In this study, a new functionalized multi-walled carbon nanotubes (MWCNTs) were synthesized by chemical modification with 3-aminopyrazole (MWCNTs-f) and applied as an efficient adsorbent. In order to comparative study, carboxylated MWCNTs (MWCNTs-COOH) and MWCNTs-f were applied to the mercury and arsenic adsorption from aqueous solutions. The adsorbents were characterized by FT-IR, EDX, FE-SEM, TGA, and BET. The effects of pH, adsorbent dose, and initial ions concentration on the adsorption efficiency were investigated by central composite design and optimum conditions were achieved. The quadratic model was applied for analysis of variance and indicated that pH was the most effective parameter in the adsorption process. Also, the pseudo-second-order model has achieved from the adsorption kinetic studies. Furthermore, the experimental data was well fitted with the Langmuir isotherm and the maximum adsorption capacities were obtained 112 and 133 mg.g−1 for the adsorption of Hg(II) and As(III) by MWCNTs-f. Moreover, negative values of ∆G° and positive values of ∆H° and ∆S° revealed that the adsorption reactions were spontaneous and endothermic with the increase in randomness. Furthermore, desorption study revealed the favorable regeneration ability of MWCNTs-f even after 3 adsorption-desorption cycles.
Keywords: Adsorption, mercury, arsenic, MWCNTs, 3-aminopyrazole, central composite design.
1. Introduction
Industrial developments are not possible without the indispensable element water. Various types of pollutants and impurities present in water such as dissolved solids, synthetic dyes, agriculture runoffs, industrial effluents, and microorganisms. The presence of heavy metals is an add-on to this list, and these are so dangerous that they may actually lead to death, even at very low concentration, because they are non-biodegradable and accumulate in living tissues [1].
Various techniques such as precipitation, cementation, membrane separation, ion exchange, solvent, and extraction have been used to remove heavy metal pollutants from contaminated sewages. Most of these techniques are ineffective or extremely expensive in terms of energy and reagent consumption, especially when concentrations of dissolved metals are in the order of 1–100 mg.l-1 [1]. But adsorption technique is the promising process for removal of heavy metal ions from water and wastewater. it has been widely employed due to its cost-effectiveness, efficiency, and simplicity of operation, Several adsorbents have been applied for adsorption of Hg and As ions such as activated carbon [2-4], fly ash [5-7], zeolites[8, 9], biomaterials [10-13], polymers [14-17], nanoparticles [18-23], soils [24, 25], and resins [26-29]. But efforts dedicated to exploring new effective adsorbents continue to grow.
Since the discovery of carbon nanotubes (CNTs) by Iijima in 1991[30], the researcher tries to discover application and property of this component. Due to the high chemical, mechanical and thermal stability, a large specific area, layered structures and electrical properties, CNTs are the ideal Nanomaterials [31]. That’s why they have been widely employed to remove heavy metal ions from aqueous solution. But all the suggested applications have so far been limited by their practical insolubility in organic and aqueous solvents. Some of these limitations can be overcome by the controlled defect and sidewall functionalization of CNTs [31]. The chemical functionalization has been proposed as a common technique to increase the interactivity of CNT. To enhance the sorption capacity of CNT, the covalent and noncovalent functionalizations were suggested as two efficient approaches [32]. surface carboxylation of CNTs under oxidizing conditions by different acids such as HNO3, KMnO4, H2SO4, KOH, and H2O2 has been reported previously as the effective methods to enhanced the adsorption capabilities of CNTs [33-36]. Besides, researchers introduced amines groups as different functional groups to growth the potential of CNTs for metal ions sorption [37, 38].
In this study, authors prepared new modified multi-walled carbon nanotubes by 3-aminopyrazole (MWCNTs-f) as an effective adsorbent for Hg(II) and As(III) removal from aqueous solutions. The goal of this work provides a comparative study of two functionalized multi-walled carbon nanotubes (MWCNTs-COOH and MWCNTs-f) for the adsorption of Hg(II) and As(III) from aqueous solutions. Also, we investigate of the optimum conditions to the adsorption process by applying central composite design (CCD) a subset of response surface methodology (RSM). The CCD was chosen to investigate the interaction of the adsorption parameters such as adsorbent dosage, pH, and initial ions concentrations. Furthermore, isotherms, kinetics, thermodynamic, and desorption studies were carried out.
2. Materials and methods
2.1. Materials and characterization
Thionyl chloride (bp 76°C, density 1.64 g.cm-3), ethanol, Dimethylformamide (DMF), Dimethylsulfoxide (DMSO), tetrahydrofuran (THF), mercury chloride (HgCl2), and nitric acid were supplied from Merck Company. MWCNTs-COOH (%95 purity, OD: 10-20 nm, Length: 0.5-2 µm) from Neutrino Co., Ltd, 3-aminopyrazole and sodium arsenite (NaAsO2.7H2O) from Sigma-Aldrich were purchased and used as received. Field emission scanning electron microscopy (FE-SEM; MIRA3 TESCAN) was used for morphological analyses. In order to identify the surface functional groups of the as-synthesized samples Fourier-transform infrared (FT-IR) spectra was applied by using KBr tablets on a Thermo Nicolet Nexus 870 FT-IR spectrometer in the range of 700-3650 cm-1. Thermogravimetric analyses (TGA) was performed on an Erkin Elmer, Yris diamond from 25 to 600◦C at a heating rate of 10◦C.min-1 and a nitrogen flow of 200 ml.min-1. A furnace atomic absorption instrument (AAS, Thermo Electron Corporation M series) was employed to the determination of metal ion concentration. Energy dispersive X-ray spectroscopy (EDS; SAMx, France) was employed to determine the elemental composition of the precursor and the prepared MWCNTs. The surface area of the adsorbents were measured by nitrogen adsorption at 77 K using a BELJAPAN surface area analyzer (BELSORP mini II) and using a 16-point BET. A pp-201 portable pH meter (GOnDO Electronic Co., Ltd) was employed to adjust the pH of the solutions. A mini centrifuge (model: MSI 8, D.T.A.P Co. Ltd.) was used for separating solid from solution. The IKA RCT basic (Germany) was applied to heating and stirring the solutions.
2.2. Preparation of functional MWCNTs-f
MWCNTs-COOH (250 mg) were suspended in SOCl2 (40 mL) and DMF (1 mL). The mixture at 70°C was stirred for 48 h under reflux. Afterwards, the residual SOCl2 was removed by reduced pressure distillation to produce the Acyl Chloride-functionalized MWCNTs (MWCNTs-COCl). MWCNTs-COCl were mixed with 3-aminopyrazole (400 mg) in DMSO (20 mL) as a solvent and the mixture was stirred at 100°C for 96 h. The mixture was cooled at room temperature for one-half hour. Afterward, solid phase was separated by centrifuging and washed thoroughly with ethanol, THF, and water. Finally, the black solid was dried at room temperature for 8 h under vacuum condition.
2.3. Preparation of metal Stock Solution
It is very important before use of the laboratory glassware, rinsed it with a solution of 2% nitric acid, due to removing all impurities from the glassware and to prevent further adsorption of metal ions on the walls of the glassware. A metal stock solutions were prepared by dissolving salt metals to distilled water. In order to the required metal ions concentrations in the adsorption experiments, the stock solutions were further diluted with deionized water.
2.4. Adsorption experiments
Batch experiments were performed by shaking (250 rpm) 50 ml of a metal ions solutions in 100 ml Erlenmeyer flask for 2 h at room temperature and different pH, adsorbent dosage, and initial ions concentrations. The initial pH of each solution was adjusted by adding 0.1 N NaOH and 0.1 N HCl solution before adsorption. Finally, the mixtures were filtered, and concentration was measured by AAs. The percentage of removal (%) and sorption capacity qe (mg.g-1) was obtained as follows:
(1)
Where C0 and Ce are the initial and final concentrations (mg.l-1) of metal ions in the aqueous solution, respectively, V (L) is the volume of metal ion solution, and m (g) is the weight of sorbent.
