Polyvinyl Alcohol Assisted Iron–Zinc Nanocomposite for Enhanced Optimized Rapid Removal of Malachite Green Dye

Eliminating hazardous contaminants is a necessity for maintaining a healthy environment on Earth. This work used a sustainable method to create Iron–Zinc nanocomposites with polyvinyl alcohol assistance. Mentha Piperita (mint leaf) extract was used as a reductant in the green synthesis of bimetallic nanocomposites. Doping with Poly Vinyl Alcohol (PVA) caused a reduction in crystallite size and greater lattice parameters. XRD, FTIR, EDS, and SEM techniques were used to establish their surface morphology and structural characterization. The high-performance nanocomposites were used to remove malachite green (MG) dye using the ultrasonic adsorption technique. Adsorption experiments were designed by central composite design and optimized by response surface methodology. According to this study, 77.87% of the dye was removed at the optimum optimized parameters (10.0 mg L−1 was the concentration of MG dye at a time of 8.0 min, pH 9.0, and 0.02 g of adsorbent amount) with adsorption capacity up to 92.59 mg·g−1. The dye adsorption followed Freundlich’s isotherm model and the pseudo-second-order kinetic model. Thermodynamic analysis affirmed the spontaneous nature of adsorption due to negative ΔGo values. As a result, the suggested approach offers a framework for creating an effective and affordable technique to remove the dye from a simulated wastewater system for environmental conservation.


Introduction
Nanomaterials have captivating interest largely due to their distinctive physicochemical characteristics, defined by their size, form, content, and crystallinity, which may all be determined. They are used in various fields, such as water treatment, biotechnology, catalysis, and medicine [1]. The metallic nanocomposites (MNCs), among the synthetic nanomaterials so far produced, have distinct characteristics such as catalysis, high chemical stability, electrical conductivity, and antibacterial properties [2]. On the basis of dimensions, nanocomposites are classified within four main categories, i.e., 3D (e.g., embedded network), 2D (e.g., nanoscale coatings), 1D (e.g., nanotubes), and 0D (e.g., embedded cluster). Classic methods for nanomaterials synthesis such as the chemical reduction method, polyol method, microemulsions, etc. are expensive, time-consuming, make use of toxic chemicals, and their application on a large scale is uneconomic; so, the usage of green strategies is becoming more popular due to their numerous advantages. The most commonly used optimal responses, which are generated by the specified parameters. RSM is a category of design of experiments that assembles an experimental system using statistical and mathematical tools within a certain number of pre-planned trials to assess the impact of the experimental factors and identify the best responses. Evaluation of response variations in relation to the modification of the experimental conditions is also helpful [28]. RSM is the basis of CCD (central composite design), which is typically utilized for experiment development with the fewest number of experimental runs. The simultaneous evaluation of the interdependent factorial replies is appropriate, and the potential for data redundancy is eliminated [25].
To eradicate the unwanted MG dye from the effluent, nanocomposites were created utilizing a mint extract, which is a natural source. The current study emphasizes the sustainable growth of polymer-aided Fe-Zn nanocomposites, i.e., (Fe-Zn-PVA-NCs). The application of the RSM was used to optimize the trials. With two central points, the CCD was used to assess the association between the removal of MG and four variables (concentration of dye, amount of adsorbent, pH, and time). Different kinetics and isotherm models were used to show and explain equilibrium data. Studies on regeneration and desorption were also carried out.
A pH meter (HANNA pH 211 microprocessor pH meter, Nusfalau, Romania) was used for pH measurements. The ultrasound-assisted adsorption procedure was carried out in an ultrasonicator (Elma-Hans Schmid Bauer Gmbh & Co, Singen, Germany) with a heating system. A UV-Vis spectrophotometer (T80, PG instruments LTD, Lutterworth, UK) at a wavelength of 618 nm, as shown in Figure 1, was used to measure the MG concentrations.
the treatment assembly to ultrasonic waves, the speed of adsorption is increased during the ultrasonication process. Because of the convection, this increases the rate of mass transfer. Additionally, it stimulates the adsorbent surface that was created by physical phenomena [27]. The response surface method (RSM) of optimization is used to obtain the optimal responses, which are generated by the specified parameters. RSM is a category of design of experiments that assembles an experimental system using statistical and mathematical tools within a certain number of pre-planned trials to assess the impact of the experimental factors and identify the best responses. Evaluation of response variations in relation to the modification of the experimental conditions is also helpful [28]. RSM is the basis of CCD (central composite design), which is typically utilized for experiment development with the fewest number of experimental runs. The simultaneous evaluation of the interdependent factorial replies is appropriate, and the potential for data redundancy is eliminated [25].
To eradicate the unwanted MG dye from the effluent, nanocomposites were created utilizing a mint extract, which is a natural source. The current study emphasizes the sustainable growth of polymer-aided Fe-Zn nanocomposites, i.e., (Fe-Zn-PVA-NCs). The application of the RSM was used to optimize the trials. With two central points, the CCD was used to assess the association between the removal of MG and four variables (concentration of dye, amount of adsorbent, pH, and time). Different kinetics and isotherm models were used to show and explain equilibrium data. Studies on regeneration and desorption were also carried out.
A pH meter (HANNA pH 211 microprocessor pH meter, Nusfalau, Romania) was used for pH measurements. The ultrasound-assisted adsorption procedure was carried out in an ultrasonicator (Elma-Hans Schmid Bauer Gmbh & Co, Singen, Germany) with a heating system. A UV-Vis spectrophotometer (T80, PG instruments LTD, Lutterworth, UK) at a wavelength of 618 nm, as shown in Figure 1, was used to measure the MG concentrations.
The MG dye solution of 500 mg L −1 , as stock, was made in deionized water, which was then diluted to prepare the working standards.  The MG dye solution of 500 mg L −1 , as stock, was made in deionized water, which was then diluted to prepare the working standards.
Energy Dispersive X-ray Spectroscopy (EDS; Jeol Japan EX-54175 jMU, Tokyo, Japan) and Scanning Electron Microscopy (SEM; Jeol Japan JFC-1500) were used to study the composition and morphology of the Fe-Zn-PVA NCs, respectively, under an accelerated voltage of 15 kV. Fourier Transform Infrared Spectroscopy (FTIR; IR-Prestige 21-Schimadzu, Kyoto, Japan) was used to study the orientations and vibrations present and its chemical composites. The orientation and crystalline structure of the Fe-Zn-PVA NCs were analyzed by X-Ray Diffraction (XRD; Jeol, Japan, JDX 3532) using Cu Kα (λ = 1.54 Å).

