İstilacı Centaurea Solstitialis Bitkisi Kullanılarak Sulu Çözeltilerden Malahit Yeşil Boya Giderimi ve Tepki Yüzey Yöntemi ile Optimizasyon: Kinetik, İzoterm ve Termodinamik Çalışma

Istilaci bitkiler, tarim urunlerinin gelisimini engelleyerek urun verimini azaltmaktadir. Bu calismada, sentetik bir boya olan Malachite Green (MG)'nin sulu cozeltiden uzaklastirilmasinda tarimsal atik olan istilaci Centaurea Solstitialis (CS) bitkisi adsorban olarak  kullanilmistir. Adsorbent - adsorbate iliskisi, spektrofotometrik yontem ve Fourier donusum infrared (FTIR) ile incelenmistir. Yuzey morfolojisi taramali elektron mikroskobu (SEM) ile belirlenmistir. CS yuzey alani Brunauer, Emmett ve Teller (BET) analizi ile olculmustur. Deney tasarimi ve modellemesinde RSM kullanilmis ve korelasyon faktoru 0,984 olarak bulunmustur. Optimum calisma kosulu olan 312.5 mg.L -1 MG iceren cozeltiye (pH=7.5) 0.325 g CS (30 mesh) koyularak 150 rpm’de 75 dakika calkalanmistir. Calismada CS tanecik boyutu 30 mesh ve calkalama hizi 150 rpm sabit tutulmustur. Optimum calisma kosullarinda CS’nin maksimum adsorpsiyon kapasitesi 90.816 mg.g -1 bulunmustur. Sonuclar Langmuir, Freundlich, Temkin ve Dubinin-Radushkevich (D-R) izotermleri ile degerlendirilmistir. Sonuc olarak adsorpsiyon isleminin Temkin izotermine (0,997 R 2 ) uyan yalanci ikinci dereceden (R 2 = 0.9982) bir kinetik reaksiyon oldugu bulunmustur. Intrapartikul difuzyon modeli, sirasiyla sinir difuzyon tabakasi ve partikul ici difuzyon etkilerini gostermistir. Adsorpsiyonun -4,47 KJ.mol -1 Gibbs serbest enerjisi ile endotermik oldugu bulunmustur. Desorpsiyon calismasinda maksimum %2.16’lik bir verim elde edilmistir. Bu sonuc CS uzerine adsorbe olan MG’nin tekrar suya karisma ihtimalinin dusuk oldugunu gostermektedir.  Daha sonra kolon calismalarina gecilmistir.  Kolonun verimliligi, 40. dakikada % 99.5'e ulasmis ve bundan sonra sabit kalmistir. Sonuc olarak, CS’nin, sulu cozeltiden MG gideriminde etkili bir adsorbent oldugu bulunmustur.


