Optimization, Equilibrium and Kinetic Modeling of Methylene Blue Removal from Aqueous Solutions Using Dry Bean Pods Husks Powder

In this research, dry bean pods husks (DBPH) were used as an adsorbent material after minimum processing (without chemical substances consumption and without thermal treatment) to remove methylene blue from aqueous solutions. The adsorbent surface characteristics were investigated using SEM and FTIR analysis. For maximum removal efficiency, several parameters that influence the dye adsorption were optimized using the Taguchi method. Equilibrium and kinetic modeling, along with thermodynamic studies, were conducted to elucidate the adsorption mechanism. Taguchi experimental design showed that the factor with the highest influence was the adsorbent dose, with a percent contribution established by the ANOVA analysis of 40.89%. Langmuir isotherm and pseudo-second order kinetic model characterizes the adsorption process. The maximum adsorption capacity, 121.16 (mg g−1), is higher than other similar adsorbents presented in scientific literature. Thermodynamic parameters indicate a spontaneous, favorable and endothermic adsorption process, and their values show that physical adsorption is involved in the process. The obtained results, and the fact that adsorbent material is inexpensive and easily available, indicate that DBPH powder represents an effective absorbent for treating waters containing methylene blue. Additionally, the Taguchi method is very suitable to optimize the process.

Methylene blue (MB) dye is used in many industrial fields as well as in medicine. Even if it is not very toxic and dangerous, this compound can cause negative effects on human health: respiratory problems, vomiting, increased heart rate, cyanosis, eye irritation, nausea, methemoglobinemia, diarrhea and jaundice [2,9,[11][12][13][14][15][16][17]. Therefore, its removal from process effluents is absolutely required.
The Taguchi method is used to optimize various processes without increasing costs. It allows understanding the effect of variable process parameters in a small number of experimental tests, and the results obtained lead to improved process performance [19,[21][22][23][24][25][26]. The Taguchi method uses an orthogonal array that distributes the variables in a balanced way, and the experimental results are converted into a signal-to-noise ratio (S/N). The optimal value of the process parameters is given by the highest S/N ratio [19,21,22].
Common bean (Phaseolus vulgaris L.) is an important legume that grows in subtropical and temperate regions. Its seeds are a very important source of food for large parts of the population due to its nutritional properties (high source of dietary fiber and proteins, low level of lipids), low cost and long-term storage possibility. Almost 27 million tones of beans are produced annually in the world [27][28][29]. If common bean is grown for dry beans, a significant amount of dry pod husks results after harvesting and separation of the grains.
The aim of the present paper was to use this product, after minimum processing (without chemical substances consumption and without thermal treatment), to remove methylene blue dye from aqueous solutions by adsorption. Scanning electron microscopy (SEM) and Fourier transform infrared spectroscopy (FTIR) were carried out to study the adsorbent surface characteristics. For maximum removal efficiency the parameters that influenced the adsorption process were optimized by using the Taguchi method. For a description of the adsorption process, equilibrium, kinetics and thermodynamic parameters were calculated and discussed.

Adsorbent Preparation and Characterization
Dry bean pods husks (DBPH) were provided by a local agricultural producer from Cerneteaz village, Romania. The husks were first washed using distilled water, then dried at room temperature for three days and then at 90 • C for 24 h. The next operations to which the material was subjected were grinding (with an electric mill), passing over a sieve (with a mesh diameter of 2 mm), washing with distilled water (for turbidity and color removal) and drying at 105 • C for 5 h.
SEM analysis (Quanta FEG 250 microscope, at 1600× magnitude) was used to study the surface morphology of the adsorbent material. Identification of functional groups of the adsorbent was performed by FTIR spectroscopy (Shimadzu Prestige-21 FTIR spectrophotometer). The FTIR spectrum was recorded after the solid adsorbent sample was ground with IR transparent potassium bromide (KBr) and then pressed into a pellet.