Preparation of Mint Extract
The green extract was prepared using mint leaves (Mentha Piperita). About 20.0 g of fresh leaves of mint were purchased from a market in Karachi city. After cleaning, leaves were washed with deionized distilled water, air dried, cut into fine pieces, and later crushed using a mortar and pestle. The crushed leaves were added to 100 mL of deionized water. The process of extraction progressed further by heating this solution for 15 min at 353 K. After that, the contents were cooled and filtered. As a result, the extract that was recovered was employed as a reducing agent [29].

Preparation of Fe-Zn-PVA NCs
The nanoparticle of Zinc and Iron was produced independently, after the addition of ZnSO 4 ·7H 2 O (0.1 M) and FeSO 4 ·7H 2 O (0.1 M) at a volume ratio of 1:2 to the extract. At room temperature, the mixtures were swirled for 30 min. Bimetallic nanocomposites were created by combining the independently produced zinc and iron nanoparticles in a 1: For the preparation of the solution of PVA, about 10 g of PVA was added to 100 mL of deionized distilled water and stirred at a temperature of 80 • C until it was completely dissolved. The produced PVA solution was combined at a 1:1 proportion with the synthesized Fe-Zn nanocomposites, and the mixture was agitated for 30 min at room temperature. The synthesized nanocomposites were vigorously stirred with polymer solution for a particular time period for the completion of the reaction. The evaporation of the filtrate proceeded on a hot plate; after that, the Fe-Zn-PVA NCs were synthesized. They were then rinsed several times with ethanol and deionized water, and dried in an oven. The preparation scheme is shown in Figure 2 [30][31][32].
the composition and morphology of the Fe-Zn-PVA NCs, respectively, under an accelerated voltage of 15 kV. Fourier Transform Infrared Spectroscopy (FTIR; IR-Prestige 21-Schimadzu, Kyoto, Japan) was used to study the orientations and vibrations present and its chemical composites. The orientation and crystalline structure of the Fe-Zn-PVA NCs were analyzed by X-Ray Diffraction (XRD; Jeol Japan, JDX 3532) using Cu Kα (λ = 1.54 Å).