2.5. Experimental Design
Central composite design (CCD) is a subset of response surface methodology (RSM) that was used to investigate the effects of variables, including the adsorbent dosage, pH, and initial ions concentration in the randomized fashion. In the CCD each factor has experiments performed at one of five levels, thus allowing one to fit a quadratic model. The CCD is specified via three operations namely: 2n factorial runs, 2n axial runs and six center runs, where n is the factors number [39]. In this study, it divided into 8, 6, and 6 factorial points, axial points, and replicates at the central point, respectively, which gives a total of 20 experiments. For statistical calculations, the variables X were coded as X according to the following equation:
(2)
Experimental parameters, coded variables, and their levels of the Hg(II) and As(III) ions adsorption process by the adsorbents are given in Table 1.
Table 1. Experimental parameters and their levels.
In order to analyze the results of experiments and considering all the linear, linear by linear interaction and square items, the quadratic model was applied to the prediction of each response. This model is as follows:
(3)
Where, R is the adsorption percent; Xi and Xj are independent variables; β0 is the model constant; βi is the linear, βii is the quadratic, and βij is the linear interaction effect of the input factor. Experimental data were evaluated with Design-Expert 8.0.7.1 (trial version) containing ANOVA in order to approximate estimation of the model quality and interaction among variables and response [40]. The quality of the model was examined with a correlation coefficient (R2) and F-test was used to check the statistical significance.
2.6. Desorption studies
Desorption experiments were carried out with metal-loaded to adsorbents in acid solution under sonication for 3 min. To optimize the concentration of the acid, different concentrations (0.1, 0.2, 0.5, 1.0) of HCl were considered. The adsorption (at optimum conditions) and desorption procedures were repeated three times, each time with fresh solutions of Hg(II) and As(III). The metal recovery was calculated by the following equation [41]:
(4)
3. Results and discussion
3.1. Characterization of modified MWCNTs
Fig1. demonstrates the schematic procedure for functionalization of MWCNTs-COOH by 3-aminopyrazole to obtained the modified adsorbent, MWCNTs-f. The product was characterized by FT-IR, FE-SEM, EDS, and TGA.
Fig 1. The modification route of the MWCNTs-COOH with 3-aminopyrazole.
The FT-IR analysis is one of the most significant characterization techniques applied to explain the changes in chemical structures. The FT-IR spectra of MWCNTs-COOH and MWCNTs-f were illustrated to investigate the surface functional groups at each step in the chemical functionalization (Fig2.a). In the spectra of MWCNTs-COOH, the peak around 1715 cm−1 can be assigned to C═O stretching of the carboxylic acid group. Also, peaks at 1083 and 3445 cm−1 can be attributed to C–O and OH stretching vibrations of the carboxylic group, respectively. Also, in the spectra of MWCNTs-f, the appearance of the new peaks at 1615 and 1500-1580 cm-1 can be assigned to the amide carbonyl groups [–C(═O)NH linkage] and C=N or C=C bonds that confirmed the formation of MWCNTs-f [37]. In addition, the peaks at 3300-3500 and 1200-1400 cm-1 can be assigned to the N–H and C-N stretching modes, respectively [42]. Thus, successful modification of MWCNTs-COOH by 3-aminopyrazole was confirmed by FT-IR spectra confirm.
Most of the functional groups in MWCNTs surface were thermally unstable, thus, TGA is the good evidence for the functionalization of MWCNTs. As shown in Fig2.b, the TGA curve of MWCNTs-COOH was stable and hardly decomposed below 600°C (weight loss was less than 11%). In TGA curve of MWCNTs-COOH, the weight loss about 5% below 200°C is observed which can be attributed to the thermal decomposition of carboxyl and hydroxyl groups. In addition, the TGA curve of MWCNTs-f shows two decomposition regions (Fig2.b). The first region at below 175°C may be attributed to the decomposition of hydroxyl groups due to the adsorbed water or humidity on the nanotube and the second region at 175-320°C could correspond to decomposition of 3-aminopyrazole. On the basis, the weight loss of the MWCNTs-COOH and MWCNTs-f at 320°C was calculated 6 and 46%, respectively.
Fig2.c shows the FE-SEM images together with EDS spectra that can supply the helpful information such as chemical composition of MWCNTs before and after functionalization. The EDS spectra of MWCNTs-f as compared to EDS of MWCNTs-COOH shows that nitrogen atoms in addition to carbon and oxygen are detected which can be due to the presence of the attached 3-aminopyrazole to MWCNTs. Also, as can be seen from this figure, the FE-SEM images of MWCNTs-COOH show tangled and agglomerated morphology. Moreover, in the MWCNTs-COOH surface, some swelling was observed which can be due to the covalent attachment of carboxyl groups [43]. The FE-SEM image of MWCNTs-f (Fig2.c) exhibits polymerized beehive like structure and the MWCNTs surface appears to be grown well with increased thickness.
Specific surface area of the adsorbents was determined by the Brunauer–Emmett–Teller (BET) method (acquisition and reduction). The BET surface areas of MWCNTs-COOH and MWCNT-f were 166.2 m2.g-1, and 7.3127 m2.g-1, respectively. Nevertheless, various groups introduced on the surface of MWCNTs-COOH provide numerous sorption sites, thereby enhanced the sorption capacities of MWCNTs-f. Similar observations were reported in the literature [44-47]. Fig2.d shows the average pore diameters (DBJH) of MWCNTs-f (24.78 nm) are evidently much larger than that of the MWCNTs-COOH (1.64 nm). The average pore diameter of 24.78 nm indicates that the MWCNTs-f prepared was in the mesopores region according to the IUPAC classification [48]. Also, Fig2.d revealed the N2 adsorption–desorption isotherms of MWCNTs-COOH and MWCNTs-f obtained at 77 K. The isotherms obtained from this figure could be classified as a type IV isotherm with H1 type hysteresis loops according to the original IUPAC classification [48]. However, according to the extended classification of adsorption isotherms, the isotherms are type IIb isotherms [49].
Fig 2. FT-IR spectra (a), TGA curve (b), EDS spectra and FE-SEM images (c), and N2 adsorption-desorption isotherm and pore size distribution (inset) (d) of functionalized MWCNTs.
3.2. Adsorption Process Modeling by CCD
The optimization of three variables (pH, adsorbent dosage, and initial ions concentration) on the adsorption efficiency of Hg(II) and As(III) ions from aqueous solutions was performed by employing CCD under RSM. In order to explore the mathematical relation between the response (percentage of adsorption) and variables, the quadratic model was chosen. The experimental results are tabulated in Table 2. According to this table, MWCNTs-f demonstrates suitable removal performance which can be due to the functionalized groups on the MWCNTs surface (the q¬¬e was 177.65 and 129.92 mg.g-1 for adsorption of Hg(II) and As(III) ions, respectively).
Table 2. CCD results for Hg(II) and As(III) removal by MWCNTs-COOH and MWCNTs-f as adsorbents.
In addition, the CCD technique and ANOVA were applied to develop the correlation between the selected factors in Table 1 and removal efficiency in this study. The model equations for coded values of the quadratic models for fitting experimental results of Hg(II) and As(III) removal by MWCNTs-COOH and MWCNTs-f, were expressed as:
(5)
(6)
(7)
(8)
Where X1, X2, and X3 are the coded values of pH, adsorbent dosage, and initial ions concentration, respectively and Y is log R. The sign of regression coefficient has a direct effect on the response, so it is important. The negative and positive signs before the terms demonstrate the antagonistic and synergistic effects of the variables [50]. On the one hand, two independent variables (pH and adsorbent dose) exhibited the positive significant effect on Hg(II) and As(III) removal for the two adsorbents. Also, the initial ions concentration in this study and all the quadratic terms showed the reducing effects for metals adsorption.
Table 3 show the statistical significance of the quadratic models evaluated with ANOVA. According to this table, the linear terms of variables (pH, adsorbent dosage, and initial ions concentration) and the quadratic terms of pH are statistically significant (p value<0.05) in two adsorbents. Besides, adjusted-R2 more than 0.9 indicates a desirable for models validation. The prediction-R2 is in reasonable agreement with the adjusted-R2 (Table 3). The coefficient of variation (C.V.%) is also less than 10%.
Table 3. ANOVA for the response surface quadratic model for adsorption of Hg(II) and As(III) by MWCNTs-COOH and MWCNTs-f.