Taguchi Experimental Design
The Taguchi (L27) orthogonal array was used to establish the optimum conditions for the dye removal by adsorption. The effect of five factors, at three levels, on the removal efficiency of dye was studied. Table 1 presents the controllable factors and their levels, which were used in the Taguchi design. The Taguchi approach converted the obtained experimental results into a signal-tonoise (S/N) ratio, which was used to analyze the experiment quality and the validity of the result. The terms "signal" and "noise" represent the desirable value (mean) and the undesirable value (standard deviation) for the output characteristic, respectively. To evaluate the experimental results, the highest adsorption efficiency was considered. In analyzing the signal-to-noise ratio of the Taguchi method, the "larger-the-better" option (Equation (1)) was chosen [19,[30][31][32]: where: n represents the repetitions number under similar experimental conditions, and y i represents the experimental response. The S/N ratio was calculated and analyzed using the Minitab19 software.
To establish the percentage contribution of each factor to the efficiency of MB removal and to evaluate the results of the Taguchi model, an analysis of variance (ANOVA-General Linear Model) was used [19,[30][31][32][33]. The necessary calculations were performed with the Minitab 19 software.

Adsorption Experimental Studies
The adsorption studies were performed at constant mixing intensity, in Erlenmeyer flasks, using 50 mL of MB solution together with the adsorbent material. The pH adjustment was realized with dilute solutions of NaOH and HCl (0.1 N). The dye concentration was determined by a UV-VIS spectrophotometer at 664 nm wavelength.
The adsorption capacity at equilibrium, (q e ), and at time t, (q t ), were calculated with Equations (2) and (3), while the dye removal percentage R(%) was calculated with Equation (4) [4,9,10,12,14]: where: C 0 represents the initial MB concentration (mg L −1 ), C e represents the MB equilibrium concentration (mg L −1 ), C t represents the MB concentration at time t (mg L −1 ), V represents the solution volume (L) and m represents the mass of adsorbent (g).

Equilibrium and Kinetic Modeling
Adsorption isotherms are very important for describing the solid-liquid adsorption process. The Langmuir and Freundlich isotherms were used to investigate the adsorption behavior. Their equations are presented below: Freundlich isotherm : where: q m represents the maximum absorption capacity (mg g −1 ), K L represents the Langmuir constant, K F represents the Freundlich constant and 1/n F represents an empirical constant indicating the adsorption intensity [34][35][36][37][38].
A lot of useful information on the mechanism and efficiency of adsorption used to design an industrial treatment plant is provided by kinetics studies. The pseudo-first-order and pseudo-second-order models were used to model the experimental data.
To establish the best-fitting kinetic and equilibrium models, the values of determination coefficient (R 2 ), sum of square error (SSE), chi-square (χ 2 ) and average relative error (ARE) were determined with the equations described below [37]. The higher value for R 2 and the smaller values for SSE, χ 2 and ARE were taken into account when choosing the most suitable models.
where: y i,exp represents the independent variable experimental value, y i,mod represents the modeled value, y i,exp represents the observed values mean o and n is the information total number.

Thermodynamic Parameter Determination
The data of methylene blue adsorption, at different temperatures (285, 296 and 306 K), were used to determine Gibbs free energy change, enthalpy change and entropy change, according to the following equations [37,39,40]: where: R represents the universal gas constant, K L represents the Langmuir constant and T represents the absolute temperature.

Desorption Experimental Studies
In the desorption studies, the MB-loaded adsorbent was mixed with various desorption agents: 0.1 N HCl, 0.1 N NaOH and distilled water.
The desorption percent D(%) of dye was calculated with Equation (15): where: m d represents amount of dye liberated by desorbing agent, and m a represents amount of dye adsorbed on adsorbent.

Adsorbent Material Characterization
SEM analysis of the adsorbent material showed the presence of irregular pores with different shapes and sizes on its surface (Figure 1a). After adsorption, the surface was modified, and the pores were covered by dye molecules (Figure 1b). amount of dye adsorbed on adsorbent.