Preparation of Mint Extract
The green extract was prepared using mint leaves (Mentha Piperita). About 20.0 g of fresh leaves of mint were purchased from a market in Karachi city. After cleaning, leaves were washed with deionized distilled water, air dried, cut into fine pieces, and later crushed using a mortar and pestle. The crushed leaves were added to 100 mL of deionized water. The process of extraction progressed further by heating this solution for 15 min at 353 K. After that, the contents were cooled and filtered. As a result, the extract that was recovered was employed as a reducing agent [29].

Preparation of Fe-Zn-PVA NCs
The nanoparticle of Zinc and Iron was produced independently, after the addition of ZnSO4•7H2O (0.1 M) and FeSO4•7H2O (0.1 M) at a volume ratio of 1:2 to the extract. At room temperature, the mixtures were swirled for 30 min. Bimetallic nanocomposites were created by combining the independently produced zinc and iron nanoparticles in a 1: For the preparation of the solution of PVA, about 10 g of PVA was added to 100 mL of deionized distilled water and stirred at a temperature of 80 °C until it was completely dissolved. The produced PVA solution was combined at a 1:1 proportion with the synthesized Fe-Zn nanocomposites, and the mixture was agitated for 30 min at room temperature. The synthesized nanocomposites were vigorously stirred with polymer solution for a particular time period for the completion of the reaction. The evaporation of the filtrate proceeded on a hot plate; after that, the Fe-Zn-PVA NCs were synthesized. They were then rinsed several times with ethanol and deionized water, and dried in an oven. The preparation scheme is shown in Figure 2 [30][31][32].

pH at Point of Zero Charge (pH pzc )
The pH value determined using the pH drift technique when the net charge of the nanocomposite surface becomes neutral is known as the Point of Zero Charges (pH pzc ). We added specific amounts of NaOH and HCl; the pH of the starting set of samples of about 50 mL of NaCl (0.05 M) concentration was kept within 2-12. An appropriate quantity of Fe-Zn-PVA NCs was added to the respective system and pH of the solutions was measured after 24 h and 48 h intervals [24].

Ultrasonic-Assisted Adsorption Process
Adsorption of MG dye over the surface of Fe-Zn-PA NCs was accomplished in batch mode with the aid of the ultrasonication method. A solution with known concentrations of MG dye (5,10,15,20, and 25 mg L −1 ) of around 50 mL was made. The solution's pH was held steady at 9. Approximately, Fe-Zn-PVA NCs (0.02 g) were dissolved in the dye solution and ultrasonically processed (8.0 min) at room temperature after measuring the preliminary absorbance at 618 nm with a UV-Visible Spectrophotometer. The following equation was used to calculate the MG dye's % removal [24].
where A f and A i show absorbance (final and initial), respectively. The adsorption capacity of nanocomposite (Fe-Zn-PVA) was computed as [24] where C e and C o symbolize the equilibrium and initial dye's concentration (mg L −1 ), respectively; V is the volume in L (L); and W shows the adsorbent mass in g (g).

Central Composite Design (CCD)
The most pertinent type of RSM design, central composite design (CCD), was used for optimization and modelling to impacts of 4 variables (independent)-sonication time (A), pH (B), Fe-Zn-PVA NCs quantity (C), and MG concentration-on the adsorption (with ultrasonic assistance) of dye (MG) against Fe-Zn-PVA NCs [27]. In order to remove the MG dye, the correlation between the four parameters described above and the two core areas was examined.
Analysis of Variance (ANOVA), response surface plots, and regression analysis were utilized for studying the effects of adsorption factors and determining the prerequisites for the highest possible adsorption efficiency [33].

Adsorption Kinetics, Isotherms, and Thermodynamics Studies
In order to conduct the adsorption kinetics investigation, 50 mL of the MG solution (10 mg L −1 ) was mixed with Fe-Zn-PVA NCs (0.02 g) and then sonicated for 1 to 10 min while the pH was maintained at 9.
The process spontaneity was examined using thermodynamic studies. ∆S • , ∆G • , and ∆H • were evaluated by mapping ln K D vs. T −1 .

