Research Article  Open Access
Vo Thi Thanh Chau, Huynh Thi MinhThanh, Pham Dinh Du, Tran Thanh Tam Toan, Tran Ngoc Tuyen, Tran Xuan Mau, Dinh Quang Khieu, "MetalOrganic Framework101 (MIL101): Synthesis, Kinetics, Thermodynamics, and Equilibrium Isotherms of Remazol Deep Black RGB Adsorption", Journal of Chemistry, vol. 2018, Article ID 8616921, 14 pages, 2018. https://doi.org/10.1155/2018/8616921
MetalOrganic Framework101 (MIL101): Synthesis, Kinetics, Thermodynamics, and Equilibrium Isotherms of Remazol Deep Black RGB Adsorption
Abstract
In the present paper, the synthesis of metalorganic framework101 (MIL101) and Remazol Deep Black RGB (RDB) adsorption on MIL101 were demonstrated. The kinetics of RDB adsorption on MIL101 was studied using Weber’s intraparticle diffusion model and the pseudofirst and pseudosecondorder kinetic models. Particularly, the statistical method of piecewise linear regression and multinonlinear regression was employed to analyse the adsorption data according to the previously mentioned kinetic models. The results indicated that the adsorption process followed the threestep pseudofirstorder kinetic equation, which was consistent with the results of the intraparticle diffusion model with three linear segments. This model best described the experimental data. In addition, the adsorption isotherm data were studied using five adsorption models, namely, Langmuir, Freundlich, Redlich–Peterson, Toth, and Sips in nonlinear forms, and the Langmuir model is the most appropriate for the experimental data. The values of energies of activation of adsorption were calculated, and they revealed that the adsorption process was of endothermic chemical nature. A statistical comparison using Akaike information criterion to estimate the goodness of fit of the kinetic and isotherm models was presented.
1. Introduction
It is well known that wastewater from textile industries, pulp mills, and dyestuff manufacturing has been a potential threat to environment [1, 2]. Various treatment processes such as physical separation, chemical oxidation, and biological degradation have been widely investigated to remove dyes from wastewaters [3]. Among these processes, adsorption technology is considered as one of the most competitive methods because it does not require high operating temperature and has a simple operation as well as low cost, and several coloring materials can be removed simultaneously [4].
MIL101 has demonstrated good performance in storage and adsorption of gas such as hydrogen storage [5], adsorption of CO_{2} and CH_{4} [6], and longchain alkanes [7]. However, there are very few reports studying the adsorption of dyes from aqueous solutions [8–10]. MOFs in general and MIL101 in particular exhibit high efficiency for adsorption of dyes from aqueous solutions due to their unique structures such as large surface areas, ordered porosity, and high density of adsorption sites (anion and cation). MIL101 demonstrates excellent adsorption properties for dyes, such as methyl orange (MO), xylenol orange (XO), and uranine. For expanding applications of MIL101, studying the adsorption of this material for Remazol Deep Black RGB (denoted as RDB) used widely in dye industry will be a concern of this work.
Azo dyes with an azo group are widely used in many textile industries due to their low cost, high solubility, and stability. These dyes and their intermediate products are toxic, carcinogenic, and mutagenic to aquatic life. Remazol Deep Black RGB is a common diazo reactive dye, widely used in textile industries [11]. It is stable and hard to degrade biologically due to the presence of aromatic rings. Thus, RDB removal from textile wastewater has drawn much attention among researchers. Several techniques including adsorption, electrochemistry, and biosorption for RDB treatment have been reported. Soloman et al. [11] applied the electrochemical treatment to hydrolyze Remazol Black. They demonstrated that the performance of the batch recirculation system was better than other reactor configurations studied in terms of capacity utilization and energy consumption. Brazilian pinefruit shells (Araucaria angustifolia) in their natural form were used as an adsorbent for the removal of RDB dye from aqueous effluents [12]. A biosorption process to discard azo dyes by fungi (Aspergillus flavus) was investigated in batch reactors [13]. Ninety percent of the dye in a 100 mg/L solution was removed. Recently, Thanh et al. [14] reported that iron doping to ZIF8 significantly enhances RDB adsorption capacity. Fe–ZIF8 also exhibits photocatalytic degradation of RDB under visible light [15].
Batch adsorption studies focus on two main trends: (i) designing and optimizing experiments with the evaluation of the influence of the experimental variables—this approach enables to estimate the magnitude of the influence of the factors affecting the process and their interactions [16]—and (ii) kinetics, thermodynamics, and equilibrium isotherm adsorption studies [14, 17]. For the latter, several models are used to study adsorption kinetics and isotherms. The parameters in these models are calculated with linear or nonlinear regression approaches. However, the number of parameters in each mode is different. For example, the Langmuir isotherm model contains two parameters, while the Redlich–Peterson isotherm model has three parameters. It is obvious that the greater the number of model parameters, the lower the relative errors (REs) or the sum of squared errors (SSEs). Therefore, the model compatibility needs to be evaluated including SSEs or REs and the number of model parameters as well as the experimental points. However, in the majority of current publications, the goodness of fit for models is estimated based on only the REs or SSEs. To the best of our knowledge, the research on this issue is limited.
In the present study, MIL101 was employed as an adsorbent for removing RDB dye. The effects of initial concentration, adsorbent particle size, agitation speed, temperature, and pH on the adsorption behavior of RDB onto MIL101 were investigated. The adsorption kinetic and isothermal studies and the goodness of fit for models were addressed.
