, the ferricyanide adsorption process onto ARM is a two step process in which the adsorption of cyanide is fast due to the ferricyanide adsorption onto ARM in the first step, while the second step is slow due to the ferricyanide ions diffusion into the ARM inner adsorption sites. Isotherm studies showed that the ferricyanide adsorption process onto ARM indicates being homogeneous with monolayer adsorption. The results show that the mechanism of ferricyanide adsorption onto the ARM is a physico-chemical process.

Gök, A., Omastová, M., Prokeš, J., 2007. Synthesis and characterization of red mud/ polyaniline composites: electrical properties and thermal stability. Eur. Polym. J. 43, 2471–2480. Hameed, B., Rahman, A., 2008. Removal of phenol from aqueous solutions by adsorption onto activated carbon prepared from biomass material. J. Hazard. Mater. 160, 576–581. Han, R., Zhang, J., Han, P., Wang, Y., Zhao, Z., Tang, M., 2009. Study of equilibrium, kinetic and thermodynamic parameters about methylene blue adsorption onto natural zeolite. Chem. Eng. J. 145, 496–504. Hanela, S., Durán, J., Jacobo, S., 2015. Removal of iron–cyanide complexes from wastewaters by combined UV–ozone and modified zeolite treatment. J. Environ. Chem. Eng. 3, 1794–1801. Hoseinian, F.S., Irannajad, M., Nooshabadi, A.J., 2015. Ion flotation for removal of Ni (II) and Zn (II) ions from wastewaters. Int. J. Miner. Process. 143, 131–137. Khambhaty, Y., Mody, K., Basha, S., Jha, B., 2009. Kinetics, equilibrium and thermodynamic studies on biosorption of hexavalent chromium by dead fungal biomass of marine Aspergillus Niger. Chem. Eng. J. 145, 489–495. Kumar, K.V., Porkodi, K., Rocha, F., 2008. Isotherms and thermodynamics by linear and non-linear regression analysis for the sorption of methylene blue onto activated carbon: comparison of various error functions. J. Hazard. Mater. 151, 794–804. Kumar, K.V., Sivanesan, S., 2006. Selection of optimum sorption kinetics: comparison of linear and non-linear method. J. Hazard. Mater. 134, 277–279. Kuyucak, N., Akcil, A., 2013. Cyanide and removal options from effluents in gold mining and metallurgical processes. Miner. Eng. 50, 13–29. Li, K., Zheng, Z., Huang, X., Zhao, G., Feng, J., Zhang, J., 2009. Equilibrium, kinetic and thermodynamic studies on the adsorption of 2-nitroaniline onto activated carbon prepared from cotton stalk fibre. J. Hazard. Mater. 166, 213–220. LIU, C-j, LI, Y-z, LUAN, Z-k, CHEN, Z-y, ZHANG, Z-g, JIA, Z-p, 2007. Adsorption removal of phosphate from aqueous solution by active red mud. J. Environ. Sci. 19, 1166–1170. Liu, Y., Naidu, R., Ming, H., 2011. Red mud as an amendment for pollutants in solid and liquid phases. Geoderma 163, 1–12. Logsdon, M.J., Hagelstein, K., Mudder, T., 1999. The Management of Cyanide in Gold Extraction. International Council on Metals and the Environment Ottawa. Meeussen, J.C., Keizer, M.G., De Haan, F.A., 1992. Chemical stability and decomposition rate of iron cyanide complexes in soil solutions. Environ. Sci. Technol. 26, 511–516. Mudder, T., Botz, M., 2004. Cyanide and society: a critical review. ejmp & ep. Eur. J. Miner. Process. Environ. Protect. 4, 62–74. Nandi, B.K., Goswami, A., Purkait, M.K., 2009. Adsorption characteristics of brilliant green dye on kaolin. J. Hazard. Mater. 161, 387–395. Noroozifar, M., Khorasani-Motlagh, M., Fard, P.A., 2009. Cyanide uptake from wastewater by modified natrolite zeolite–iron oxyhydroxide system: application of isotherm and kinetic models. J. Hazard. Mater. 166, 1060–1066. Nsimba, E.B., 2009. Cyanide and Cyanide Complexes in the Goldmine Polluted Land in the East and Central Rand Goldfields. (South Africa). Olgun, A., Atar, N., 2009. Equilibrium and kinetic adsorption study of Basic Yellow 28 and Basic Red 46 by a boron industry waste. J. Hazard. Mater. 161, 148–156. Rai, S., Wasewar, K., Mukhopadhyay, J., Yoo, C.K., Uslu, H., 2012. Neutralization and utilization of red mud for its better waste management. World 6, 5410. Sadeghalvad, B., Azadmehr, A., Hezarkhani, A., 2016. Enhancing adsorptive removal of sulfate by metal layered double hydroxide functionalized quartz-albitophire iron ore waste: preparation, characterization and properties. RSC Adv. 6, 67630–67642. Sari, A., Mendil, D., Tuzen, M., Soylak, M., 2009. Biosorption of palladium (II) from aqueous solution by moss (Racomitrium lanuginosum) biomass: equilibrium, kinetic and thermodynamic studies. J. Hazard. Mater. 162, 874–879. Schwarz, M., Lalík, V., 2012. Possibilities of Exploitation of Bauxite Residue from Alumina Production. Recent Researches in Metallurgical Engineering-From Extraction to Forming. InTech. Shifrin, N.S., Beck, B.D., Gauthier, T.D., Chapnick, S.D., Goodman, G., 1996. Chemistry, toxicology, and human health risk of cyanide compounds in soils at former manufactured gas plant sites. Regul. Toxicol. Pharmacol. 23, 106–116. Sparks, D.L., 1986. Kinetics of ionic reactions in clay minerals and soils. Adv. Agron. 38, 231–266. Sutar, H., Mishra, S.C., Sahoo, S.K., Maharana, H., 2014. Progress of Red Mud Utilization: an Overview. Tan, I., Ahmad, A.L., Hameed, B., 2008. Adsorption of basic dye on high-surface-area activated carbon prepared from coconut husk: equilibrium, kinetic and thermodynamic studies. J. Hazard. Mater. 154, 337–346. Wang, S., Boyjoo, Y., Choueib, A., Zhu, Z., 2005. Removal of dyes from aqueous solution using fly ash and red mud. Water Res. 39, 129–138. Ye, J., Cong, X., Zhang, P., Zeng, G., Hoffmann, E., Liu, Y., Wu, Y., Zhang, H., Fang, W., Hahn, H.H., 2016. Application of acid-activated Bauxsol for wastewater treatment with high phosphate concentration: characterization, adsorption optimization, and desorption behaviors. J. Environ. Manag. 167, 1–7. Young, C., Jordan, T., 1995. Cyanide remediation: current and past technologies. In: Proceedings of the 10th Annual Conference on Hazardous Waste Research. Kansas State University, Manhattan, KS, pp. 104–129. Zhu, S., Hou, H., Xue, Y., 2008. RETRACTED: kinetic and isothermal studies of lead ion adsorption onto bentonite. Appl. Clay Sci. 40, 171–178.