The data obtained from the adsorption experiments and the values predicted by the response surface model for the MWCNTs-COOH and MWCNTs-f adsorbents are shown in Fig3. in Online Resource 1. The values of the R2 coefficient greater than 0.95 indicate a good correlation between the experimental data and predicted ones. Also, the second-order quadratic models for the observed results are significant because the p values of the models are small.
In order to evaluation of the effect of each factor on response parameter (removal %), the percent contribution was measured and tabulated in Table 4 in Online Resource 2 [51]. According to this table, more than 44 and 17% contribution in the case of Hg adsorption and 31 and 44% contribution in the case of As(III) adsorption by MWCNTs-COOH and MWCNTs-f, were obtained for individual parameters, respectively. But less than 1% of the contribution was for combine interaction in all cases. Among the parameters, pH has the largest contribution (in the case of Hg(II) adsorption: 38.9 and 13.24%, in the case of As(III) adsorption: 22.44 and 38.44% by MWCNTs-COOH and MWCNTs-f, respectively). Therefore, in these experiments, pH has effective roles in the adsorption process.
The perturbation plot (Fig4. in Online Resource 3) shows the comparative effects of all independent variables on the removal efficiency. The sharp curvature of the A factor (pH) in all cases shows that the response, heavy metals removal efficiency, was very sensitive to this variable. Also, the maximum removal efficiency occurred as the adsorbent dose reached the maximum level. Furthermore, with increasing metal ions concentration, the response achieved to the minimum level.
3.3. Response Surface Plotting
3.3.1. Effect of pH and initial ions concentration on adsorption process
Fig5.a is 3D response surface plot displaying the interactive effect of pH and metal ions concentration for Hg(II) and As(III) removal. This figure revealed that the adsorption process extremely depends on the solution pH in all cases. The optimum pH for Hg(II) and As(III) removal by the adsorbents was found to be in the range of 6.0-8.0. At pH below 6.0, the removal efficiency significantly increased with the pH growth. Because, the formation of mercuric hydroxide was increased, thus the sorption of Hg(OH)2 on the surface of adsorbents increased [52]. Besides, at pH less than 6.0, there is the strong competition between H+ and As(III) on the adsorption sites [53]. On the other hand, at pH higher than 8.0, the As(III) removal efficiency dropped due to the ionization of H3AsO3, that causes more competition between arsenite and OH- anions [54]. Also, at this pH, the Hg(II) removal efficiency reduced with increasing pH. It could be due to the complex formation of Hg(OH)3- and Hg(OH)4- in the solution that these complex could not be retained on the surface of the adsorbents because they are soluble in water [55]. Also, Fig5.a reveals that with decreasing the initial ions concentration from 60 to 10 ppm, the removal% was increased to the maximum amount at the constant adsorbent dose (20 mg). This sorption characteristic demonstrated that the saturation of surface was related to the initial ions concentrations. The initial ions concentration provides a significant driving force to overcome all mass transfer resistance of metal ion between the solid phases and aqueous [56].
3.3.2. Effect of pH and adsorbent dose on Hg(II) and As(III) adsorption
The 3D plots for the combined effects of adsorbent dose and pH on the Hg(II) and As(III) uptake by the MWCNTs-COOH and MWCNTs-f are represented in Fig5.b. Depending on the type of adsorbent and functional groups, the pH of the solution has a different effect on adsorption. Fig5.b represents that the metals adsorption extremely depends on the initial pH of the solution. Under acidic condition (pH<4.0), both adsorbents and metal ions are positively charged, so there were unfavorable electrostatic interactions as well as competition between H+ and metal cations for adsorption on available surface sites [57]. Thus, adsorption of Hg(II) and As(III) by the adsorbents was reduced at low pH. Also, at pH higher than 9.0, the Hg ions precipitation occurred that caused absorption efficiency decreased [58]. Also, at this pH, the fraction of the total As(III) as H2AsO3- will be decreased [59]. In addition, it could be seen that the adsorption rate smoothly increase when adsorbent dose increases from 10 to 20 mg and reached a maximum of 20 mg for the two adsorbents. It could be explained by easier penetration of metal ions to active sites due to increasing in the active sites in the higher amounts of adsorbents [60]. Similar observations are reported in the literature [55, 61, 62].
Fig 5. The 3D plots are showing the interactive effect of pH and initial ions concentration (a) and the interactive effect of pH and adsorbent dose (b). (1) Hg(II) removal by MWCNTs-COOH, (2) Hg(II) removal by MWCNTs-f, (3) As(III) removal by MWCNTs-COOH, and (4) As(III) removal by MWCNTs-f.
3.4. Process optimization
The optimization of Hg(II) and As(III) removal process by the adsorbents was carried out by using the Design Expert 8.0.1 software (trial version). In the optimization analysis, the goal criterion was set as maximum values for the responses (removal %). The optimum adsorption conditions achieved are presented in Table 5, with the desirability of one. This table reveals that the models supported the experimental data with a relatively small error less than 1.0%. In order to decrease the experimental errors for the evaluation of the isotherm models, kinetic parameters, and thermodynamic properties, the optimum conditions were chosen as Table 6 in Online Resource 4.
Table 5. The optimum adsorption conditions and models validation.
3.5. Adsorption isotherms
The equilibrium characteristics of adsorption were described through Langmuir, Freundlich, and Temkin isotherm models. As known, Langmuir model assumes a monolayer adsorption onto a surface containing the finite number of adsorption sites, specific homogeneous sites and finally, all sites are identical and energetically equivalent [63]. The linear form of Langmuir model is given as:
(9)
Where qmax (mg.g-1) is the maximum amount of sorbed metal per unit mass of adsorbent and KL (l.mg-1) is the Langmuir constants related to the adsorption heat. The dimensionless separation factor, RL, is the essential characteristics of Langmuir model. Its linear form is given by the following equation:
(10)
Where C0 (mg.l-1) is the initial concentration of adsorbate in solution. The RL values reveal the type of isotherm to be irreversible (RL = 0) , favorable (0 < RL < 1), , linear (RL = 1) or unfavorable (RL > 1) [64].
The Freundlich model widely uses to explain the adsorption characteristics of the heterogeneous surface and reversible adsorption. The linear form of this empirical model is given as [64, 65]:
(11)
Where Kf (mg.g-1) is related to the sorption capacity of adsorbent and n is related to the intensity of sorption and varies with surface heterogeneity and affinity. The n-1 value lies between 0 and 1 demonstrates an amount of adsorption intensity or heterogeneity of the surface. If n-1 values were near to the zero, heterogeneity of the surface was increased. Also, the n-1 value less than one suggest chemisorption process, instead, the value above unity imply the cooperative adsorption [64, 66].
Moreover, Temkin isotherm model was considered to the evaluation of adsorption isotherm. It assumes that heat of adsorption of all molecules in the layer would decrease linearly rather than logarithmic with coverage and uniform distribution of binding energy between adsorbent and adsorbate [67] and its linear form is given as:
(12)
(13)
Where B (J.mol-1) is constant related to the heat of adsorption, b is the Temkin isotherm constant, and KT is the equilibrium binding constant (l.g-1) [64, 68, 69].
The slope and the intercept of each linear plot of the isotherms are used to calculate the isotherm model parameters and listed in Table 7.
The R2 values in Table 7 and Fig6. in Online Resource 5 revealed that experimental data in all cases were fitted well with Langmuir equation in the studied temperature range (R2>0.92). This could be indicated the monolayer adsorption onto the surface of the adsorbents [70].
According to the Table 7, the KL values increased with increasing temperature from 25 to 45°C that revealed favorable metal ions adsorption would be expected at the higher temperature. This represents that the adsorption process by the adsorbents was endothermic [71]. However, the qmax values decreased with increasing solution temperature. Therefore, less adsorption of Hg(II) and As(III) ions would be expected at low temperature. The RL values were less than one (0< RL <1) that revealed Langmuir isotherm was favorable for adsorption of Hg(II) and As(III) on the adsorbents.