Adsorbent Material Characterization
SEM analysis of the adsorbent material showed the presence of irregular pores with different shapes and sizes on its surface (Figure 1a). After adsorption, the surface was modified, and the pores were covered by dye molecules (Figure 1b). FTIR spectroscopy was used to identify the presence of different functional groups on the surface of adsorbent. The FTIR spectra illustrated in Figure 2 suggest cellulose and hemicellulose as main components. The differences between the peak's wavenumber before and after adsorption are less than 10 cm −1 , indicating an adsorption mechanism that could include physical interaction or ion-exchange mechanism [41]. FTIR spectroscopy was used to identify the presence of different functional groups on the surface of adsorbent. The FTIR spectra illustrated in Figure 2 suggest cellulose and hemicellulose as main components. The differences between the peak's wavenumber before and after adsorption are less than 10 cm −1 , indicating an adsorption mechanism that could include physical interaction or ion-exchange mechanism [41]. The specific peaks of the main functional group are: 3406 cm −1 -strong absorption peaks of O-H stretching vibration [42]; 2924 cm −1 -the peak belongs to -CH2 groups of cellulose [43]; 1738 cm −1 -C=O stretching vibration of carboxylic groups of hemicellulose [44]; 1636 cm −1 -O-H bending vibration of water sorption characteristics of cellulose [45]; 1453 cm −1 -the peak belongs to the bending of -CH groups of cellulose [43]; 1255 cm −1 -C-O stretching and CH or OH bending of hemicellulose structures [40,46]; 1026 cm −1 -C-O, C-O-H, C-O-C, C-C, ring stretching vibration in cellulose and hemicellulose [47]; 609 cm −1 -the bending modes of aromatic compounds of cellulose [48]. The specific peaks of the main functional group are: 3406 cm −1 -strong absorption peaks of O-H stretching vibration [42]; 2924 cm −1 -the peak belongs to -CH 2 groups of cellulose [43]; 1738 cm −1 -C=O stretching vibration of carboxylic groups of hemicellulose [44]; 1636 cm −1 -O-H bending vibration of water sorption characteristics of cellulose [45]; 1453 cm −1 -the peak belongs to the bending of -CH groups of cellulose [43]; Materials 2021, 14, 5673 6 of 16 1255 cm −1 -C-O stretching and CH or OH bending of hemicellulose structures [40,46]; 1026 cm −1 -C-O, C-O-H, C-O-C, C-C, ring stretching vibration in cellulose and hemicellulose [47]; 609 cm −1 -the bending modes of aromatic compounds of cellulose [48].

Optimization of Adsorption Parameters
Five controllable factors at three levels were used in the Taguchi design to estimate the optimum conditions for MB adsorption. Table 2 shows the L27 orthogonal array and results obtained after each run. Using the rank of S/N ratio, along with total increments (delta values), the order of the controllable factors' significance was determined ( Table 3). The delta value measures the magnitude of the effect considering the difference between the highest and lowest characteristic average for a controllable factor [19]. The factor that had the greatest influence on the process was the adsorbent dose, while the factor with the least influence was temperature. The optimum conditions of adsorption are also marked in Table 3.  The adsorbent dose has a greater influence on process efficiency due to the adsorption surface area and the number of sites available for adsorption increase with the adsorbent material dose [14,36,49]. Another parameter that has a great influence on the process is pH. Dye removal efficiency increases with the increasing of pH in the range 2-10. At lower values of pH the adsorbent surface is positively charged, but with the increase in this parameter the adsorbent surface became negatively charged and favored the electrostatic attraction with MB cations, resulting in a better efficiency [14,36,39]. With the increasing of the initial dye concentration, almost all the adsorption sites on the adsorbents became saturated due to the accumulation of dye molecules on the surface of the adsorbent particle and the removal percentage of the dye decreasing [1]. The impact of contact time was significant at the beginning of the process when a large number of active sites on the adsorbent surface were available for MB adsorption that generated a rapid increase in dye removal efficiency, until the equilibrium was reached after 30 min. The temperature had a lower influence on the dye removing process from the aqueous solution. The increase in temperature reduces the solution viscosity and has a positive effect on the mobility of the dye cations [39].
The order of the controllable factor influence predicted by the Taguchi design was confirmed by analysis of variance (ANOVA-General Linear Model). Figure 4 illustrates the specific influence of each factor on dye removal by adsorption on DBPH powder. Even though the Taguchi experimental design is based on a limited number of experiments, by analyzing the correlation of the predicted MB removal efficiency with experimental results, it can be observed that the accuracy of the Taguchi method prediction was very good ( Figure 5). The adsorbent dose has a greater influence on process efficiency due to the adsorption surface area and the number of sites available for adsorption increase with the adsorbent material dose [14,36,49]. Another parameter that has a great influence on the process is pH. Dye removal efficiency increases with the increasing of pH in the range 2-10. At lower values of pH the adsorbent surface is positively charged, but with the increase in this parameter the adsorbent surface became negatively charged and favored the electrostatic attraction with MB cations, resulting in a better efficiency [14,36,39]. With the increasing of the initial dye concentration, almost all the adsorption sites on the adsorbents became saturated due to the accumulation of dye molecules on the surface of the adsorbent particle and the removal percentage of the dye decreasing [1]. The impact of contact time was significant at the beginning of the process when a large number of active sites on the adsorbent surface were available for MB adsorption that generated a rapid increase in dye removal efficiency, until the equilibrium was reached after 30 min. The temperature had a lower influence on the dye removing process from the aqueous solution. The increase in temperature reduces the solution viscosity and has a positive effect on the mobility of the dye cations [39].
The order of the controllable factor influence predicted by the Taguchi design was confirmed by analysis of variance (ANOVA-General Linear Model). Figure 4 illustrates the specific influence of each factor on dye removal by adsorption on DBPH powder. Even though the Taguchi experimental design is based on a limited number of experiments, by analyzing the correlation of the predicted MB removal efficiency with experimental results, it can be observed that the accuracy of the Taguchi method prediction was very good ( Figure 5).