Introduction
Dyes are material used for coloring purposes. It can be synthetic or natural (Saxena and Raja 2014). Although Synthetic dyes are more economically and beneficiary than natural, the natural dyes are more favorable environmentally. Synthetic dyes may affect human (Bapat and Jaspal 2016) and the environment seriously (Robinson et al. 2001). Paper, hair color, food, plastic, and textile industries are examples for the dyes uses (Mittal et al. 2010). Dyes manufacturing has increased rapidly, with a production of 555 million tons/year for over 1000,000 types. Most of these types are related to water pollution (Mittal et al. 2009;Saravanan et al. 2016). In Turkey, The productions of dyes have reached 903 thousand tons for 2018 (Paint Industry in Turkey and the World 2018).
The green crystal powder dye named Malachite Green (MG) is soluble in water and belongs to tri phenyl methane category (Raval et al. 2016). It is not used in dying only, but it used as antiseptic, antiphrastic, antibacterial and antifungal. It can cause nervous system, skin, eye, kidney, and heart damages. So a proper treatment should be chosen to protect the human and the environment (Raval et al. 2017;Khan et al. 2011).
Several treatment methods were utilized to remove MG from aqueous solution. Levin et al. (2004) study the biodegradation of MG by white root fungi. The Malachite Green decolorization by Algal species was tested (Abd-El-Kareem and Taha, 2012). Man et al. (2012) examined the thermolysis -coagulation-flocculation process for MG removal. Chemical methods like ozonation and Fenton's reagent showed capabilities in treating MG (Zhou et al. 2013;Bai et al. 2013;Chen et al. 2002). Photocatalytic, photooxidative, and solar degradations were also examined (Hasanat et al. 2003;Yang et al. 2005;Aliyan et al. 2013;Bojinova and Dushkin 2011;Modirshahla and Behnajady 2006;Pirsaheb et al. 2015). Efficiency, engineering perspective, value, and effectiveness still have limited on the above techniques (Raval et al. 2017).
The adsorption process is the removing of the particles employing the transfer of mass from the liquid phase onto the solid phase (Crittenden et al. 2012). The process of adsorption is more favorable than the chemical methods, it is simpler and more flexible (Tobías et al. 2018). In addition to that, the adsorption process does not generate any sludge as in the biological process (Khamparia and Jaspal 2017). For these reasons, the adsorption process became the most frequent process.Activated carbon had been used widely as an adsorbent. Because of its cost, researchers attitude to use low-cost adsorbents (Gupta 2009;Salleh et al. 2011).
Time, pH, the concentration of dye, and adsorbent mass affect the adsorption process (Saleh et al. 2019). The relation between each factor and the adsorption process should be predicted. In the past, the effect of each factor was studied individually at the time (Gadekar and Ahammed 2019). Researchers had developed a collection of statistical and mathematical techniques called Response surface method (RSM) (Bezerraa et al. 2008). RSM had solved the problem by creating a model predicting the relation between them.
Many crops are cultivated in Mersin due to its location in the Mediterranean region, climate, soil structure, and geographical features. Due to the same reasons, there are many invasive plant species grows around the crops. These plants are no useful use. They are removed from agricultural areas and disposed of as waste. One of these is Centaurea solstitialis (CS). Farmers try to dispose of it by chemical and physical techniques. This study aims to investigate the ability of removal of MG from aqueous solution by CS plants and optimize the process by using RSM.

Dye Preparation
The organic chloride salt dye named Malachite green was purchased from Sigma-Aldrich. The cationic MG dye has a chemical formula of C23H25CIN2 (Figure2) and a molecular weight of 346.9 g. mol -1 . A stock solution with 1000 ppm was prepared and kept till the end of experiments. The stock was used to prepare the proposed concentrations. The desired pHs were modified using 0.1 M sodium hydroxide (NaOH) and 0.1 M hydrochloric acid (HCl). The concentration changes were obtained using UV-visible spectroscopy at a wavelength of 618 nm (Raval et al. 2017).  (Raval et al. 2017).

Adsorbent Characterization
Adsorbent and dye weights were measured by a precision scale (Uhaus Corp Adventure); the orbital shaker (Stuart Biolab) was used to mix the solution; pH meter (ISOLAB) was used to measure the Hydrogen ions concentration in the solution. The concentrations of dyes were measured using UV-Visible spectrophotometer (T 90); Fourier transform infrared spectrophotometer was utilized to determine the CS infrared spectra. In addition to that, the surface area and surface morphology were identified by BET analysis and scanning electron microscope (SEM) (Gemini Zeiss Supra 55).

Batch Study
In the batch study, the common affecting factors such as pH, the mass of adsorbent, the initial concentration and the contact time were investigated (Gupta et al. 2015). The volume of solution was 100 mL for all the experiment. Samples were drained into the 250 mL flask to avoid any missing in the solution quantity. The efficiency and the adsorbent capacity (qe) were calculated using equation 1 and equation 2, respectively (Bagheri et al. 2019): Where, Ci is the MG initial concentration (mg.L -1 ), Cf is the final concentration (mg.L -1 ); respectively. V is the solution volume (L); m is the CS mass (g).
To calculate the recovery percentage, desorption experiments were conducted by soaking the exhausted adsorbent in 0.1M sodium hydroxide (NaOH) and 0.1M hydrochloric acid (HCl). The solutions were waved at 150 rpm. And the recovery percentage was calculated as shown in equation 3.