References Altundoğan, H.S., Altundoğan, S., Tümen, F., Bildik, M., 2002. Arsenic adsorption from aqueous solutions by activated red mud. Waste Manag. 22, 357–363. Başar, C.A., 2006. Applicability of the various adsorption models of three dyes adsorption onto activated carbon prepared waste apricot. J. Hazard. Mater. 135, 232–241. Behnamfard, A., Salarirad, M.M., 2009. Equilibrium and kinetic studies on free cyanide adsorption from aqueous solution by activated carbon. J. Hazard. Mater. 170, 127–133. Bhatnagar, A., Vilar, V.J., Botelho, C.M., Boaventura, R.A., 2011. A review of the use of red mud as adsorbent for the removal of toxic pollutants from water and wastewater. Environ. Technol. 32, 231–249. Botz, M.M., 2001. Overview of Cyanide Treatment Methods. Mining Environmental Management, Mining Journal Ltd., London, UK, pp. 28–30. Brunori, C., Cremisini, C., Massanisso, P., Pinto, V., Torricelli, L., 2005. Reuse of a treated red mud bauxite waste: studies on environmental compatibility. J. Hazard. Mater. 117, 55–63. Cao, J.-L., Yan, Z.-L., Deng, Q.-F., Wang, Y., Yuan, Z.-Y., Sun, G., Jia, T.-K., Wang, X.-D., Bala, H., Zhang, Z.-Y., 2014. Mesoporous modified-red-mud supported Ni catalysts for ammonia decomposition to hydrogen. Int. J. Hydrogen Energy 39, 5747–5755. Castaldi, P., Silvetti, M., Garau, G., Deiana, S., 2010. Influence of the pH on the accumulation of phosphate by red mud (a bauxite ore processing waste). J. Hazard. Mater. 182, 266–272. Castaldi, P., Silvetti, M., Santona, L., Enzo, S., Melis, P., 2008. XRD, FTIR, and thermal analysis of bauxite ore-processing waste (red mud) exchanged with heavy metals. Clay Clay Miner. 56, 461–469. Dash, R., Balomajumder, C., Kumar, A., 2005. An overview of removal methods of cyanide from industrial wastewater. In: 7th International Congress on Civil Engineering. Dash, R.R., Gaur, A., Balomajumder, C., 2009. Cyanide in industrial wastewaters and its removal: a review on biotreatment. J. Hazard. Mater. 163, 1–11. Deihimi, N., Irannajad, M., Rezai, B., 2018. Characterization studies of red mud modification processes as adsorbent for enhancing ferricyanide removal. J. Environ. Manag. 206, 266–275. Doğan, M., Alkan, M., Türkyilmaz, A., Özdemir, Y., 2004. Kinetics and mechanism of removal of methylene blue by adsorption onto perlite. J. Hazard. Mater. 109, 141–148. Donato, D., Nichols, O., Possingham, H., Moore, M., Ricci, P., Noller, B., 2007. A critical review of the effects of gold cyanide-bearing tailings solutions on wildlife. Environ. Int. 33, 974–984. Donia, A.M., Atia, A.A., Moussa, E.M., El-Sherif, A.M., El-Magied, M.O.A., 2009. Removal of uranium (VI) from aqueous solutions using glycidyl methacrylate chelating resins. Hydrometallurgy 95, 183–189. Dzombak, D.A., Ghosh, R.S., Wong-Chong, G.M., 2005. Cyanide in Water and Soil: Chemistry, Risk, and Management. CRC press. El Nemr, A., 2009. Potential of pomegranate husk carbon for Cr (VI) removal from wastewater: kinetic and isotherm studies. J. Hazard. Mater. 161, 132–141. Fulazzaky, M.A., Khamidun, M.H., Omar, R., 2013. Understanding of mass transfer resistance for the adsorption of solute onto porous material from the modified mass transfer factor models. Chem. Eng. J. 228, 1023–1029. Gebresemati, M., Gabbiye, N., Sahu, O., 2017. Sorption of cyanide from aqueous medium by coffee husk: response surface methodology. J. Appl. Res. Technol. 15, 27–35. Gimbert, F., Morin-Crini, N., Renault, F., Badot, P.-M., Crini, G., 2008. Adsorption isotherm models for dye removal by cationized starch-based material in a single component system: error analysis. J. Hazard. Mater. 157, 34–46.

285