Equilibrium Modeling
While Langmuir isotherm assumes a monolayer adsorption on a homogeneous surface, Freundlich isotherm considers a multilayer adsorption on a heterogeneous surface, with the sites having different affinity [2,7,19].

Equilibrium Modeling
While Langmuir isotherm assumes a monolayer adsorption on a homogeneous surface, Freundlich isotherm considers a multilayer adsorption on a heterogeneous surface, with the sites having different affinity [2,7,19].

Equilibrium Modeling
While Langmuir isotherm assumes a monolayer adsorption on a homogeneous surface, Freundlich isotherm considers a multilayer adsorption on a heterogeneous surface, with the sites having different affinity [2,7,19].
Both isotherms are comparatively illustrated in Figure 6. The values of the isotherm's constants, summarized in Table 4, indicate that the adsorption process follows the Langmuir isotherm. The maximum adsorption capacity of 121.16 (mg g −1 ) is comparable to those previously reported in the literature, even higher than other similar adsorbents (Table 5). Both isotherms are comparatively illustrated in Figure 6. The values of the isotherm's constants, summarized in Table 4, indicate that the adsorption process follows the Langmuir isotherm. The maximum adsorption capacity of 121.16 (mg g −1 ) is comparable to those previously reported in the literature, even higher than other similar adsorbents (Table 5).

Kinetic Modeling
The pseudo-first-order and pseudo-second-order models were used in the kinetic study ( Figure 7). Kinetic parameters for these models were summarized in Table 6. The pseudo-second-order kinetic model had a higher value for R 2 and smaller values for SSE, χ 2 and ARE. Therefore, this model is best suited to describe the adsorption process.  The pseudo-first-order and pseudo-second-order models models tested for MB adsorption on DBPH powder.

Desorption Studies
In order to consider the regeneration possibility of adsorbent material, the desorption studies were performed in three different media (acid, neutral and basic). Figure 9 shows the desorption efficiency of the tested regeneration agents. The acid was found as being the best desorbing reagent with efficiency of about 85%.

Desorption Studies
In order to consider the regeneration possibility of adsorbent material, the desorption studies were performed in three different media (acid, neutral and basic). Figure 9 shows the desorption efficiency of the tested regeneration agents. The acid was found as being the best desorbing reagent with efficiency of about 85%.