Isotherms
To clarify the affinity between the adsorbate and the adsorbent, adsorption isotherms were performed by exposing 200 mg.L -1 of MG in 100 mL volume to various dosage of CS. Langmuir (Langmuir 1918), Freundlich (Freundlich 1906), Tempkin (Tempkin and Pyzhev 1940), and Dubinin-Radushkevich (D-R) (Dabrowski 2001) isotherms were studied. Langmuir describes the equilibrium between the solid surface and the liquid. In this case, it describes the equilibrium between the surface of CS and the MG. The formation of MG monolayer on the outer surface of the adsorbent CS was quantified. Equation 4 shows Langmuir isotherm linear form (Langmuir 1918). Freundlich isotherm (Equation 5) which is proposed originally as an empirical formula, describes the heterogonous of the CS (Freundlich 1906). Temkin isotherm describes the adsorption heat. It assumes that the heat of the adsorbent will decrease linearly for all molecules in the layer as shown in Equation 6 (Aharoni and Ungarish 1977). The distribution of Gaussian energy onto a heterogeneous surface was described by Dubinin-Radushkevich (D-R) as shown in Equation 7 (Gunay et al. 2007). The linear form of the used isotherm represented in Table 1.

Isotherm Linear form Equation Reference
Langmuir  CA is the concentration of adsorbate A at the equilibrium (mg.L -1 )  bA is the Langmuir constant for the adsorbate A (L.mg -1 )  QM maximum adsorbent capacity at saturation (mg.g -1 )  KA is the Freundlich adsorption capacity parameter (mg.g -1 ) (L.mg -1 ) 1/n  1/n is the intensity parameter  R is the universal gas constant (8.314J K -1 M -1 )  qs is the theoretical saturation capacity (mg.g -1 )  Kad is the isotherm constant (mol 2 .kj -2 )  T is the absolute temperature

Adsorption Kinetics
Kinetic studies were performed by analysing the sample over time. The sampling process continued until the concentration of the sample became closer. The samples were fitted by Lagergren's pseudo first order model (Pal and Pal, 2017) and pseudo second-order model (Robati, 2013) as shown in Equation 8 and 9 respectively.
Where qe and qt are the adsorbent (CS) capacity at the equilibrium and at time (t) respectively (mg.g -1 ). k1 is Lagergren's constant (min -1 ), k2 is the constant of second order rate (mg.g -1 .min -1 ). In addition to that, intraparticle diffusion model was plotted to as shown in equation 10 (Belhouchat and Zaghouane-Boudiaf 2017).
Where ki is the intraparticle diffusion rate constant (mg.g.min -1/2 ), C is a constant related to thickness of the boundary layer

Adsorption Thermodynamics
The feasibility, spontaneity, and randomness of the adsorption process can be identified by thermodynamics concepts. It realizes the relationship between the adsorbent and the adsorbate. Thermodynamics parameters were calculated as shown in Equation 11-13.
Where ∆G is the Gibbs free energy changes; ∆H and ∆S is the Enthalpy and Entropy respectively; b is the Langmuir constant. R is the universal gas constant (8.314 J.K -1 .M -1 ) and T refers to temperature in Kelvin.

Response Surface Method (RSM)
Response surface method was applied the same as explained in the previous study by Saleh et al. (2019). Time, pH, the concentration of MG, and the mass of CS effects were modelled by Design expert v11 program. The factors ranges were determined and entered into the model (Saleh et al. 2019). Table 2 shows the factors and level of interest.

Column Study
Column experiment is an important issue to decide if the adsorbent can be realized or not. The experiment was conducted in 25 cm long column with an internal diameter of 1cm. The adsorbent was inserted into the column for a height of 10 cm and compacted to fill the spaces. The column was fed from the bottom flowrate of 1.8 mL.min -1 by a peristaltic pump. The influent and the effluent were kept constant. The samples were collected from the top of the column. Every 5 mL the samples were analysed and the efficiency of the column system was calculated as shown in Equation 1. In the studies carried out with similar plants, the removal of dyes with concentratıon of 50 ppm from lıquıd solutıon was accomplished in column systems. 50 ppm as initial concentration was accepted and a column study was performed.