In Freundlich isotherm, the values of KF which calculated for MWCNTs-f, were larger than MWCNTs-COOH indicating that MWCNTs-f has the high affinity toward heavy metal ions. Also, deviation of the n-1 values from unity demonstrates a non-linear adsorption that takes place on heterogeneous surfaces. Furthermore, the n-1 values of MWCNTs-f were lower than MWCNTs-COOH denoting that MWCNTs-f surface was more heterogeneous than MWCNTs-COOH [72].
The values of b in Table 7 express the heat of metal adsorption by the adsorbents. The b values increased with increasing solution temperature from 25 to 45°C, indicating endothermic sorption. Also, it has been reported that the b values is less than 8.0 kJ.mol-1 which denoting a physical adsorption process. Thus, the low values in this study indicate a weak interaction between sorbate and sorbent, supporting a physisorption mechanism for this study [73].
Table 7. Adsorption isotherm constant values of Hg(II) and As(III) adsorption by MWCNTs-COOH.
3.6. Adsorption kinetics
Different models can be applied to analyze the kinetics of sorption process. In this work pseudo-first-order, pseudo-second-order, and Elovich kinetic models were applied to kinetic studies. Pseudo-first-order proposed a rate equation for the sorption of adsorbate from a liquid solution. This equation is as follows:
(14)
Integrating from t=0 to t=t and q=0 to q=q gives:
(15)
Where q and qe are the values of amount adsorbed per unit mass of adsorbent (mg.g-1) at any time (min) and at equilibrium, respectively, and k1 is the pseudo-first-order rate constant (min-1) [74, 75]. In this model by plotting the values of log (qe-q) vs. t qe and k1 values can be determined from the intercept and slope, respectively.
Another model for the analysis of sorption kinetic is pseudo-second-order and the rates law is given as:
(16)
Integrating from t=0 to t=t and q=0 to q=q gives:
(17)
(18)
Where k2 (g.mg−1.min−1) is the rate constant of the pseudo-second-order adsorption and h is the initial sorption rate (mg.g−1.min−1). Moreover, equation (17) can be rearranged to achieve a linear form [76, 77]:
(19)
In this model by plotting the values of t/q vs. t, it can be determined the qe and k2 values from the slope and intercept, respectively.
In recent years, the Elovich model has been used to describe the pollutants adsorption from aqueous solutions. This semi-empirical model usually employed to describing the chemisorption process assuming that the adsorbent surfaces are energetically heterogeneous. However, this model does not recommend any definite mechanism for adsorbent-adsorbate. The linear form of this model is as follow [78]:
(20)
Where α is initial sorption rate (mg.g−1.min−1) and β is desorption constant (g.mg−1) during any one experiment. The plot of qt vs. lnt was used to obtain the β and α constant.
In order to study the equilibrium time and adsorption rates, experiments were carried out at optimum conditions at different time intervals up to 300 min. The calculated design parameters (k1, k2, qe, α, β) and correlation coefficient (R2) could be obtained by the slope and intercept of each linear plot as given in Table 8 in Online Resource 6. Also, the equilibrium times were found to be 100 min for the adsorption of Hg(II) and As(III) ions by MWCNTs-COOH and MWCNTs-f. As can be seen from Table 8, the R2of the second order rate equation is more than 0.998 for all cases. Besides, the calculated qe from the second order rate equation for MWCNTs-COOH and MWCNTs-f were very close to the experimental values with the mean absolute relative error less than 4 percent (Table 8). So, the adsorption of mercury and arsenic ions by MWCNTs-COOH and MWCNTs-f can be well displayed by the pseudo-second-order kinetic model. Also, this model suggested that the rate controlling step is chemisorption in the adsorption process. Our findings are in good agreements with those reported in the literature [52, 53, 55].
3.7. Thermodynamic studies
In order to the evaluation of thermodynamic properties such as free energy change (∆G°), enthalpy change (∆H°), and entropy change (∆S°), the following equations were considered:
(21)
(22)
(23)
Where C0 and Ce (mg.l−1) are the initial and equilibrium metal ions concentration in the solution, respectively. Kd is the equilibrium constant, T is temperature of the solution (K) and R is the universal gas constant (8.314, J.K−1.mol−1) [79]. The values of ∆H° and ∆S° could be determined from slope and intercept of the plotting of ln kd vs. 1/T. The calculated thermodynamic parameters are given in Table 9.
Table 9. Thermodynamic parameters for the adsorption of Hg(II) and As(III) on the adsorbents at optimum conditions.
In thermodynamics, ∆G° is used to the evaluation of the spontaneity of the adsorption process; the more energetically desirable process has the higher negative ∆G° value. The negative values of ∆G° in Table 9 indicate that adsorption of Hg(II) and As(III) are the spontaneous processes. Further, it is considerable that the ∆G° values for mercury and arsenic adsorption decrease with increasing temperature from 25 to 45°C, denoting that the process was more efficient at the higher temperature [80]. Also, the ∆G° values, less than −15 kJ.mol−1, are regarded with the physical interactions between adsorbent sites and metal ions [65]. However, ∆H° and ∆S° values are positive, it represents that adsorption process is endothermic and randomness increases at the solid-solution interface within the adsorption [80]. Similar results were reported in the literature for the adsorption of Hg(II) ions from aqueous solution by MnO2/CNTs [81] and adsorption of As(III) ions from aqueous solution by iron oxide impregnated activated alumina [61].
3.8. Desorption study
Renewability is one of the important factors to the applicability of adsorbents. To investigation of this property, desorption study was carried out. According to the Fig7. in Online Resource 7, with increasing the concentration of HCl, the metal-loaded to the adsorbents were more desorbed. Further, the Hg(II) and As(III) removal efficiency reduced about 10 and 6% during consecutive cycles from 1 to 3 for MWCNTs-COOH and MWCNTs-f, respectively. The renewability of the adsorbents after 3 cycles in this work, is comparable to the literature [55, 58, 82-84].
3.9. Comparison of adsorbent performance with literature
The comparison between the performances of MWCNTs-f for the adsorption of Hg(II) and As(III) ions with different adsorbents are presented in Table 10. Due to the different experimental conditions that were employed in those obtained in literature, the direct comparison was difficult. However, the qmax values were widely different for various adsorbents. Ordinarily, functionalized MWCNTs were exhibited significant higher adsorption capacities than the pristine MWCNTs as well as activated carbon. Also, MWCNTs-f shows comparable adsorption capacity with other functionalize MWCNTs adsorbents that were used yet to mercury and arsenic adsorption. However, it was the deduction that MWCNTs-f has an applicability to the adsorption of As(III) in compared with other adsorbents that were used previously. Eventually, this comparison indicates that the MWCNTs-f exhibited an acceptable capacity for Hg(II) and As(III) adsorption from aqueous solutions.
Table 10. Comparison of adsorption capacities of various adsorbents for Hg(II) and As(III) ions adsorption.
4. Conclusions
In this work, the new functionalized MWCNTs was successfully synthesized and used for mercury and arsenic ions adsorption from aqueous solutions. FT-IR, TGA, EDS and FE-SEM analysis results confirmed the chemical modification process. Response surface methodology was applied for the evaluation of optimum values of the effective factors such as pH, adsorbent dose, and initial ions concentration. For correlating the experimental data, the quadratic model was applied. This model and ANOVA analysis results confirmed that the pH has the major effect on the adsorption process and the optimum conditions for adsorption of Hg(II) and As(III) by MWCNTs-COOH and MWCNTs-f were evaluated. Furthermore, the experimental data were in good agreement with the data predicted by the models with relatively small errors. The maximum removal efficiency of Hg(II) and As(III) under the optimum conditions were achieved as 80.5 and 72.4% by MWCNTs-f, respectively. Between the isotherms models, equilibrium data fitted well by the Langmuir isotherm for the mercury and arsenic adsorption. Also, the kinetic adsorption could be well explained by the pseudo-second-order kinetic equation. In addition, thermodynamic properties, ∆G°, ∆H°, and ∆S° were determined and exhibited that the adsorption processes were more efficient at the higher temperature and were endothermic as well as increasing in randomness. The adsorption–desorption experiments showed a small loss in the adsorption capacity of 6% after three cycles, demonstrating a good regeneration capacity of MWCNTs-f for treating wastewater containing mercury and arsenic ions. Comparison of MWCNTs-f adsorption capacity with other adsorbents in the literature revealed that there are the great potentials and applications of the new modified adsorbent of this work (MWCNTs-f) for adsorbing Hg(II) and As(III) ions from aqueous solutions.