2. Experimental
2.1. Materials
Chromium (III) nitrate nonahydrates (Cr(NO_{3})_{3}·9H_{2}O, Merck, Germany), terephthalic acid, (C_{6}H_{4}(COOH)_{2}, Merck, Germany) (H_{2}BDC), and hydrofluoric acid (HF, 40%, Merck, Germany) were utilized in this study. Remazol Deep Black RGB (C_{26}H_{21}N_{5}Na_{4}O_{19}S_{6}), (molecular weight = 991.82) was supplied by Thuy Duong Textile Company, Vietnam. The molecular structure of RDB is shown in Scheme 1.
2.2. Apparatus
The powder Xray diffraction (XRD) pattern was recorded by means of a D8 Advance (Bruker, Germany) with CuKα radiation (λ = 1.5406 Å). The morphology of the obtained samples was determined using transmission electron microscope (TEM) on JEOL JEM2100F. The specific surface area of the samples was determined by means of nitrogen adsorption/desorption isotherms using a Micromeritics 2020 volumetric adsorption analyzer system. Visible spectrophotometry was measured by using Lambda 25 Spectrophotometer (PerkinElmer, Singapore) at λ_{max} of RDB dye (600 nm).
2.3. Preparation of MIL101
MIL101 was synthesized from chromium (III) nitrate nonahydrates, H_{2}BDC, and HF using the hydrothermal method [18]. Three samples of MIL101 with different molar ratios of HF/H_{2}BDC = 0.00, 0.25, and 0.75 were prepared. In a typical procedure, a mixture of 10 mmol of H_{2}BDC, 12.5 mmol of Cr(NO_{3})_{3}·9H_{2}O, x mmol of HF (x = 0.00, 0.25, and 0.75), and 350 mmol of H_{2}O was heated in a Teflonlined stainless steel autoclave at 200°C for 8 h. The resulting green solid was filtered and then washed with ethanol in a Soxhlet apparatus for around 12 h to completely remove the unreacted amount of H_{2}BDC. The obtained MIL101 was denoted as MHF0 for the molar ratio of HF/H_{2}BDC = 0, MHF0.25 for 0.25, and MHF0.75 for 0.75.
2.4. Adsorption Kinetic and Thermodynamic Studies
Kinetic studies were carried out in a 3 L plastic beaker. This beaker was equipped with a stainless steel flatblade impeller using an electric motor to stir the dye solution and a tap near the bottom to withdraw the liquid at any time. MIL101 (0.50 g) was mixed thoroughly with 1000 mL of the dye solution in the beaker at room temperature. 10 mL of the mixture was withdrawn periodically, and MIL101 was removed by using a centrifuge. The final dye concentration was determined using the spectrophotometric method. The adsorption capacity of the adsorbent was calculated according to the following formula:where is the adsorption capacity (mg·g^{−1}) at time, is the initial dye concentration (mg·L^{−1}), is the dye concentration (mg·L^{−1}) at time , is the volume of dye solution (L), and is the mass of the adsorbent (g).
For the adsorption isotherm study, an amount of 30, 40, 50, 60, 70, 80, 90, and 110 mg of MIL101 was added to 8 stopper 250 mL Erlenmeyer flasks containing 100 mL of 100 mg/L RDB solution. The flasks were then placed into a shaker bath at 28 ± 1°C for 24 hours. Thereafter, the supernatant liquid was separated by centrifugation and the final dye concentration was determined with the method mentioned above.
In order to study formal and diffusion kinetics, Weber’s intraparticle diffusion model and pseudofirst and pseudosecondorder kinetic models were used. Weber’s intraparticle diffusion model is described as in the following equation [19]:where is intraparticle diffusion rate constant (mg·g^{−1}·min^{−0.5}) and is the intercept which represents the thickness of the boundary layer.
If intraparticle diffusion is the ratelimiting step, then the plot of q versus t^{0.5} will give a straight line with a slope that equals k_{i} and an intercept equal to zero.
The pseudofirst and pseudosecondorder kinetic models in the nonlinear form are expressed as follows [20]:where and are the adsorption capacity at time (min) and at equilibrium time, respectively, is the rate constants of the pseudofirstorder model (min^{−1}), and is the rate constant of the pseudosecondorder kinetic model (g·mg^{−1}·min^{−1}).
For the thermodynamic study, the experiments were conducted in the same way as in the adsorption kinetics study. However, the adsorption temperature was fixed at 298, 308, 318, and 328 K. The activation energy, , was determined using the Arrhenius equation [21]:where is the rate constant, is the frequency factor, is the gas constant (8.314J mol^{−1}·K^{−1}), and is the absolute temperature in Kelvin. Taking the natural logarithm of both sides of (4), one obtains
By linearly plotting versus 1/T, one could obtain from the slope ().
In order to evaluate whether the adsorption process is spontaneous, the adsorption thermodynamic parameters are needed. The standard Gibbs free energy of adsorption () is given by the following expression:where , , and are the change of standard Gibbs energy, enthalpy, and entropy, respectively.
is given by the van’t Hoff equation:where is the distribution coefficient of the solute ions and equal to [22, 23] and the others are described earlier.
By replacing (6) to (7), one obtains
The value of and is calculated from the slope and intercept of the linear plot of ln versus .
2.5. Adsorption Isotherm Study
Experimental data were analysed according to five isotherm models by Langmuir, Freundlich, Redlich–Peterson, Sips, and Toth.
Langmuir isotherm: Langmuir model is valid for monolayer sorption onto the surface. It could be expressed as follows [24]:where is the maximum monolayer capacity amount (mg·g^{−1}), is the equilibrium constant, is the equilibrium adsorption capacity (mg·g^{−1}), and is the equilibrium concentration of adsorbate (mg·L^{−1}).
Freundlich isotherm: The Freundlich equation is an empirical relation based on the sorption onto a heterogeneous surface. It is commonly represented as [25]where ((mg·g^{−1})·(mg·L^{−1})^{n}) and n are the Freundlich parameters related to adsorption capacity and adsorption intensity, respectively.