Desorption Studies
In order to consider the regeneration possibility of adsorbent material, the desorption studies were performed in three different media (acid, neutral and basic). Figure 9 shows the desorption efficiency of the tested regeneration agents. The acid was found as being the best desorbing reagent with efficiency of about 85%.  The efficiency of the adsorption process after several adsorption-regeneration cycles was studied in the next stage of experimental determinations. The results are shown in Figure 10. It can be seen that the efficiency of the dye removal process decreased after each adsorption-regeneration cycle. If initially this parameter was 87% after the first cycle, it decreased to 73%, and after the second cycle it reached 15%. Practically only the first regeneration cycle is recommended, after which the adsorbent material has a lower performance. DBPH powder is cheap and easily available in large quantities; therefore, we consider that regeneration is not mandatory.
was studied in the next stage of experimental determinations. The results are shown in Figure 10. It can be seen that the efficiency of the dye removal process decreased after each adsorption-regeneration cycle. If initially this parameter was 87% after the first cycle, it decreased to 73%, and after the second cycle it reached 15%. Practically only the first regeneration cycle is recommended, after which the adsorbent material has a lower performance. DBPH powder is cheap and easily available in large quantities; therefore, we consider that regeneration is not mandatory. Figure 10. Dye removal efficiency after each adsorption-regeneration cycle.

Conclusions
The dye removal efficiency by adsorption on dry bean pods husks powder is influenced by solution pH, contact time, initial dye concentration and adsorbent dose. The Taguchi (L27) experimental design showed that the most influential factor was adsorbent dose (with a percent contribution, established by the ANOVA (General Linear Model) analysis, of 40.89%), followed by pH, initial dye concentration, time and temperature. The accuracy of the Taguchi method prediction was very good (R 2 = 0.998). Langmuir isotherm and pseudo-second-order kinetic model describe the adsorption process. These kinetic and equilibrium models had greater values for R 2 and smaller values for SSE, χ 2 and ARE. The maximum adsorption capacity, 121.16 (mg g −1 ), was higher compared to other similar adsorbents reported in the literature. The thermodynamic parameter values indicate a spontaneous, favorable and endothermic (ΔG 0 < 0, ΔH 0 > 0) adsorption process. Furthermore, they denote those electrostatic forces and van der Waals interactions are implied in the physical adsorption process (ΔH 0 < 20 kJ mol −1 ). The desorption studies performed in three different media (acid, neutral and basic) showed that the acid was the best desorbing reagent with an efficiency of about 85%. The efficiency of the adsorption process after several adsorption-regeneration cycles decreased after each cycle. After the second cycle it reached 15%. The obtained results, and the fact that adsorbent material is cheap and easily available, recommend DBPH powder as an efficient, low-cost adsorbent for methylene blue removal from aqueous media and the Taguchi method as a very suitable adsorption optimization algorithm.

Conclusions
The dye removal efficiency by adsorption on dry bean pods husks powder is influenced by solution pH, contact time, initial dye concentration and adsorbent dose. The Taguchi (L27) experimental design showed that the most influential factor was adsorbent dose (with a percent contribution, established by the ANOVA (General Linear Model) analysis, of 40.89%), followed by pH, initial dye concentration, time and temperature. The accuracy of the Taguchi method prediction was very good (R 2 = 0.998). Langmuir isotherm and pseudo-second-order kinetic model describe the adsorption process. These kinetic and equilibrium models had greater values for R 2 and smaller values for SSE, χ 2 and ARE. The maximum adsorption capacity, 121.16 (mg g −1 ), was higher compared to other similar adsorbents reported in the literature. The thermodynamic parameter values indicate a spontaneous, favorable and endothermic (∆G 0 < 0, ∆H 0 > 0) adsorption process. Furthermore, they denote those electrostatic forces and van der Waals interactions are implied in the physical adsorption process (∆H 0 < 20 kJ mol −1 ). The desorption studies performed in three different media (acid, neutral and basic) showed that the acid was the best desorbing reagent with an efficiency of about 85%. The efficiency of the adsorption process after several adsorption-regeneration cycles decreased after each cycle. After the second cycle it reached 15%. The obtained results, and the fact that adsorbent material is cheap and easily available, recommend DBPH powder as an efficient, low-cost adsorbent for methylene blue removal from aqueous media and the Taguchi method as a very suitable adsorption optimization algorithm.