Acknowledgements
The financial and encouragement support provided by the Research Vice Presidency of Semnan University, (www.semnan.ac.ir), zip code: 35131-19111, Tel/Fax: +98 2333654120, Semnan, Iran and Ayatollah Amoli Branch, Islamic Azad University, zip code: 46351-43358, Tel.: +98(11) 43217000-3, Fax:+98(11)43217041, info@iauamol.ac.ir, Amol, Iran.
1. Introduction
Industrial developments are not possible without the indispensable element water. Various types of pollutants and impurities present in water such as dissolved solids, synthetic dyes, agriculture runoffs, industrial effluents, and microorganisms. The presence of heavy metals is an add-on to this list, and these are so dangerous that they may actually lead to death, even at very low concentration, because they are non-biodegradable and accumulate in living tissues [1].
Various techniques such as precipitation, cementation, membrane separation, ion exchange, solvent, and extraction have been used to remove heavy metal pollutants from contaminated sewages. Most of these techniques are ineffective or extremely expensive in terms of energy and reagent consumption, especially when concentrations of dissolved metals are in the order of 1–100 mg.l-1 [1]. But adsorption technique is the promising process for removal of heavy metal ions from water and wastewater. it has been widely employed due to its cost-effectiveness, efficiency, and simplicity of operation, Several adsorbents have been applied for adsorption of Hg and As ions such as activated carbon [2-4], fly ash [5-7], zeolites[8, 9], biomaterials [10-13], polymers [14-17], nanoparticles [18-23], soils [24, 25], and resins [26-29]. But efforts dedicated to exploring new effective adsorbents continue to grow.
Since the discovery of carbon nanotubes (CNTs) by Iijima in 1991[30], the researcher tries to discover application and property of this component. Due to the high chemical, mechanical and thermal stability, a large specific area, layered structures and electrical properties, CNTs are the ideal Nanomaterials [31]. That’s why they have been widely employed to remove heavy metal ions from aqueous solution. But all the suggested applications have so far been limited by their practical insolubility in organic and aqueous solvents. Some of these limitations can be overcome by the controlled defect and sidewall functionalization of CNTs [31]. The chemical functionalization has been proposed as a common technique to increase the interactivity of CNT. To enhance the sorption capacity of CNT, the covalent and noncovalent functionalizations were suggested as two efficient approaches [32]. surface carboxylation of CNTs under oxidizing conditions by different acids such as HNO3, KMnO4, H2SO4, KOH, and H2O2 has been reported previously as the effective methods to enhanced the adsorption capabilities of CNTs [33-36]. Besides, researchers introduced amines groups as different functional groups to growth the potential of CNTs for metal ions sorption [37, 38].
In this study, authors prepared new modified multi-walled carbon nanotubes by 3-aminopyrazole (MWCNTs-f) as an effective adsorbent for Hg(II) and As(III) removal from aqueous solutions. The goal of this work provides a comparative study of two functionalized multi-walled carbon nanotubes (MWCNTs-COOH and MWCNTs-f) for the adsorption of Hg(II) and As(III) from aqueous solutions. Also, we investigate of the optimum conditions to the adsorption process by applying central composite design (CCD) a subset of response surface methodology (RSM). The CCD was chosen to investigate the interaction of the adsorption parameters such as adsorbent dosage, pH, and initial ions concentrations. Furthermore, isotherms, kinetics, thermodynamic, and desorption studies were carried out.
2. Materials and methods
2.1. Materials and characterization
Thionyl chloride (bp 76°C, density 1.64 g.cm-3), ethanol, Dimethylformamide (DMF), Dimethylsulfoxide (DMSO), tetrahydrofuran (THF), mercury chloride (HgCl2), and nitric acid were supplied from Merck Company. MWCNTs-COOH (%95 purity, OD: 10-20 nm, Length: 0.5-2 µm) from Neutrino Co., Ltd, 3-aminopyrazole and sodium arsenite (NaAsO2.7H2O) from Sigma-Aldrich were purchased and used as received. Field emission scanning electron microscopy (FE-SEM; MIRA3 TESCAN) was used for morphological analyses. In order to identify the surface functional groups of the as-synthesized samples Fourier-transform infrared (FT-IR) spectra was applied by using KBr tablets on a Thermo Nicolet Nexus 870 FT-IR spectrometer in the range of 700-3650 cm-1. Thermogravimetric analyses (TGA) was performed on an Erkin Elmer, Yris diamond from 25 to 600◦C at a heating rate of 10◦C.min-1 and a nitrogen flow of 200 ml.min-1. A furnace atomic absorption instrument (AAS, Thermo Electron Corporation M series) was employed to the determination of metal ion concentration. Energy dispersive X-ray spectroscopy (EDS; SAMx, France) was employed to determine the elemental composition of the precursor and the prepared MWCNTs. The surface area of the adsorbents were measured by nitrogen adsorption at 77 K using a BELJAPAN surface area analyzer (BELSORP mini II) and using a 16-point BET. A pp-201 portable pH meter (GOnDO Electronic Co., Ltd) was employed to adjust the pH of the solutions. A mini centrifuge (model: MSI 8, D.T.A.P Co. Ltd.) was used for separating solid from solution. The IKA RCT basic (Germany) was applied to heating and stirring the solutions.
2.2. Preparation of functional MWCNTs-f
MWCNTs-COOH (250 mg) were suspended in SOCl2 (40 mL) and DMF (1 mL). The mixture at 70°C was stirred for 48 h under reflux. Afterwards, the residual SOCl2 was removed by reduced pressure distillation to produce the Acyl Chloride-functionalized MWCNTs (MWCNTs-COCl). MWCNTs-COCl were mixed with 3-aminopyrazole (400 mg) in DMSO (20 mL) as a solvent and the mixture was stirred at 100°C for 96 h. The mixture was cooled at room temperature for one-half hour. Afterward, solid phase was separated by centrifuging and washed thoroughly with ethanol, THF, and water. Finally, the black solid was dried at room temperature for 8 h under vacuum condition.
2.3. Preparation of metal Stock Solution
It is very important before use of the laboratory glassware, rinsed it with a solution of 2% nitric acid, due to removing all impurities from the glassware and to prevent further adsorption of metal ions on the walls of the glassware. A metal stock solutions were prepared by dissolving salt metals to distilled water. In order to the required metal ions concentrations in the adsorption experiments, the stock solutions were further diluted with deionized water.
2.4. Adsorption experiments
Batch experiments were performed by shaking (250 rpm) 50 ml of a metal ions solutions in 100 ml Erlenmeyer flask for 2 h at room temperature and different pH, adsorbent dosage, and initial ions concentrations. The initial pH of each solution was adjusted by adding 0.1 N NaOH and 0.1 N HCl solution before adsorption. Finally, the mixtures were filtered, and concentration was measured by AAs. The percentage of removal (%) and sorption capacity qe (mg.g-1) was obtained as follows:
(1)
Where C0 and Ce are the initial and final concentrations (mg.l-1) of metal ions in the aqueous solution, respectively, V (L) is the volume of metal ion solution, and m (g) is the weight of sorbent.
2.5. Experimental Design
Central composite design (CCD) is a subset of response surface methodology (RSM) that was used to investigate the effects of variables, including the adsorbent dosage, pH, and initial ions concentration in the randomized fashion. In the CCD each factor has experiments performed at one of five levels, thus allowing one to fit a quadratic model. The CCD is specified via three operations namely: 2n factorial runs, 2n axial runs and six center runs, where n is the factors number [39]. In this study, it divided into 8, 6, and 6 factorial points, axial points, and replicates at the central point, respectively, which gives a total of 20 experiments. For statistical calculations, the variables X were coded as X according to the following equation:
(2)
Experimental parameters, coded variables, and their levels of the Hg(II) and As(III) ions adsorption process by the adsorbents are given in Table 1.
Table 1. Experimental parameters and their levels.