The maximum capacity, (mg·g^{−1}), can be calculated from the following equation [26]:where is the initial concentration.
Redlich–Peterson isotherm: Redlich–Peterson isotherm [27] contains three parameters and involves the features of both Langmuir and Freundlich isotherms. It can be described as follows:where (L·g^{−1}) and (L·mg^{−1}) are the Redlich–Peterson isotherm constants and is the exponent which lies between 0 and 1.
When approaches 1, (12) becomes the Langmuir equation. Then, the maximum adsorption capacity, (mg·g^{−1}), can be determined by the following equation:
Sips isotherm: Sips isotherm is a combination of the Langmuir and Freundlich isotherms and expected to describe heterogeneous surfaces much better [24]. The model can be written as [28]where and are the Sips constants related to the energy of adsorption. The maximum monolayer adsorption capacity could is given by .
Toth isotherm: Toth isotherm is the Langmuirbased isotherm and considers a continuous distribution of site affinities. It is expressed as [29]where is the Toth model constant and is the Toth model exponent (). It is obvious that, for , this isotherm reduces to the Langmuir equation.
2.6. Piecewise Linear Regression and Model Comparison
The application of Weber’s model often suffers from uncertainties caused by the multilinear nature of its plots. Malash and ElKhaiary [30] proposed the piecewise linear regression for the analysis of multilinearity in intraparticle diffusion and film diffusion. In this method, the experimental data could be fixed for one, two, three, or fourlinearsegment lines: onelinearsegment line: Y = B + A · X (two parameters), twolinearsegment line: Y = B + A · X + C · (X − D)·SIGN (X − D) (four parameters), threelinearsegment line: Y = B + A · X + C · (X − D)·SIGN (X − D) + E · (X − F) · SIGN (X − F) (six parameters), fourlinearsegment line: Y = B + A · X + C · (X − D)·SIGN (X − D) + E · (X − F) · SIGN (X − F) + G · (X − H)·SIGN (X − H) (eight parameters),where the values of A, B, C, D, E, F, G, and H are estimated by nonlinear regression. D, F, and H, called breakpoints, are the boundaries between the segments. The Microsoft Excel “SIGN” function is defined as follows:
The example for the threelinearsegment equation is expressed as follows:
Then, the linear equation of the first segment is y = a_{1} · x + b_{1}, where b_{1} = B − C − E and a_{1} = A + C · D + E · F. The linear equation of the second segment is y = b_{2} · x + a_{2}, where b_{2} = B + C – E and a_{2} = A + C · D + E · F. The linear equation of the third segment is y = b_{3} · x + a_{3}, where b_{3} = B + C + E and a_{3} = A − C · D − E · F.
The model’s parameters are determined using the least squares method. This is calculated by minimizing the sum of squared errors, SSE_{S}, by numerical optimization techniques using the Solver function in Microsoft Excel. The function for minimization iswhere is experimental datum and is the value estimated from the model.
The determination coefficient, , is obtained by the following expression:where is the total sum of squares equal to ( is the mean value of ).
We know that increasing the number of linear segments increases the number of regression parameters that almost universally led to the decrease of SSEs or the increase of . Therefore, the model compatibility cannot be based only on SSE or but must also include the number of regression parameters as well as the experimental points. Akaike’s information criterion (AIC) is one of the wellknown statistical methods used in this case.where is the parameter of the model. The other parameters are described above.
The is applied as is small compared with . is only computed as is at least two units greater than . The value of could be positive or negative, and the lower the value, the better it is. Another way of comparing is using the evidence ratio (ER) which is expressed as follows [17]:where Δ is the absolute value of the difference between and scores. ER means that the model with lower is 1/e^{−0.5Δ} times more likely to be correct than the alternative model.
In the present study, the comparison of the model will use or SSE if the models possess the same experimental points (N) and parameter numbers of models (). Otherwise, will be employed.
3. Results and Discussion
3.1. Characterization of MIL101 Samples
Figure 1 shows the XRD patterns of MIL101 synthesized with the HF/H_{2}BDC molar ratios of 0.00, 0.25, and 0.75. The characteristic diffractions of the samples matched well with the published XRD patterns of MIL101 [18]. This means that the obtained materials are MIL101. However, the peak intensity of the samples synthesized with HF is significantly higher than that of the sample synthesized without HF (MHF0). This could be due to fluorine that acts as a mineralizing agent in the hydrothermal synthesis for the formation of well crystalline microporous materials [31].
The morphology of the obtained material consists of octahedronshaped crystals with smooth facets (Figure 2). The particle size of MIL101 increases with the increase in the HF/H_{2}BDC ratio. The average particle size counted based on 50 particles is 234 nm for MHF0, 364 nm for MHF0.25, and 612 nm for MHF0.75 (Table 1).
(a)
(b)
(c)

The textural properties of the MIL101 samples were investigated by using nitrogen adsorption/desorption isotherms. The isotherm curves belong to type IV according to IUPAC classification (Figure 3). The BET specific surface area for MHF0.25 is the highest (Table 1), and thus, MIL101 synthesized with HF/H_{2}BDC = 0.25 was chosen for adsorption experiments.