Surface Regeneration of Fe-Zn-PVA NCs
For four cycles, the capacity of Fe-Zn-PVA NCs to regenerate for the subsequent sorption of MG was examined. In the first sorption cycle, 0.02 g of Fe-Zn-PVA NCs were treated for 8.0 min in 50 mL MG dye with a concentration of 10 mg L −1 in 50 mL at pH 9.0. Following the first cycle's effective dye sorption, Fe-Zn-PVA NCs were sorted and cleaned with distilled water. After drying, MG dye was removed from Fe-Zn-PVA NCs using a variety of solvents, including sulfuric acid (0.05 M), sodium hydroxide (0.05 M), hydrochloric acid (0.05 M), glacial acetic acid (0.05 M), and ethanol. Glacial acetic acid was the solvent used in all subsequent desorption tests since it was the most successful method for MG dye desorption. In the MG sorption and desorption investigations described above, the dye-desorbed Fe-Zn-PVA NCs were used again for the upcoming three cycles, and the sorption effectiveness of Fe-Zn-PVA NCs was noted [34]. Figure 3 shows the FTIR spectra of Fe-Zn-PVA nanocomposites to examine the bonding interactions and surface functionalization. The -OH of PVA molecules that are integrated onto Fe-Zn NCs are responsible for the wide peak with a center value of 3460 cm −1 . Zn-O and Zn-Zn vibrations caused the vibrations of stretching to manifest at 439.77 cm −1 , 493.78 cm −1 , and a tiny height at 750 cm −1 [35]. Fe-O can be detected in the peaks at 603 cm −1 and 651 cm −1 . Additionally, a large, irregular tip at 1637.56 cm −1 displays the mild twisting of the N-H group, whereas a tiny crest at 1384.89 cm −1 displays the curving of the (phenolic) OH group. Additionally, because of the significant broadening of the M-O-C bond, two peaks in the area of 1100 cm −1 namely, 1097.5 cm −1 and 1136.07 cm −1 can be observed [31,36].

Characterization of Fe-Zn-PVA NCs
cleaned with distilled water. After drying, MG dye was removed from Fe-Zn-PVA NCs using a variety of solvents, including sulfuric acid (0.05 M), sodium hydroxide (0.05 M), hydrochloric acid (0.05 M), glacial acetic acid (0.05 M), and ethanol. Glacial acetic acid was the solvent used in all subsequent desorption tests since it was the most successful method for MG dye desorption. In the MG sorption and desorption investigations described above, the dye-desorbed Fe-Zn-PVA NCs were used again for the upcoming three cycles, and the sorption effectiveness of Fe-Zn-PVA NCs was noted [34]. Figure 3 shows the FTIR spectra of Fe-Zn-PVA nanocomposites to examine the bonding interactions and surface functionalization. The -OH of PVA molecules that are integrated onto Fe-Zn NCs are responsible for the wide peak with a center value of 3460 cm −1 . Zn-O and Zn-Zn vibrations caused the vibrations of stretching to manifest at 439.77 cm −1 , 493.78 cm −1 , and a tiny height at 750 cm −1 [35]. Fe-O can be detected in the peaks at 603 cm −1 and 651 cm −1 . Additionally, a large, irregular tip at 1637.56 cm −1 displays the mild twisting of the N-H group, whereas a tiny crest at 1384.89 cm −1 displays the curving of the (phenolic) OH group. Additionally, because of the significant broadening of the M-O-C bond, two peaks in the area of 1100 cm −1 namely, 1097.5 cm −1 and 1136.07 cm −1 can be observed [31,36]. The SEM image of Fe-Zn-PVA NCs is shown in Figure 4a; the nanocomposites are between 80 to 100 nm in size and have spherical surfaces, which makes it easier for the dye to adhere to the Fe-Zn-PVA NCs.

Characterization of Fe-Zn-PVA NCs
The EDX (Energy Dispersive X-ray) spectrum, shown in Figure 4b, demonstrates the presence of carbon, oxygen, iron, and zinc in large numbers, but only small levels of calcium and chlorine, which may be present in the water utilized as interfering species [37]. The SEM image of Fe-Zn-PVA NCs is shown in Figure 4a; the nanocomposites are between 80 to 100 nm in size and have spherical surfaces, which makes it easier for the dye to adhere to the Fe-Zn-PVA NCs.  The XRD diffractogram displayed in Figure 5 shows that there were peaks at 23.87, 25.65, and 95.75, with miller indices of (100), (100), and (320) being detected at each peak, respectively. PVA's interaction with Fe-Zn nanocomposites is what causes the decline in peak intensities and amalgamation of the diffraction peaks. Therefore, the XRD measurements show that the crystallite's structural development is complicated [38]. The EDX (Energy Dispersive X-ray) spectrum, shown in Figure 4b, demonstrates the presence of carbon, oxygen, iron, and zinc in large numbers, but only small levels of calcium and chlorine, which may be present in the water utilized as interfering species [37].
The XRD diffractogram displayed in Figure 5 shows that there were peaks at 23.87, 25.65, and 95.75, with miller indices of (100), (100), and (320) being detected at each peak, respectively. PVA's interaction with Fe-Zn nanocomposites is what causes the decline in peak intensities and amalgamation of the diffraction peaks. Therefore, the XRD measurements show that the crystallite's structural development is complicated [38]. The XRD diffractogram displayed in Figure 5 shows that there were peaks at 23.87, 25.65, and 95.75, with miller indices of (100), (100), and (320) being detected at each peak, respectively. PVA's interaction with Fe-Zn nanocomposites is what causes the decline in peak intensities and amalgamation of the diffraction peaks. Therefore, the XRD measurements show that the crystallite's structural development is complicated [38].