In order to analyze the results of experiments and considering all the linear, linear by linear interaction and square items, the quadratic model was applied to the prediction of each response. This model is as follows:
(3)
Where, R is the adsorption percent; Xi and Xj are independent variables; β0 is the model constant; βi is the linear, βii is the quadratic, and βij is the linear interaction effect of the input factor. Experimental data were evaluated with Design-Expert 8.0.7.1 (trial version) containing ANOVA in order to approximate estimation of the model quality and interaction among variables and response [40]. The quality of the model was examined with a correlation coefficient (R2) and F-test was used to check the statistical significance.
2.6. Desorption studies
Desorption experiments were carried out with metal-loaded to adsorbents in acid solution under sonication for 3 min. To optimize the concentration of the acid, different concentrations (0.1, 0.2, 0.5, 1.0) of HCl were considered. The adsorption (at optimum conditions) and desorption procedures were repeated three times, each time with fresh solutions of Hg(II) and As(III). The metal recovery was calculated by the following equation [41]:
(4)
3. Results and discussion
3.1. Characterization of modified MWCNTs
Fig1. demonstrates the schematic procedure for functionalization of MWCNTs-COOH by 3-aminopyrazole to obtained the modified adsorbent, MWCNTs-f. The product was characterized by FT-IR, FE-SEM, EDS, and TGA.
Fig 1. The modification route of the MWCNTs-COOH with 3-aminopyrazole.
The FT-IR analysis is one of the most significant characterization techniques applied to explain the changes in chemical structures. The FT-IR spectra of MWCNTs-COOH and MWCNTs-f were illustrated to investigate the surface functional groups at each step in the chemical functionalization (Fig2.a). In the spectra of MWCNTs-COOH, the peak around 1715 cm−1 can be assigned to C═O stretching of the carboxylic acid group. Also, peaks at 1083 and 3445 cm−1 can be attributed to C–O and OH stretching vibrations of the carboxylic group, respectively. Also, in the spectra of MWCNTs-f, the appearance of the new peaks at 1615 and 1500-1580 cm-1 can be assigned to the amide carbonyl groups [–C(═O)NH linkage] and C=N or C=C bonds that confirmed the formation of MWCNTs-f [37]. In addition, the peaks at 3300-3500 and 1200-1400 cm-1 can be assigned to the N–H and C-N stretching modes, respectively [42]. Thus, successful modification of MWCNTs-COOH by 3-aminopyrazole was confirmed by FT-IR spectra confirm.
Most of the functional groups in MWCNTs surface were thermally unstable, thus, TGA is the good evidence for the functionalization of MWCNTs. As shown in Fig2.b, the TGA curve of MWCNTs-COOH was stable and hardly decomposed below 600°C (weight loss was less than 11%). In TGA curve of MWCNTs-COOH, the weight loss about 5% below 200°C is observed which can be attributed to the thermal decomposition of carboxyl and hydroxyl groups. In addition, the TGA curve of MWCNTs-f shows two decomposition regions (Fig2.b). The first region at below 175°C may be attributed to the decomposition of hydroxyl groups due to the adsorbed water or humidity on the nanotube and the second region at 175-320°C could correspond to decomposition of 3-aminopyrazole. On the basis, the weight loss of the MWCNTs-COOH and MWCNTs-f at 320°C was calculated 6 and 46%, respectively.
Fig2.c shows the FE-SEM images together with EDS spectra that can supply the helpful information such as chemical composition of MWCNTs before and after functionalization. The EDS spectra of MWCNTs-f as compared to EDS of MWCNTs-COOH shows that nitrogen atoms in addition to carbon and oxygen are detected which can be due to the presence of the attached 3-aminopyrazole to MWCNTs. Also, as can be seen from this figure, the FE-SEM images of MWCNTs-COOH show tangled and agglomerated morphology. Moreover, in the MWCNTs-COOH surface, some swelling was observed which can be due to the covalent attachment of carboxyl groups [43]. The FE-SEM image of MWCNTs-f (Fig2.c) exhibits polymerized beehive like structure and the MWCNTs surface appears to be grown well with increased thickness.
Specific surface area of the adsorbents was determined by the Brunauer–Emmett–Teller (BET) method (acquisition and reduction). The BET surface areas of MWCNTs-COOH and MWCNT-f were 166.2 m2.g-1, and 7.3127 m2.g-1, respectively. Nevertheless, various groups introduced on the surface of MWCNTs-COOH provide numerous sorption sites, thereby enhanced the sorption capacities of MWCNTs-f. Similar observations were reported in the literature [44-47]. Fig2.d shows the average pore diameters (DBJH) of MWCNTs-f (24.78 nm) are evidently much larger than that of the MWCNTs-COOH (1.64 nm). The average pore diameter of 24.78 nm indicates that the MWCNTs-f prepared was in the mesopores region according to the IUPAC classification [48]. Also, Fig2.d revealed the N2 adsorption–desorption isotherms of MWCNTs-COOH and MWCNTs-f obtained at 77 K. The isotherms obtained from this figure could be classified as a type IV isotherm with H1 type hysteresis loops according to the original IUPAC classification [48]. However, according to the extended classification of adsorption isotherms, the isotherms are type IIb isotherms [49].
Fig 2. FT-IR spectra (a), TGA curve (b), EDS spectra and FE-SEM images (c), and N2 adsorption-desorption isotherm and pore size distribution (inset) (d) of functionalized MWCNTs.
3.2. Adsorption Process Modeling by CCD
The optimization of three variables (pH, adsorbent dosage, and initial ions concentration) on the adsorption efficiency of Hg(II) and As(III) ions from aqueous solutions was performed by employing CCD under RSM. In order to explore the mathematical relation between the response (percentage of adsorption) and variables, the quadratic model was chosen. The experimental results are tabulated in Table 2. According to this table, MWCNTs-f demonstrates suitable removal performance which can be due to the functionalized groups on the MWCNTs surface (the q¬¬e was 177.65 and 129.92 mg.g-1 for adsorption of Hg(II) and As(III) ions, respectively).
Table 2. CCD results for Hg(II) and As(III) removal by MWCNTs-COOH and MWCNTs-f as adsorbents.
In addition, the CCD technique and ANOVA were applied to develop the correlation between the selected factors in Table 1 and removal efficiency in this study. The model equations for coded values of the quadratic models for fitting experimental results of Hg(II) and As(III) removal by MWCNTs-COOH and MWCNTs-f, were expressed as:
(5)
(6)
(7)
(8)
Where X1, X2, and X3 are the coded values of pH, adsorbent dosage, and initial ions concentration, respectively and Y is log R. The sign of regression coefficient has a direct effect on the response, so it is important. The negative and positive signs before the terms demonstrate the antagonistic and synergistic effects of the variables [50]. On the one hand, two independent variables (pH and adsorbent dose) exhibited the positive significant effect on Hg(II) and As(III) removal for the two adsorbents. Also, the initial ions concentration in this study and all the quadratic terms showed the reducing effects for metals adsorption.
Table 3 show the statistical significance of the quadratic models evaluated with ANOVA. According to this table, the linear terms of variables (pH, adsorbent dosage, and initial ions concentration) and the quadratic terms of pH are statistically significant (p value<0.05) in two adsorbents. Besides, adjusted-R2 more than 0.9 indicates a desirable for models validation. The prediction-R2 is in reasonable agreement with the adjusted-R2 (Table 3). The coefficient of variation (C.V.%) is also less than 10%.
Table 3. ANOVA for the response surface quadratic model for adsorption of Hg(II) and As(III) by MWCNTs-COOH and MWCNTs-f.
The data obtained from the adsorption experiments and the values predicted by the response surface model for the MWCNTs-COOH and MWCNTs-f adsorbents are shown in Fig3. in Online Resource 1. The values of the R2 coefficient greater than 0.95 indicate a good correlation between the experimental data and predicted ones. Also, the second-order quadratic models for the observed results are significant because the p values of the models are small.