3.2. RDB Adsorption on MIL101
3.2.1. Effect of Initial Concentrations
The RDB kinetics of adsorption on MIL101 at different initial concentrations in the range of 25–600 ppm is illustrated in Figure 4(a). It is obvious that the adsorption capacity of RDB on MIL101 increases when RDB initial concentration increases from 25 to 400 ppm. This might be due to the fact that, initially, the sites on the adsorbent surface are less occupied by the dye molecules, and increasing the concentration increases the interaction between the dye molecules and the adsorbent; thus, more dye molecules adsorb on the surface [10, 32, 33]. In addition, the mass transfer driving force becomes larger as the initial concentration increases, and this results in higher adsorption capacity [21]. At higher concentrations (>400 ppm), the adsorption takes place very fast during the first 15 minutes, and then, it slows down and reaches equilibrium at about 150 minutes. Meanwhile, for concentration at 25 ppm, the adsorption reaches equilibrium practically immediately, just after about 10 minutes. This might be because at low dye concentration, the driving force is very small and the adsorption takes place only on the surface of MIL101, whereas when the dye concentration is high with large driving force, the adsorption also occurs in the pore of the adsorbent, and due to high resistance in the pores, the adsorption becomes slower and reaches equilibrium after a longer period. Furthermore, adsorption might take place in several stages. At a very high dye concentration (600 ppm), the RDB adsorption on MIL101 does not follow the same pattern as at lower initial dye concentration (Figure 4). This might be the result of forming a colloidal solution at a high concentration of RDB [20, 34].
(a)
(b)
Therefore, the RDB initial concentration from 50 to 400 ppm is suitable for the kinetic study of adsorption on MIL101. In this kinetic study, Weber’s intraparticle diffusion model [19] was applied to study the adsorption mechanism. The values of q_{t} at different times were analysed using piecewise linear regressions based on the assumption of one, two, three, and four linear segments. The AIC_{C} value was the criterion for determining which one is the goodness of fit (Table 2).

The data indicate that the threelinearsegment model has the lowest AIC_{C}, and therefore, this Weber’s model is the most accurate because for this criterion the lower the value, the more suitable the model. Three distinct steps can be seen on the kinetic curves: (i) instantaneous adsorption of RDB molecules within the first 7–21 minutes, (ii) a gradual attainment of the equilibrium due to the utilization of the all active sites on the adsorbent surface, and (iii) an equilibrium attainment of RDB molecules onto MIL101 (Figure 4(b)). At the initial concentration of 50 mg·L^{−1}, for example, the intercept of the first linear segment is 0.05, and its 99% confidence interval is (−3.14; 3.23), indicating that the intercept is not significantly different from zero. This strongly suggests that intraparticle diffusion is the ratecontrolling mechanism during the first 15 minutes of adsorption. The next two linear segments do not pass through the origin because the 99% confidence intervals of their intersects do not contain zero, indicating that the intraparticle diffusion is not the only ratelimiting step and film diffusion or chemical reaction might take place during these periods of adsorption [20, 33]. This behavior is found for all concentrations. The intraparticle parameters are illustrated in Table 3.

The data indicate that the film thickness (intercept 2 and intercept 3) increases with the increase in initial RDB concentration. This suggests that film diffusion controls the adsorption process in the last two steps. In the first step, the intraparticle diffusion parameter, k_{p1}, increases as initial RDB concentration increases. These results suggest that intraparticle diffusion controls the rate in the initial step of the adsorption process.
In addition, the nature of the ratelimiting step is also confirmed by plotting the intraparticle diffusion constant, k_{p1}, versus the first power of the initial concentration. If k_{p1} is proportional to the initial dye concentration, the adsorption process is controlled by film diffusion, whereas, if intraparticle diffusion limits the adsorption process, the relationship between dye concentrations and k_{p1} will not be linear [20, 35]. For the adsorption of RDB onto MIL101, the plot of k_{p1} versus the initial concentration (C_{0}) is not linear (R^{2} = 0.875; ), confirming that the intraparticle diffusion mechanism controlled the adsorption in the initial step.
To study the formal kinetics of the RDB adsorption on MIL101, the experimental data were subjected to pseudofirstorder and pseudosecondorder kinetics in the nonlinear form. The results are shown in Figure 5(a) and Table 4.
(a)
(b)

The experimental points in Figure 5(a) are very far from the firstorder model curves, whereas they practically coincide with the secondorder kinetic curves. From Table 4, we can see that the pseudosecondorder kinetic model has lower AIC values than the pseudofirstorder kinetic model. This means that the pseudosecondorder kinetic model explains the experimental data more appropriately. These results are consistent with those of other reports [30, 36, 37]. The rate constant k_{2} calculated from the pseudosecondorder kinetic model decreases as the initial RDB concentration increases. This indicates that chemisorption is significant in the ratelimiting step, involving valence forces through sharing or exchange of electrons between RDB and MIL101.
However, as mentioned earlier, in the present study, the adsorption process took place in three steps. Although the pseudosecondorder kinetic model accounts for the chemisorption nature well, it does not support the multisegment adsorption process. AlGhouti et al. [20] mentioned this problem and tried to analyse the adsorption as a threestep process with three linear segments on the kinetic plots by using the graphical approach method. Therefore, in this study, we analyse the adsorption process in the same way; that is, the kinetic plot is also divided into three segments each of which follows the pseudofirstorder adsorption kinetics.
The threestep kinetic rate equation was expressed as follows:where , , and are the amount of dye adsorbed at time after the subsequent kinetic steps (mg·g^{−1}); is the amount of dye adsorbed at time = 0; and , , and are the kinetic rate constants associated with each kinetic step. At t = ∞, when the adsorption reaches equilibrium, q_{t} = q_{e}, and q_{e} = q_{0} + q_{1} + q_{2} + q_{3}. Therefore, (22) can be written as
The values of , , and and , , and can be obtained using nonlinear regression by means of the Statistical Package for Scientific Social 20 (SPSS 20).