Determination of Surface Neutrality by pHpzc of Fe-Zn-PVA NCs
The pHpzc for Fe-Zn-PVA NCs was determined by producing a graph of ΔpH vs. initial pH, and it was determined to be 7.0, as shown in Figure 6.

Determination of Surface Neutrality by pH pzc of Fe-Zn-PVA NCs
The pH pzc for Fe-Zn-PVA NCs was determined by producing a graph of ∆pH vs. initial pH, and it was determined to be 7.0, as shown in Figure 6. As mentioned, if pH > pHpzc, the nanocomposites' surface becomes negatively charged, attracting molecules with opposing charges and vice versa. Because Fe-Zn-PVA NCs are neutral at a pH value of 7.0 and MG is a dye with cationic characteristics with a surface positive charge, the two are attracted to one another. As the solution's pH rises, the particle surface turns more negatively charged, leading to a larger percentage of dye removal, as shown in Figure 7 [33]. surface positive charge, the two are attracted to one another. As the solution's pH rises, the particle surface turns more negatively charged, leading to a larger percentage of dye removal, as shown in Figure 7 [33]. As mentioned, if pH > pHpzc, the nanocomposites' surface becomes negatively charged, attracting molecules with opposing charges and vice versa. Because Fe-Zn-PVA NCs are neutral at a pH value of 7.0 and MG is a dye with cationic characteristics with a surface positive charge, the two are attracted to one another. As the solution's pH rises, the particle surface turns more negatively charged, leading to a larger percentage of dye removal, as shown in Figure 7 [33].

Design of Experiments
To ascertain the impact of the interactions between the key components for the eradication of MG, a four-factorial, five-level design was created, and the experimental runs were performed according to the central composite design provided in Table 1 along with the experimental and predicted responses.

Design of Experiments
To ascertain the impact of the interactions between the key components for the eradication of MG, a four-factorial, five-level design was created, and the experimental runs were performed according to the central composite design provided in Table 1 along with the experimental and predicted responses. Analysis Of Variance (ANOVA) was carried out to assess model significance along with the significance pertaining to all the variables and interactions. Table 2 shows that the model is considered statistically significant based on P and F values. The p-values of under 0.500 (i.e., 0.0001) and F-value (46.33) were found to indicate that the model is significant at a 95% confidence level. It is positive and demonstrates the strong predictability of the model, where the Lack of Fit F-value (F tab = 4.74 > F cal = 3.29) and p-value greater than 0.05 suggest the non-significance for the Lack of Fit that is desired [33].
It is clear from the ANOVA (shown in Table 2) that the most important factor for the MG removal is pH (p < 0.0001, SS = 2368.11, F = 501.14) and the next is time (p < 0.0001, SS = 155.14, F = 32.83). Another significant component is the interaction A 2 (p < 0.0001, SS = 129.85, F = 27.48).
The statistics of the model summary are shown in Table 2. The great R 2 (0.9774) demonstrates how well the data matched the fit regression line. The modified R 2 of 0.9563 was a fair deal with the expected R 2 of 0.8827, indicating that the regression line does not overfit the model, nor does the addition of an independent variable make the model worse [39,40]. In the above table, 'a' denotes the degree of freedom, 'b' denotes the sum of squares, and 'c' denotes the mean squares of the datasets. The model summary for the removal of MG by synthesized nanocomposites revealed an exceptional model fitting as the R 2 value was observed to be 0.9774 with an adjusted R 2 value of 0.9563 and predicted R 2 value of 0.8827. (3), is useful for identifying the comparative impact of the four factors (i.e., sonication time, pH, the concentration of MG, and amount of adsorbent) and their contact for MG removal.

The coded equation, as represented in Equation
This equation shows, because of the individual variables (sonication time, pH, amount of adsorbent and the concentration of pollutant), there has been a noticeable favorable effect for the elimination of MG dye (p < 0.05). Additionally, the reaction is inhibited by the dye concentration. Of the three factors, pH had the biggest impact on the MG removal from an aqueous solution.