In order to evaluation of the effect of each factor on response parameter (removal %), the percent contribution was measured and tabulated in Table 4 in Online Resource 2 [51]. According to this table, more than 44 and 17% contribution in the case of Hg adsorption and 31 and 44% contribution in the case of As(III) adsorption by MWCNTs-COOH and MWCNTs-f, were obtained for individual parameters, respectively. But less than 1% of the contribution was for combine interaction in all cases. Among the parameters, pH has the largest contribution (in the case of Hg(II) adsorption: 38.9 and 13.24%, in the case of As(III) adsorption: 22.44 and 38.44% by MWCNTs-COOH and MWCNTs-f, respectively). Therefore, in these experiments, pH has effective roles in the adsorption process.
The perturbation plot (Fig4. in Online Resource 3) shows the comparative effects of all independent variables on the removal efficiency. The sharp curvature of the A factor (pH) in all cases shows that the response, heavy metals removal efficiency, was very sensitive to this variable. Also, the maximum removal efficiency occurred as the adsorbent dose reached the maximum level. Furthermore, with increasing metal ions concentration, the response achieved to the minimum level.
3.3. Response Surface Plotting
3.3.1. Effect of pH and initial ions concentration on adsorption process
Fig5.a is 3D response surface plot displaying the interactive effect of pH and metal ions concentration for Hg(II) and As(III) removal. This figure revealed that the adsorption process extremely depends on the solution pH in all cases. The optimum pH for Hg(II) and As(III) removal by the adsorbents was found to be in the range of 6.0-8.0. At pH below 6.0, the removal efficiency significantly increased with the pH growth. Because, the formation of mercuric hydroxide was increased, thus the sorption of Hg(OH)2 on the surface of adsorbents increased [52]. Besides, at pH less than 6.0, there is the strong competition between H+ and As(III) on the adsorption sites [53]. On the other hand, at pH higher than 8.0, the As(III) removal efficiency dropped due to the ionization of H3AsO3, that causes more competition between arsenite and OH- anions [54]. Also, at this pH, the Hg(II) removal efficiency reduced with increasing pH. It could be due to the complex formation of Hg(OH)3- and Hg(OH)4- in the solution that these complex could not be retained on the surface of the adsorbents because they are soluble in water [55]. Also, Fig5.a reveals that with decreasing the initial ions concentration from 60 to 10 ppm, the removal% was increased to the maximum amount at the constant adsorbent dose (20 mg). This sorption characteristic demonstrated that the saturation of surface was related to the initial ions concentrations. The initial ions concentration provides a significant driving force to overcome all mass transfer resistance of metal ion between the solid phases and aqueous [56].
3.3.2. Effect of pH and adsorbent dose on Hg(II) and As(III) adsorption
The 3D plots for the combined effects of adsorbent dose and pH on the Hg(II) and As(III) uptake by the MWCNTs-COOH and MWCNTs-f are represented in Fig5.b. Depending on the type of adsorbent and functional groups, the pH of the solution has a different effect on adsorption. Fig5.b represents that the metals adsorption extremely depends on the initial pH of the solution. Under acidic condition (pH<4.0), both adsorbents and metal ions are positively charged, so there were unfavorable electrostatic interactions as well as competition between H+ and metal cations for adsorption on available surface sites [57]. Thus, adsorption of Hg(II) and As(III) by the adsorbents was reduced at low pH. Also, at pH higher than 9.0, the Hg ions precipitation occurred that caused absorption efficiency decreased [58]. Also, at this pH, the fraction of the total As(III) as H2AsO3- will be decreased [59]. In addition, it could be seen that the adsorption rate smoothly increase when adsorbent dose increases from 10 to 20 mg and reached a maximum of 20 mg for the two adsorbents. It could be explained by easier penetration of metal ions to active sites due to increasing in the active sites in the higher amounts of adsorbents [60]. Similar observations are reported in the literature [55, 61, 62].
Fig 5. The 3D plots are showing the interactive effect of pH and initial ions concentration (a) and the interactive effect of pH and adsorbent dose (b). (1) Hg(II) removal by MWCNTs-COOH, (2) Hg(II) removal by MWCNTs-f, (3) As(III) removal by MWCNTs-COOH, and (4) As(III) removal by MWCNTs-f.
3.4. Process optimization
The optimization of Hg(II) and As(III) removal process by the adsorbents was carried out by using the Design Expert 8.0.1 software (trial version). In the optimization analysis, the goal criterion was set as maximum values for the responses (removal %). The optimum adsorption conditions achieved are presented in Table 5, with the desirability of one. This table reveals that the models supported the experimental data with a relatively small error less than 1.0%. In order to decrease the experimental errors for the evaluation of the isotherm models, kinetic parameters, and thermodynamic properties, the optimum conditions were chosen as Table 6 in Online Resource 4.
Table 5. The optimum adsorption conditions and models validation.
3.5. Adsorption isotherms
The equilibrium characteristics of adsorption were described through Langmuir, Freundlich, and Temkin isotherm models. As known, Langmuir model assumes a monolayer adsorption onto a surface containing the finite number of adsorption sites, specific homogeneous sites and finally, all sites are identical and energetically equivalent [63]. The linear form of Langmuir model is given as:
(9)
Where qmax (mg.g-1) is the maximum amount of sorbed metal per unit mass of adsorbent and KL (l.mg-1) is the Langmuir constants related to the adsorption heat. The dimensionless separation factor, RL, is the essential characteristics of Langmuir model. Its linear form is given by the following equation:
(10)
Where C0 (mg.l-1) is the initial concentration of adsorbate in solution. The RL values reveal the type of isotherm to be irreversible (RL = 0) , favorable (0 < RL < 1), , linear (RL = 1) or unfavorable (RL > 1) [64].
The Freundlich model widely uses to explain the adsorption characteristics of the heterogeneous surface and reversible adsorption. The linear form of this empirical model is given as [64, 65]:
(11)
Where Kf (mg.g-1) is related to the sorption capacity of adsorbent and n is related to the intensity of sorption and varies with surface heterogeneity and affinity. The n-1 value lies between 0 and 1 demonstrates an amount of adsorption intensity or heterogeneity of the surface. If n-1 values were near to the zero, heterogeneity of the surface was increased. Also, the n-1 value less than one suggest chemisorption process, instead, the value above unity imply the cooperative adsorption [64, 66].
Moreover, Temkin isotherm model was considered to the evaluation of adsorption isotherm. It assumes that heat of adsorption of all molecules in the layer would decrease linearly rather than logarithmic with coverage and uniform distribution of binding energy between adsorbent and adsorbate [67] and its linear form is given as:
(12)
(13)
Where B (J.mol-1) is constant related to the heat of adsorption, b is the Temkin isotherm constant, and KT is the equilibrium binding constant (l.g-1) [64, 68, 69].
The slope and the intercept of each linear plot of the isotherms are used to calculate the isotherm model parameters and listed in Table 7.
The R2 values in Table 7 and Fig6. in Online Resource 5 revealed that experimental data in all cases were fitted well with Langmuir equation in the studied temperature range (R2>0.92). This could be indicated the monolayer adsorption onto the surface of the adsorbents [70].
According to the Table 7, the KL values increased with increasing temperature from 25 to 45°C that revealed favorable metal ions adsorption would be expected at the higher temperature. This represents that the adsorption process by the adsorbents was endothermic [71]. However, the qmax values decreased with increasing solution temperature. Therefore, less adsorption of Hg(II) and As(III) ions would be expected at low temperature. The RL values were less than one (0< RL <1) that revealed Langmuir isotherm was favorable for adsorption of Hg(II) and As(III) on the adsorbents.
In Freundlich isotherm, the values of KF which calculated for MWCNTs-f, were larger than MWCNTs-COOH indicating that MWCNTs-f has the high affinity toward heavy metal ions. Also, deviation of the n-1 values from unity demonstrates a non-linear adsorption that takes place on heterogeneous surfaces. Furthermore, the n-1 values of MWCNTs-f were lower than MWCNTs-COOH denoting that MWCNTs-f surface was more heterogeneous than MWCNTs-COOH [72].
The values of b in Table 7 express the heat of metal adsorption by the adsorbents. The b values increased with increasing solution temperature from 25 to 45°C, indicating endothermic sorption. Also, it has been reported that the b values is less than 8.0 kJ.mol-1 which denoting a physical adsorption process. Thus, the low values in this study indicate a weak interaction between sorbate and sorbent, supporting a physisorption mechanism for this study [73].