The threenonlinearsegment regressions using the pseudofirstorder kinetic model for the RDB adsorption on MIL101 are shown in Figure 5(b) and Table 5. High determination coefficients (0.91–0.99) indicate the appropriateness of the model. In addition, the AIC_{C} values for the pseudofirstorder kinetics with three segments are also the lowest of the three kinetic models (Table 6). This further confirms the best fit of the pseudofirstorder kinetics with three segments with the experimental data. The finding is also consistent with the analysis of the threestep adsorption process using Weber’s intraparticle diffusion model.


3.2.2. Effect of Particle Size and Agitation
To study diffusion kinetics of the RDB adsorption on MIL101 in terms of particle size, the threelinearsegment regression for the intraparticle diffusion model was applied to analyse the experimental data (Figure 6(a)). The results were the same as those of the effect of initial concentration. Intraparticle diffusion limited the adsorption rate at the initial step, and film diffusion controlled the process in the next two steps. The values of k_{p1} are 14.04, 29.94, and 13.55 mg·g^{−1}·min^{−0.5} for MHF0, MHF0.25, and MHF0.75, respectively. Theoretically, the intraparticle diffusion constant, k_{p1}, versus the inverse particle diameter, d^{−1}, did not give a straight line, and the conclusion for this is that the intraparticle diffusion was not the only operative mechanism [20]. In fact, in our study, this line was not linear () although MIL101 is a porous material; hence, we cannot rely on the particle size (external surface area) to confirm the adsorption mechanism. Therefore, the intraparticle diffusion mechanism controlled the initial step of the adsorption process. As can be seen from Figure 6(a), the adsorption capacity depends on the specific area rather than the particle size. This is because, for porous materials, the external surface contributes very little to the total surface area. In terms of particle size, only external surface area is concerned.
(a)
(b)
Stirring speed affects not only the distribution of the dye molecules in the bulk solution but also the formation of the external boundary film. Increasing stirring speed decreases the film thickness and thus the resistance to mass transfer around the adsorbent particle and increases the mobility of the whole system [38]. The effect of the stirring speed on RDB adsorption on MIL101 was carried out with three values: 200, 300, and 400 rpm (Figure 6(b)). It is evident that the adsorption capacity increased when the stirring speed increases from 200 rpm to 300 rpm and remained practically stable at 400 rpm. As the stirring speed increases the diffusion rate, the resistance of the solution becomes small. After a certain stirring rate, the external resistance no longer affects the sorption process.
3.2.3. Thermodynamic Studies
Temperature significantly affected the RDB adsorption over MIL101. When the temperature increased from 301 K to 333 K, the adsorption capacity increased rapidly from 120 mg·g^{−1} to 190 mg·g^{−1} (Figure 7(a)). This indicates that the RDB adsorption on MIL101 is an endothermic process. Similar results were observed in the adsorption of uranine [9] and methyl orange [32] on MIL101.
(a)
(b)
It is obvious that high temperature increased the diffusion rate of the dye molecules across the external boundary layer and in the internal pores of the adsorbent particle. This was the result of the decrease in the viscosity of the solution. In addition, the increase in adsorption capacity could also be ascribed to the increase in the number of active sites on the MIL101 surface due to the decrease in the hydrogen bonding between adsorbed water and the adsorbent making more sites available for RDB molecules.
The pseudosecondorder kinetic model was more consistent with the kinetic data than the pseudofirstorder kinetic model in the temperature range of 301 K to 333 K. Hence, the rate constant k_{2} was used to calculate the thermodynamic parameters. The E_{a} value obtained from the slope of the linear plot of ln k_{2} versus T^{−1} (F(3) = 59.15; R^{2} = 0.98, ) (Figure 7(b)) was 50.39 kJ·mol^{−1}. This large activation energy (over 42 kJ·mol^{−1}) implies that chemisorption controlled the adsorption of RDB on MIL101.
The thermodynamic parameters of the system, namely, ΔH^{0}, ΔS^{0}, and ΔG^{0} were evaluated using the van’t Hoff equation to determine the spontaneity of the adsorption process. The positive value of ΔH^{0} (Table 7) suggested an endothermic adsorption process. The positive value of ΔS^{0} indicated the increase in the randomness at the solidliquid interface during the adsorption of RDB molecules on the adsorbent [39]. The large negative values of ΔG^{0} strongly recommended the spontaneous RDB adsorption on MIL101. The more negative value at higher temperatures suggested that the spontaneity increased with temperature. As the change of Gibbs free energy was negative and accompanied by the positive standard entropy change, the adsorption reaction was spontaneous with high affinity.

3.2.4. Effect of pH
The pH of the solution affects the dye adsorption process because it can alternate both dye ionization and the ionic state of the surface of the adsorbent. Figure 8(a) presents the effect of pH on RDB adsorption from aqueous solutions. The RDB adsorption capacity of MIL101 increased slightly with pH from 3 to 5, followed by a significant increase with pH from 5 to 9. The pH_{pzc} (the point of zero charge) of MIL101 determined by the pH drift method [40] is around 5 (Figure 8(b)). This pH_{pzc} implies that the surface of the MIL101 is positively charged when pH of the solution is below 5, whereas the surface of adsorbent becomes negatively charged at pH above 5.
(a)
(b)
Increasing pH led to an increase in adsorption capacity, suggesting that the adsorption could follow a mechanism other than electrostatic interaction. The ππ stacking interaction between the aromatic rings of the RDB and the aromatic rings of terephthalate in the MIL101 framework was also thought to contribute to the RDB adsorption capacity. In addition, the coordination of the oxygen of the carboxyl group in the RDB molecules with the unsaturated Cr(III) ions in the MIL101 framework is also responsible for more efficient adsorption. A possible mechanism of RDB adsorption on MIL101 is illustrated in Scheme 2.