Validation and Optimization of Fe-Zn-PVA NCs
The following factors were established to be the most efficient for removing MG dye: 8.0 min duration; pH 9.0; 0.02 g of the nanocomposite; MG dye concentration 10.0 mg L −1 ; and the highest removal of the model, which was anticipated to be 85.16%. The best conditions were verified by conducting a series of five tests at room temperature, and the observed results were 77.87% with a variance coefficient of 0.989%.

Effect of Parameters on MG Dye Removal
The 3D representations of surface graphs shown in Figure 8a-c illustrate the effects of four distinct factors on the responses: the quantity of adsorbent, time, pH, and adsorbate (dye) concentration. that as pH rises, adsorption effectiveness rises as well. MG is a cationic dye, and when the pH > pHpzc, the adsorbent takes on a negative charge. As a result, electrostatic attraction causes a higher percent removal [33,37]. The reaction was observed to be at its optimum with a rise in pH and interaction duration, as illustrated in Figure 8c. A graph of the normal probability of residuals for the MG's removal is shown in Figure 8d, and it represents that the actual points of data are evenly spaced out all over the center axis.

Adsorption Kinetics
In Figure 9a-d, different models (Elovich model, intra-particle diffusion, pseudo-first-order, and pseudo-second-order) of kinetic studies were analyzed to observe the process mechanism. According to the results provided in Table 3, the pseudo-second-order was the model that best fit the data when compared to the other three models, with an R 2 value of 0.9970. Furthermore, there was a fair agreement between the calculated and experimental Qe values. However, even though the R 2 value The relationship between the volume of adsorbent and the duration of the ultrasonication is shown in Figure 8a. It was noted that as the adsorbent's amount increased, the process efficiency grew due to the accessibility of more active sites, which increases the likelihood of adsorption. Additionally, the sonication period has a limited positive impact on the effectiveness of dye removal and might have negative impacts if it is prolonged due to system saturation and dye molecule desorption [37].
Adsorbate molecules are present at lower concentrations in the vicinity of accessible sites, and the percentage removal of adsorption is higher at lower than higher concentrations of adsorbate (MG), as seen in Figure 8b. As a result, the percentage of molecules that are adsorbed will be higher at lower concentrations than at higher concentrations. The reaction is inversely influenced by the solution's pH. It was found that as pH rises, adsorption effectiveness rises as well. MG is a cationic dye, and when the pH > pH pzc , the adsorbent takes on a negative charge. As a result, electrostatic attraction causes a higher percent removal [33,37].
The reaction was observed to be at its optimum with a rise in pH and interaction duration, as illustrated in Figure 8c. A graph of the normal probability of residuals for the MG's removal is shown in Figure 8d, and it represents that the actual points of data are evenly spaced out all over the center axis.

Adsorption Kinetics
In Figure 9a-d, different models (Elovich model, intra-particle diffusion, pseudo-firstorder, and pseudo-second-order) of kinetic studies were analyzed to observe the process mechanism. According to the results provided in Table 3, the pseudo-second-order was the model that best fit the data when compared to the other three models, with an R 2 value of 0.9970. Furthermore, there was a fair agreement between the calculated and experimental Q e values. However, even though the R 2 value for the pseudo-first-order model was also high (0.983), there was a significant discrepancy between the estimated and observed Q e values, suggesting that the process of adsorption does not closely follow the Lagergran equation. The intra-particle diffusion model and Weber and Morris' equation were used to analyze the likelihood of intra-particle diffusion for the examination of a diffusioncontrolled adsorption system. Figure 9c explains that the plot will not cross through the origins once extended back to the y-axis, indicating that boundary layer diffusion affects the adsorption. Additionally, a tiny curve in the straight line indicates that pore diffusion also occurred [41,42].
The Elovich model is used to analyze the processes involving chemical adsorption. It is also useful to define the extent of chemisorption.
The values of respective constants are represented in Table 3. The R 2 value shows that this study followed the respective model very well [27]. Table 3. Adsorption kinetics parameters for MG adsorption onto Fe-Zn-PVA NCs.