Table 7. Adsorption isotherm constant values of Hg(II) and As(III) adsorption by MWCNTs-COOH.
3.6. Adsorption kinetics
Different models can be applied to analyze the kinetics of sorption process. In this work pseudo-first-order, pseudo-second-order, and Elovich kinetic models were applied to kinetic studies. Pseudo-first-order proposed a rate equation for the sorption of adsorbate from a liquid solution. This equation is as follows:
(14)
Integrating from t=0 to t=t and q=0 to q=q gives:
(15)
Where q and qe are the values of amount adsorbed per unit mass of adsorbent (mg.g-1) at any time (min) and at equilibrium, respectively, and k1 is the pseudo-first-order rate constant (min-1) [74, 75]. In this model by plotting the values of log (qe-q) vs. t qe and k1 values can be determined from the intercept and slope, respectively.
Another model for the analysis of sorption kinetic is pseudo-second-order and the rates law is given as:
(16)
Integrating from t=0 to t=t and q=0 to q=q gives:
(17)
(18)
Where k2 (g.mg−1.min−1) is the rate constant of the pseudo-second-order adsorption and h is the initial sorption rate (mg.g−1.min−1). Moreover, equation (17) can be rearranged to achieve a linear form [76, 77]:
(19)
In this model by plotting the values of t/q vs. t, it can be determined the qe and k2 values from the slope and intercept, respectively.
In recent years, the Elovich model has been used to describe the pollutants adsorption from aqueous solutions. This semi-empirical model usually employed to describing the chemisorption process assuming that the adsorbent surfaces are energetically heterogeneous. However, this model does not recommend any definite mechanism for adsorbent-adsorbate. The linear form of this model is as follow [78]:
(20)
Where α is initial sorption rate (mg.g−1.min−1) and β is desorption constant (g.mg−1) during any one experiment. The plot of qt vs. lnt was used to obtain the β and α constant.
In order to study the equilibrium time and adsorption rates, experiments were carried out at optimum conditions at different time intervals up to 300 min. The calculated design parameters (k1, k2, qe, α, β) and correlation coefficient (R2) could be obtained by the slope and intercept of each linear plot as given in Table 8 in Online Resource 6. Also, the equilibrium times were found to be 100 min for the adsorption of Hg(II) and As(III) ions by MWCNTs-COOH and MWCNTs-f. As can be seen from Table 8, the R2of the second order rate equation is more than 0.998 for all cases. Besides, the calculated qe from the second order rate equation for MWCNTs-COOH and MWCNTs-f were very close to the experimental values with the mean absolute relative error less than 4 percent (Table 8). So, the adsorption of mercury and arsenic ions by MWCNTs-COOH and MWCNTs-f can be well displayed by the pseudo-second-order kinetic model. Also, this model suggested that the rate controlling step is chemisorption in the adsorption process. Our findings are in good agreements with those reported in the literature [52, 53, 55].
3.7. Thermodynamic studies
In order to the evaluation of thermodynamic properties such as free energy change (∆G°), enthalpy change (∆H°), and entropy change (∆S°), the following equations were considered:
(21)
(22)
(23)
Where C0 and Ce (mg.l−1) are the initial and equilibrium metal ions concentration in the solution, respectively. Kd is the equilibrium constant, T is temperature of the solution (K) and R is the universal gas constant (8.314, J.K−1.mol−1) [79]. The values of ∆H° and ∆S° could be determined from slope and intercept of the plotting of ln kd vs. 1/T. The calculated thermodynamic parameters are given in Table 9.
Table 9. Thermodynamic parameters for the adsorption of Hg(II) and As(III) on the adsorbents at optimum conditions.
In thermodynamics, ∆G° is used to the evaluation of the spontaneity of the adsorption process; the more energetically desirable process has the higher negative ∆G° value. The negative values of ∆G° in Table 9 indicate that adsorption of Hg(II) and As(III) are the spontaneous processes. Further, it is considerable that the ∆G° values for mercury and arsenic adsorption decrease with increasing temperature from 25 to 45°C, denoting that the process was more efficient at the higher temperature [80]. Also, the ∆G° values, less than −15 kJ.mol−1, are regarded with the physical interactions between adsorbent sites and metal ions [65]. However, ∆H° and ∆S° values are positive, it represents that adsorption process is endothermic and randomness increases at the solid-solution interface within the adsorption [80]. Similar results were reported in the literature for the adsorption of Hg(II) ions from aqueous solution by MnO2/CNTs [81] and adsorption of As(III) ions from aqueous solution by iron oxide impregnated activated alumina [61].
3.8. Desorption study
Renewability is one of the important factors to the applicability of adsorbents. To investigation of this property, desorption study was carried out. According to the Fig7. in Online Resource 7, with increasing the concentration of HCl, the metal-loaded to the adsorbents were more desorbed. Further, the Hg(II) and As(III) removal efficiency reduced about 10 and 6% during consecutive cycles from 1 to 3 for MWCNTs-COOH and MWCNTs-f, respectively. The renewability of the adsorbents after 3 cycles in this work, is comparable to the literature [55, 58, 82-84].
3.9. Comparison of adsorbent performance with literature
The comparison between the performances of MWCNTs-f for the adsorption of Hg(II) and As(III) ions with different adsorbents are presented in Table 10. Due to the different experimental conditions that were employed in those obtained in literature, the direct comparison was difficult. However, the qmax values were widely different for various adsorbents. Ordinarily, functionalized MWCNTs were exhibited significant higher adsorption capacities than the pristine MWCNTs as well as activated carbon. Also, MWCNTs-f shows comparable adsorption capacity with other functionalize MWCNTs adsorbents that were used yet to mercury and arsenic adsorption. However, it was the deduction that MWCNTs-f has an applicability to the adsorption of As(III) in compared with other adsorbents that were used previously. Eventually, this comparison indicates that the MWCNTs-f exhibited an acceptable capacity for Hg(II) and As(III) adsorption from aqueous solutions.
Table 10. Comparison of adsorption capacities of various adsorbents for Hg(II) and As(III) ions adsorption.
4. Conclusions
In this work, the new functionalized MWCNTs was successfully synthesized and used for mercury and arsenic ions adsorption from aqueous solutions. FT-IR, TGA, EDS and FE-SEM analysis results confirmed the chemical modification process. Response surface methodology was applied for the evaluation of optimum values of the effective factors such as pH, adsorbent dose, and initial ions concentration. For correlating the experimental data, the quadratic model was applied. This model and ANOVA analysis results confirmed that the pH has the major effect on the adsorption process and the optimum conditions for adsorption of Hg(II) and As(III) by MWCNTs-COOH and MWCNTs-f were evaluated. Furthermore, the experimental data were in good agreement with the data predicted by the models with relatively small errors. The maximum removal efficiency of Hg(II) and As(III) under the optimum conditions were achieved as 80.5 and 72.4% by MWCNTs-f, respectively. Between the isotherms models, equilibrium data fitted well by the Langmuir isotherm for the mercury and arsenic adsorption. Also, the kinetic adsorption could be well explained by the pseudo-second-order kinetic equation. In addition, thermodynamic properties, ∆G°, ∆H°, and ∆S° were determined and exhibited that the adsorption processes were more efficient at the higher temperature and were endothermic as well as increasing in randomness. The adsorption–desorption experiments showed a small loss in the adsorption capacity of 6% after three cycles, demonstrating a good regeneration capacity of MWCNTs-f for treating wastewater containing mercury and arsenic ions. Comparison of MWCNTs-f adsorption capacity with other adsorbents in the literature revealed that there are the great potentials and applications of the new modified adsorbent of this work (MWCNTs-f) for adsorbing Hg(II) and As(III) ions from aqueous solutions.
Acknowledgements
The financial and encouragement support provided by the Research Vice Presidency of Semnan University, (www.semnan.ac.ir), zip code: 35131-19111, Tel/Fax: +98 2333654120, Semnan, Iran and Ayatollah Amoli Branch, Islamic Azad University, zip code: 46351-43358, Tel.: +98(11) 43217000-3, Fax:+98(11)43217041, info@iauamol.ac.ir, Amol, Iran.