3.2.5. Adsorption Isotherms of RDB on MIL101
To describe the adsorption isotherms, Langmuir, Freundlich, Redlich–Peterson, Sips, and Toth equations were selected for use in this study. The determination of parameters of isotherm models is often based on the linear regression. However, linear regression requires the transformation of the original equation into a linear form that induces a problem related to abuse R^{2}. For example, the popular linear form of the Langmuir model is , in which is present in both independent and dependent variables [41]. Some papers [42, 43] reported that the linear form is less accurate than the nonlinear form in some cases of isotherm sorption as well as sorption kinetics. For these reasons, the isotherm equations in the nonlinear form are used in this study. Figure 9 shows the graphs that plot q_{e} versus C_{e} using the Langmuir, Freundlich, Redlich–Peterson, Sips, and Toth models. These models displayed lines around the experimental data, indicating that they all could describe the experimental data well.
The parameters of the isotherm models calculated using nonlinear regression are listed in Table 8. Except q_{m} derived from the Sips model (290.15 mg·g^{−1}), the values of q_{m} from the Redlich–Peterson and Toth models are fairly similar to that of q_{m} obtained from the Langmuir model due to the parameter of n being close to 1.

Based on the values of SSE as well as the coefficient of determination, we can see that the Langmuir, Redlich–Peterson, and Toth models provide a higher goodness of fit than the Sips and Freundlich models (Table 8).
To compare models with the same parameters and experimental points, SSE or R^{2} are frequently utilized to estimate the goodness of fit. As a result, it is obvious that the experimental data fit the Langmuir model better than the Freundlich model. However, for models with a different degree of freedom, Akaike’s information criterion [44] is used instead.
Table 9 shows the comparison of the Langmuir model with other models in the study. The value of ER (15.1, 15.8, 28.6, 1875.1) indicates that the Langmuir model is more appropriate than the Toth, Redlich–Peterson, and Freundlich models, implying that the monomolecularlayer nature is prevalent for the adsorption of RDB onto MIL101.

It is obvious that the piecewise linear regression is a useful approach to analyse the multilinearity. In order to find out the parameters in regression equations, the initial variables should be provided. If the initial variable is as far away as the real parameters, the solving equation is easy to go wrong due to finding the local minimum. This problem needs to study further. In this study, using AIC is proved to be effective for evaluating the goodness of fit for models which have a different number of parameters and experimental points. However, the application of AIC in adsorption filed is limited. Clarification of the meaning behind the AIC should be clarified.
3.2.6. Reusability of MIL101
In order evaluate the reusability of MIL101 in the removal of RDB, used MIL101 samples were regenerated by washing with a 0.25 M NaOH solution for 5 h under sonication and drying for 10 hours at 100°C. Three generations were performed, and the RDB adsorption capacity decreases slightly (96.5%) and remained at around 120 mg·g^{−1} (Figure 10(a)). Furthermore, MIL101 seems to be stable under adsorption conditions since the XRD patterns of generated MIL101 samples remained practically unchanged (Figure 10(b)).
(a)
(b)
4. Conclusions
MIL101 was synthesized using the hydrothermal process. The particle size of MIL101 could be controlled by adjusting the molar ratio of HF/H_{2}BDA. MIL101 can serve as a useful adsorbent for RDB under batch conditions. The synthesized MIL101 material displays high adsorption capacity and can be reused. Piecewise linear regression allows to objectively analyse the experimental data using Weber’s intraparticle diffusion and pseudokinetic adsorption models during the sorption process. The results of kinetic analysis suggested that the mechanism of the sorption of RDB on MIL101 might take place throughout the three steps: (i) film diffusion that dominates at the beginning of the process, (ii) chemisorption that monitors the subsequent period of the process, and (iii) intraparticle diffusion, where the adsorption significantly slowed down. The best fit of the pseudofirstorder kinetics with three segments with the experimental data is consistent with the analysis of the threestep adsorption process using Weber’s intraparticle diffusion model. Akaike’s information criterion was employed to compare different isotherm models with a different degree of freedom. The equilibrium data of RDB onto MIL101 fitted well to the Langmuir model rather than the Freundlich, Sips, Toth, and Redlich–Peterson models. As the change of Gibbs free energy was negative and accompanied by the positive standard entropy change, the adsorption process was spontaneous with high affinity.
Data Availability
The data used to support the findings of this study are available from the corresponding author upon request.
Conflicts of Interest
The authors declare that they have no conflicts of interest.