Models Equations Parameters Values
Pseudo-first-order log(Q e − Q t ) = log Q e − model and Weber and Morris' equation were used to analyze the likelih intra-particle diffusion for the examination of a diffusion-controlled adsorption Figure 9c explains that the plot will not cross through the origins once extended the y-axis, indicating that boundary layer diffusion affects the adsorption. Additio tiny curve in the straight line indicates that pore diffusion also occurred [41,42]. The Elovich model is used to analyze the processes involving chemical adsor is also useful to define the extent of chemisorption.  Table 3. The R 2 value that this study followed the respective model very well [27]. Table 3. Adsorption kinetics parameters for MG adsorption onto Fe-Zn-PVA NCs.

Models Equations Parameters V
Pseudo-first-order log Q Q log Q k 2.303t Various parameters are involved in different models in the above table. Here, Q e , Q t , k 1 , k 2 , K id , and C denote the amount of MG adsorbed at equilibrium, amount of MG adsorbed at ultrasonicated time "t", pseudo-first-order rate constant, pseudo-second-order rate constant, intra-particle diffusion rate constant, and thickness of the boundary layer, respectively.

Adsorption Isotherms
The adsorption equilibrium at various temperatures was ascertained using the Langmuir, Freundlich, Temkin, and D-R isotherm models, as shown in Figure 10a-d.

Langmuir Isotherm Model
The monolayer adsorption that occurs on the homogeneous surface with limited identical adsorption sites is described by the Langmuir adsorption isotherm model [27,42].
The intercept (1/K L Q m ) and the slope (1/Q m ) of the graph were used to calculate the constant for the model parameters. The values of Q m and K L decreased as temperature rose, indicating that low temperatures are best for adsorbent and adsorbate interaction. The R L values, which were observed to be below 1, suggested that the Langmuir model for the adsorption process was feasible [43]. It has been reported in the Literature [43] that if R L > 1, the process is unfavorable; if R L = 1, the process is linear; if 0 < R L < 1, the process is favorable; and if R L = 0, the process is irreversible. Nanomaterials 2023, 13, x FOR PEER REVIEW 1   The higher R 2 values up to 0.9920 showed that the Langmuir isotherm model wellrepresented the adsorption equilibrium data.

Freundlich Isotherm Model
To analyze the heterogeneity in the adsorbent surface, which results from the presence of several functional groups on the adsorbent surface, the Freundlich adsorption isotherm model is used [42].
The model's K F and n parameters were determined using the plot's intercept (log K F ) and slope (1/n), respectively, as presented in Table 4. Since n > 1, as documented in the Lature, the adsorption nature is considered to be favorable; "n" reflects the favorability of the adsorption, which was more than 1. The finding that the value of K F decreases as temperature rises indicates that adsorption is more effective at decreased temperatures. The large values of R 2 up to 0.9943 indicate that the Freundlich isotherm may be effectively applied to fit equilibrium data to MG dye adsorption onto the surface of Fe-Zn-PVA NCs [42].

Temkin and Dubinin-Radushkevich Isotherm Models
The Temkin isotherm model is employed to study the effect of absorption heat, which changed during the adsorbate-adsorbate interactions. The constants of model parameters (K T and B) were determined using the slope of the plot (B) and intercept (B ln K T ), as illustrated in (Table 4). Additionally, the R 2 > 0.8 indicates that this model is quite favorable.
The heterogeneous surface adsorption is described by the D-R isotherm model. Additionally, the energy of sorption, or E, predicts the kind of adsorption. When it is less than 8.00 kJ mol −1 , physical adsorption is assumed to occur. Since, in Table 4, the E value is less than 8.00 kJ mol −1 , it is clear that physical force has a greater influence on the current adsorption system [24,33]. In the above table, various isotherm parameters are represented by Q m , K L , K F , n, K T , B, R, T, β, Q s , ε, and E, which denote the amount of adsorbate per unit mass of adsorbent at equilibrium, Langmuir constant, Freundlich constant, constant related to adsorption capacity, Temkin isotherm equilibrium constant, Temkin constant related to heat adsorption, universal gas constant, absolute temperature, D-R isotherm constant, theoretical saturation capacity of the isotherm, D-R constant called Polanyi potential, and Mean free energy of sorption.

Adsorption Thermodynamics
To establish the temperature impact on the viability of MG dye adsorption onto Fe-Zn-PVA NCs, parameters of thermodynamics containing ∆G (free energy change), ∆S (entropy), and ∆H (enthalpy) were investigated [29]. The free energy change was calculated using the following equation. The change in free energy was estimated by the given equation: where R shows the universal gas constant (8.314 J mol −1 K −1 ), K D shows the thermodynamic equilibrium constant, and T shows the temperature (K). Moreover, in order to study the impact of temperature on the thermodynamic equilibrium constant, the equations used are given as follows: K D , R, and T have the same representation as described in the above equation. Figure 11 shows the linear plot of ln K D versus 1/T, and the values of intercept and slope were utilized to determine ∆S and ∆H, respectively. According to the values of ∆G and ∆H shown in (Table 5), the adsorption mechanism was exothermic and spontaneous, respectively.