CS Characterization
In this study, the characterization of the CS adsorbent was carried out to understand the properties of the adsorbent. According to SEM results, the surface of the adsorbent can be considered cruelty, porous and heterogeneous. The SEM was carried to the sample before and after the adsorption process. Figure 3.a shows the raw CS while Figure3.b shows the CS after the adsorption. The surface area of the adsorbent CS was measured by BET analysis. The surface area and the Micro pore volume for the adsorbent before the adsorption process were found to be 4.97 m 2 .g -1 and 0.37 mm 3 .g -1 . This value is 2 times larger than Xanthium Italicum (Saleh et al. 2019). The surface area decreased at the end of the adsorption to reach 2.7 m 2 .g -1 . The adsorption process also affected the Micro pore volume since there was no Micro pore detected in the sample.
The Fourier transform infrared spectra for the adsorbent CS was determined using FTIR analysis. The peak related to the functional group of N-H stretching changed from the 3325.64 cm -1 to 3276.90 cm -1 . The intensity of these peaks increased from 30.45 to 33.14. The intensity of C-H stretching at band 2916.81 cm -1 has increased from 24.43 to 29.88. As pointed in Figure 4, the chemical structure of the adsorbent had been changed. Two new peaks were marked. The first peak related to the C-H stretching functional group was noticed at a band of 2849.31 cm -1 . The second was at the wavelength of 893.84 cm -1 related to the functional group of C-Cl stretching. Three peaks were not noticed after the adsorption process. Those were at the band of 593.97, 1157.08, 1714.41 cm -1 .Over the bands from 4000 -400 cm -1 , many peaks had been shifted. These results confirm the effect of the functional group. Since the structure of the adsorbent had been changed, the adsorption of MG onto CS can be considered chemisorption.

Response Surface Method (RSM)
The batch experiments were designed by RSM. The performance of MG removal from aqueous solutions by CS was modelled (Appendix A shows the experiments and their responses). The four factors effects were studied by Central Composite Design (CCD). The response from the CCD method ranges from 8.14-122.5. The ratio between the maximum and the minimum value was larger than 10, which required a transformation method. The inverse square root transform was selected. Box-Cox graph for power transformation was plotted to compare the current lambda with the best one. Box-Cox plot is shown in Figure 5.

Figure 5. Box-Cox plot for power transformation
The results also were fitted by linear, 2FI, Quadratic and cubic. Since Quadratic has the largest correlation factor, it was chosen to fit the data. Model statistic summary and Sequential Model Sum of Squares are shown in Appendix B.1 and B.2 respectively. Analysis of variance (ANOVA) for the quadratic model was carried out (Appendix B.3). The significant factors were determined via P and Fvalues. MG removal by CS model was significant since it has an F-value of 68.59. The P-value smaller than 0.01, thus the model is significant at a 99% confidence level (Roosta et al. 2014). In addition to that, the model parameters those have P-value smaller than 0.05 are significant. In this case A, C, D, AC, B², C², D² are significant model terms.
The developed model was used to predict the CS capacity response to the change in the defined parameters. The relationship between the factors and the capacity of the adsorbent is shown in equation 14. Where A, B, C, and D are pH, time (min), mass (g), and concentration (mg.L -1 ) respectively.
The regression coefficient (R 2 ) for the created model was found to be 0.9846. The difference between R 2 and the adjusted R 2 was 0.0143, which could be neglected. Same results were obtained by Khamparia and Jaspal (2017). Since the difference between the adjusted R 2 and the predicted R 2 was 0.0589, which is lesser than 0.2. Furthermore, the model showed a good coincidence between the observed and the predicted as shown in Figure 6 (Khamparia and Jaspal 2017).

Model Optimization
The model was optimized by changing the values of the parameters. The capacity of 90.816 mg.g -1 was achieved when MG solution of 312.5 mg.L -1 concentration was adsorbed onto 0.325 g CS for 75 min at a pH of 7.5. The desirability of these conditions reached 0.992.

Effect of Parameters on the Capacity
In this study, the range of pH was between 3 and 9. Because of the bonds between the surface of the CS and the Mg functional group, the CS capacity increases with the pH until pH =7.5. Sartape et al. (2017) used Limonia acidissima (wood apple) shell to remove MG from the aqueous solution and the maximum efficiency was when pH ranged between 7.5 and 8. The capacity of the CS has an inverse relationship with the mass of adsorbent. The maximum capacity obtained at 0.325g. Figure 7 shows the relationship between the pH, adsorbent mass and the capacity of CS.

Isotherm
Langmuir, Freundlich, Temkin, and Dubinin-Radushkevich (D-R) isotherms were plotted (as shown in Appendix C). The slope and y-axis intercept were determined to find the isotherms coefficients and summarized in Table 3. Temkin isotherm had the highest European Journal of Science and Technology e-ISSN: 2148-2683 regression value of 0.997, while the Langmuir isotherm had the lowest regression value of 0.697. Temkin isotherm had the best fit than the others. Multi-layers of MG adsorbed onto CS with the heterogeneous surface.