References
 D. Z. Van, P. Frank, and S. Villaverde, “Combined anaerobic–aerobic treatment of azo dyes—a short review of bioreactor studies,” Water Research, vol. 39, no. 8, pp. 1425–1440, 2005. View at: Publisher Site  Google Scholar
 W. H. Cheung, Y. S. Szeto, and G. McKay, “Intraparticle diffusion processes during acid dye adsorption onto chitosan,” Bioresource Technology, vol. 98, no. 15, pp. 2897–2904, 2007. View at: Publisher Site  Google Scholar
 J. Labanda, J. Sabaté, and J. Llorens, “Experimental and modeling study of the adsorption of single and binary dye solutions with an ionexchange membrane adsorber,” Chemical Engineering Journal, vol. 166, no. 2, pp. 536–543, 2011. View at: Publisher Site  Google Scholar
 S. Chen, J. Zhang, C. Zhang, Q. Yue, Y. Li, and C. Li, “Equilibrium and kinetic studies of methyl orange and methyl violet adsorption on activated carbon derived from Phragmites australis,” Desalination, vol. 252, no. 1–3, pp. 149–156, 2010. View at: Publisher Site  Google Scholar
 Y. Li and R. T. Yang, “Hydrogen storage in metalorganic and covalentorganic frameworks by spillover,” AIChE Journal, vol. 54, no. 1, pp. 269–279, 2008. View at: Publisher Site  Google Scholar
 P. L. Llewellyn, S. Bourrelly, C. Serre et al., “High uptakes of CO_{2} and CH_{4} in mesoporous metalorganic frameworks MIL100 and MIL101,” Langmuir, vol. 24, no. 14, pp. 7245–7250, 2008. View at: Publisher Site  Google Scholar
 T. K. Trung, N. A. Ramsahye, P. Trens et al., “Adsorption of C_{5}–C_{9} hydrocarbons in microporous MOFs MIL100(Cr) and MIL101(Cr): a manometric study,” Microporous and Mesoporous Materials, vol. 134, no. 1–3, pp. 134–140, 2010. View at: Publisher Site  Google Scholar
 E. Haque, J. E. Lee, I. T. Jang et al., “Adsorptive removal of methyl orange from aqueous solution with metalorganic frameworks, porous chromiumbenzenedicarboxylates,” Journal of Hazardous Materials, vol. 181, no. 1–3, pp. 535–542, 2010. View at: Publisher Site  Google Scholar
 F. Leng, W. Wang, X. J. Zhao, X. L. Hu, and Y. F. Li, “Adsorption interaction between a metal–organic framework of chromium–benzenedicarboxylates and uranine in aqueous solution,” Colloids and Surfaces A: Physicochemical and Engineering Aspects, vol. 441, pp. 164–169, 2014. View at: Publisher Site  Google Scholar
 C. Chen, M. Zhang, Q. Guan, and W. Li, “Kinetic and thermodynamic studies on the adsorption of xylenol orange onto MIL101(Cr),” Chemical Engineering Journal, vol. 183, pp. 60–67, 2012. View at: Publisher Site  Google Scholar
 P. A. Soloman, C. A. Basha, M. Velan, V. Ramamurthi, K. Koteeswaran, and N. Balasubramanian, “Electrochemical degradation of Remazol black B dye effluent,” Clean–Soil, Air, Water, vol. 37, no. 11, pp. 889–900, 2009. View at: Publisher Site  Google Scholar
 N. F. Cardoso, R. B. Pinto, E. C. Lima et al., “Removal of remazol black B textile dye from aqueous solution by adsorption,” Desalination, vol. 269, no. 1–3, pp. 92–103, 2011. View at: Publisher Site  Google Scholar
 V. P. Ranjusha, R. Pundir, K. Kumar, M. G. Dastidar, and T. R. Sreekrishnan, “Biosorption of Remazol black B dye (Azo dye) by the growing Aspergillus flavus,” Journal of Environmental Science and Health. Part A, vol. 45, no. 10, pp. 1256–1263, 2010. View at: Publisher Site  Google Scholar
 M. T. Thanh, T. V. Thien, V. T. T. Chau, P. D. Du, N. P. Hung, and D. Q. Khieu, “Synthesis of iron doped zeolite imidazolate framework8 and its remazol deep black RGB dye adsorption ability,” Journal of Chemistry, vol. 2017, Article ID 504597, 18 pages, 2017. View at: Publisher Site  Google Scholar
 M. T. Thanh, T. V. Thien, P. D. Du, and D. Q. Khieu, “Iron doped zeolitic imidazolate framework (FeZIF8): synthesis and photocatalytic degradation of RDB dye in FeZIF8,” Journal of Porous Materials, vol. 25, no. 3, pp. 857–869, 2017. View at: Publisher Site  Google Scholar
 M. Roosta, M. Ghaedi, and A. Asfaram, “Simultaneous ultrasonicassisted removal of malachite green and safranin O by copper nanowires loaded on activated carbon: central composite design optimization,” RSC Advances, vol. 5, no. 70, pp. 57021–57029, 2015. View at: Publisher Site  Google Scholar
 H. Motulsky and A. Christopoulos, Fitting Models to Biological Data Using Linear and Nonlinear Regression: A Practical Guide to Curve Fitting, Oxford University Press, Oxford, UK, 2004.