ΔG
RTlnK (6) KD, R, and T have the same representation as described in the above equation. Figure  11 shows the linear plot of ln KD versus 1/T, and the values of intercept and slope were utilized to determine ΔS and ΔH, respectively. According to the values of ΔG and ΔH shown in (Table 5), the adsorption mechanism was exothermic and spontaneous, respectively. Figure 11. Thermodynamics studies of MG on Fe-Zn-PVA NCs. Figure 11. Thermodynamics studies of MG on Fe-Zn-PVA NCs.

Desorption and Surface Regeneration Studies
The studies for desorption were carried out to meet the demand economically and to assess an adsorbent's reusability. Figure 12 shows that, when compared to the other studied solvents, glacial acetic acid's maximal desorption capability after 8 min of ultrasonication was observed to be 51.97%.
The effectiveness of Fe-Zn-PVA NCs to adsorb MG was evaluated by conducting four cycles of the adsorption-desorption process in batch mode. MG sorption was discovered to be 76.55% in the first cycle, but after three cycles of subsequent adsorption and desorption, the % adsorption was reduced to 70.47, as shown in Figure 13. The strength of the binding forces that exist between the molecules of adsorbate and adsorbent determines the degree of reversibility of adsorption. Because of this, MG molecules that were adsorbed physically from Fe-Zn-PVA NCs departed the surface after 8 min but chemisorbed molecules found it difficult to do so, resulting in a higher percentage of adsorption than desorption, as seen in Figures 12 and 13 [2].

Desorption and Surface Regeneration Studies
The studies for desorption were carried out to meet the demand economically and to assess an adsorbent's reusability. Figure 12 shows that, when compared to the other studied solvents, glacial acetic acid's maximal desorption capability after 8 min of ultrasonication was observed to be 51.97%. The effectiveness of Fe-Zn-PVA NCs to adsorb MG was evaluated by conducting four cycles of the adsorption-desorption process in batch mode. MG sorption was discovered to be 76.55% in the first cycle, but after three cycles of subsequent adsorption and desorption, the % adsorption was reduced to 70.47, as shown in Figure 13. The strength of the binding forces that exist between the molecules of adsorbate and adsorbent determines the degree of reversibility of adsorption. Because of this, MG molecules that were adsorbed physically from Fe-Zn-PVA NCs departed the surface after 8 min but chemisorbed molecules found it difficult to do so, resulting in a higher percentage of adsorption than desorption, as seen in Figures 12 and 13 [2]. The mechanism of adsorption and desorption can be easily understood by the following scheme given in Figure 14. The mechanism of adsorption and desorption can be easily understood by the following scheme given in Figure 14. Table 6 presents the adsorption capacity for malachite green of various synthesized adsorbents with the Fe-Zn-PVA nanocomposites. It can easily be seen from the table that the nanocomposites were highly efficient in the adsorption of malachite green from the simulated wastewater. The mechanism of adsorption and desorption can be easily understood by the following scheme given in Figure 14.

Conclusions
In this study, green chemistry and nanotechnology were united in a manner by which Fe-Zn-PVA NCs were made using Mentha Piperita extract as a green source and then characterized using FTIR, SEM, EDS, and XRD. It was observed that the nanoparticles were homogeneous and between 80 and 100 nm in size. Response surface methodology was used to calculate the optimum operating parameters, i.e., 10.0 mg L −1 was the concentration of MG dye at a time of 8.0 min, pH 9.0, and 0.02 g of adsorbent amount, for the removal of malachite green dye. Under the optimum operational circumstances mentioned, the removal was discovered to be 77.87%. The Freundlich isotherm model and the pseudosecond-order model, respectively, were shown to be well matched to the equilibrium data by isotherm and kinetics analyses. Thermodynamic studies affirmed the spontaneity of the adsorption of MG dye over the surface of the green synthesized nanocomposites. A comparison with earlier research affirms that the nanocomposites are highly efficient in the adsorption of malachite green dye from aqueous medium. It has been determined that Fe-Zn-PVA NCs may be effectively used to remove dye from wastewater from the home and industrial sources.