Kinetics
Lagergren Pseudo first-order and Pseudo second-order were plotted according to the Equations 8 and 9 as shown in Figures 9 and  10. The correlation factor of the linearity for each one was determined. The adsorption of MG onto CS was fitted to the Pseudo second-order. The correlation factor of the second order was 0.9982, and a rate constant of 0.0072 (gm.g -1 .min -1 ). The first-order kinetic had correlation factor of 0.9756, and a rate constant of 0.04145 min -1 . The intraparticle diffusion plots reflected the adsorption steps. It involved two steps; the first linear part comprehended the boundary diffusion layer, while the second portion introduced the intraparticle diffusion effects (Figure11).

Thermodynamics
Adsorption Thermodynamic was identified by the known parameters includes Gibbs free energy, enthalpy, and entropy. The parameters values were summarized as shown in Table 4. The Gibbs energy values for the temperatures 30, 35, 40 were -4.47,-5.66, and -6.46 KJ.mol -1 respectively. Although the Gibbs energy located in the range of 0-20KJ.mol -1 , the adsorption can be considered chemisorption. The negative sign also showed that the adsorption of MG onto CS was feasible and spontaneous in nature. From the enthalpy, the adsorption was endothermic since it had a positive value. The positive sign of entropy reflected the randomness at the solid phase interface had increased during the adsorption process.

Desorption
Desorption processes were conducted at the end of the adsorption process. The adsorbent was soaked into 0.1M HCL and 0.1 M NaOH solutions separately. It was noted that the adsorption process was irreversible. The max percentage of desorption process was 2.16 % at 180 min in acid batch. After this time, the adsorption process has started again.

Column Study
As mentioned previously, the column experiments were conducted to realize the adsorption process. A solution of 50 ppm was prepared and pumped from the bottom of the column with a flow of 1.8m L.min -1 . The samples were analysed Every 5 mL. The efficiency of the column system was determined as shown in Figure 12. The results show an opportunity to realize the adsorption process of MG onto CS. The first sample reached and collected after 35 min from the pumping process. The efficiency reached 99.5% after 40 min and stilled constant.

Previous Studies
Many researchers studied the removal of MG by low cost adsorbent. A comparison between them and proposed adsorbent in this study is shown in Table 5.

Conclusion
In this study, invasive CS plant which is agricultural waste was used as adsorbent for removal of MG from aqueous solution. The experiments were designed and modelled by RSM. The correlation factor of the developed model was 0.984. Temkin isotherm had the highest R 2 of 0.997. The kinetics models of the adsorption process were investigated and found to be fit with pseudo-second-order kinetic (R 2 =0.9982). The intraparticle diffusion model reflected the involved two steps; the first comprehended the boundary diffusion layer, while the second introduced the intraparticle diffusion effects. The adsorption was found to be endothermic. Although the Gibbs energy located in the range of 0-20 KJ.mol -1 , the adsorption can be considered chemisorption. It was noted that the adsorption process was irreversible with a max percentage of desorption process of 2.16%. Column experiment realized the adsorption process. The efficiency of the column reached 99.5% after 40 min and stilled constant.
Centaurea solstitialis (CS) was used successfully in adsorption of the Malachite Green dye. Reducing or eliminating the negative environmental impact of dye-containing wastewater is of great importance for the textile industry today. Chemical treatment is the most effective and common method used for removing dyes from wastewater. Due to the chemical content of the generated sludge from the treatment system, the development of more environmentally friendly technologies is an important research topic today. Adsorption is another effective dye removal process. One of the most important problems in this process is the elimination of the dyefilled adsorbent. One of the most important features of using plant origin materials as adsorbents is that the plant does not allow the adsorbed dyes to return to the water. In this study, the dye desorption of the CS plant was found to be very low. Also, the environment can be inoperative by adsorbent after the adsorption process by burning plant wastes in emission controlled industrial facilities. The used CS in this study is an invasive plant. After the removal of invasive plants from the agricultural fields, they are left to rot or burn around the field. Emissions from uncontrolled combustion in the open area contribute to air pollution. There is no use of CS as animal feed.