 G. Férey, C. MellotDraznieks, C. Serre et al., “A chromium terephthalatebased solid with unusually large pore volumes and surface area,” Science, vol. 309, no. 5743, pp. 2040–2042, 2005. View at: Publisher Site  Google Scholar
 W. J. Weber and J. C. Morris, “Kinetics of adsorption on carbon from solution,” Proceedings of the American Society of Civil Engineers/Journal of the Sanitary Engineering Division, vol. 89, pp. 31–59, 1963. View at: Google Scholar
 M. A. AlGhouti, M. A. M. Khraisheh, M. N. M. Ahmad, and S. Allen, “Adsorption behaviour of methylene blue onto Jordanian diatomite: a kinetic study,” Journal of Hazardous Materials, vol. 165, no. 1–3, pp. 589–598, 2009. View at: Publisher Site  Google Scholar
 V. Vadivelan and K. V. Kumar, “Equilibrium, kinetics, mechanism, and process design for the sorption of methylene blue onto rice husk,” Journal of Colloid and Interface Science, vol. 286, no. 1, pp. 90–100, 2005. View at: Publisher Site  Google Scholar
 M. F. Sawalha, J. R. PeraltaVidea, J. RomeroGonzález, and J. L. GardeaTorresdey, “Biosorption of Cd(II), Cr(III), and Cr(VI) by saltbush (Atriplex canescens) biomass: thermodynamic and isotherm studies,” Journal of Colloid and Interface Science, vol. 300, no. 1, pp. 100–104, 2006. View at: Publisher Site  Google Scholar
 R. R. Sheha and A. A. ElZahhar, “Synthesis of some ferromagnetic composite resins and their metal removal characteristics in aqueous solutions,” Journal of Hazardous Materials, vol. 150, no. 3, pp. 795–803, 2008. View at: Publisher Site  Google Scholar
 I. Langmuir, “The constitution and fundamental properties of solids and liquids,” Journal of the American Chemical Society, vol. 38, no. 11, pp. 2221–2295, 1916. View at: Publisher Site  Google Scholar
 H. M. F. Freundlich, “Over the adsorption in solution,” Journal of Physical Chemistry, vol. 57, pp. 385–471, 1915. View at: Google Scholar
 G. D. Halsey, “The role of surface heterogeneity in adsorption,” Advances in Catalysis, vol. 4, pp. 259–269, 1952. View at: Publisher Site  Google Scholar
 O. Redlich and D. L. Peterson, “A useful adsorption isotherm,” Journal of Physical Chemistry, vol. 63, no. 6, pp. 10–24, 1959. View at: Publisher Site  Google Scholar
 R. Sips, “On the structure of a catalyst surface,” Journal of Chemical Physics, vol. 16, no. 5, pp. 490–495, 1948. View at: Publisher Site  Google Scholar
 J. Toth, “State equations of the solid gas interface layer,” Acta Chimica Hungarica, vol. 69, pp. 311–317, 1971. View at: Google Scholar
 G. F. Malash and M. I. ElKhaiary, “Piecewise linear regression: a statistical method for the analysis of experimental adsorption data by the intraparticlediffusion models,” Chemical Engineering Journal, vol. 163, no. 2010, pp. 256–263, 2010. View at: Publisher Site  Google Scholar
 T. Loiseau and G. Férey, “Crystalline oxyfluorinated openframework compounds: silicates, metal phosphates, metal fluorides and metalorganic frameworks (MOF),” Journal of Fluorine Chemistry, vol. 128, no. 4, pp. 413–422, 2007. View at: Publisher Site  Google Scholar
 N. Mohammadi, H. Khani, V. K. Gupta, E. Amereh, and S. Agarwal, “Adsorption process of methyl orange dye onto mesoporous carbon material–kinetic and thermodynamic studies,” Journal of Colloid and Interface Science, vol. 362, no. 2, pp. 457–462, 2011. View at: Publisher Site  Google Scholar
 X. Peng, D. Huang, T. OdoomWubah, D. Fu, J. Huang, and Q. Qin, “Adsorption of anionic and cationic dyes on ferromagnetic ordered mesoporous carbon from aqueous solution: Equilibrium, thermodynamic and kinetics,” Journal of Colloid and Interface Science, vol. 430, pp. 272–282, 2014. View at: Google Scholar
 S. J. Allen, L. J. Whitten, M. Murray, O. Duggan, and P. Brown, “The adsorption of pollutants by peat, lignite and activated chars,” Journal of Chemical Technology and Biotechnology, vol. 68, no. 4, pp. 442–452, 1997. View at: Publisher Site  Google Scholar
 M. Korczak and J. Kurbiel, “New mineralcarbon sorbent: Mechanism and effectiveness of sorption,” Water Research, vol. 23, no. 8, pp. 937–946, 1989. View at: Publisher Site  Google Scholar
 Z. Hasan, J. Jeon, and S. H. Jhung, “Adsorptive removal of naproxen and clofibric acid from water using metalorganic frameworks,” Journal of Hazardous Materials, vol. 209–210, pp. 151–157, 2012. View at: Publisher Site  Google Scholar
 S. Lin, Z. Song, G. Che et al., “Adsorption behavior of metal–organic frameworks for methylene blue from aqueous solution,” Microporous and Mesoporous Materials, vol. 193, pp. 27–34, 2014. View at: Publisher Site  Google Scholar
 C.K. Lee, K.S. Low, and L.C. Chung, “Removal of some organic dyes by hexaneextracted spent bleaching earth,” Journal of Chemical Technology and Biotechnology, vol. 69, no. 1, pp. 93–99, 1997. View at: Publisher Site  Google Scholar
 E. I. Unuabonah, K. O. Adebowale, and B. I. OluOwolabi, “Kinetic and thermodynamic studies of the adsorption of lead (II) ions onto phosphatemodified kaolinite clay,” Journal Hazardous Materials, vol. 144, pp. 386–395, 2007. View at: Publisher Site  Google Scholar
 A. Kumar, B. Prasad, and I. M. Mishra, “Adsorptive removal of acrylonitrile by commercial grade activated carbon: Kinetics, equilibrium and thermodynamics,” Journal of Hazardous Materials, vol. 152, no. 2, pp. 589–600, 2008. View at: Publisher Site  Google Scholar
 M. I. ElKhaiary and G. F. Malash, “Common data analysis errors in batch adsorption studies,” Hydrometallurgy, vol. 105, no. 3–4, pp. 314–320, 2011. View at: Publisher Site  Google Scholar
 A. Günay, E. Arslankaya, and I. Tosun, “Lead removal from aqueous solution by natural and pretreated clinoptilolite: adsorption equilibrium and kinetics,” Journal of Hazardous Materials, vol. 146, no. 12, pp. 362–371, 2007. View at: Publisher Site  Google Scholar
 K. V. Kumar and S. Sivanesan, “Pseudosecond order kinetic models for safranin onto rice husk: Comparison of linear and nonlinear regression analysis,” Process Biochemistry, vol. 41, no. 5, pp. 1198–1202, 2006. View at: Publisher Site  Google Scholar
 H. Akaike, “A new look at the statistical model dentification,” IEEE Transactions on Automatic Control, vol. 19, no. 6, pp. 716–723, 1974. View at: Publisher Site  Google Scholar
Copyright
Copyright © 2018 Vo Thi Thanh Chau et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.