3.1 Characteristics of adsorbent
In Fig. 1(I) the X-ray diffraction spectrum is visible where the presence of broadband between 15 and 400 is observed, which indicates that no mineral phase is present. Therefore, the leaf powder has a highly amorphous structure. The amorphous phase is characterized by the broad peak that can be attributed to the lignin content [43]. Similar XRD patterns were also observed by other authors using structures of materials of natura plant origin [26, 44, 45].
In the FT-IR spectrophotometer of the leaf powder, several functional groups present on the surface of the adsorbent were observed. As can be seen in Fig. 1(II), the strong band at 3442 cm− 1 can be attributed to the vibrations of the hydroxyl group attached to the cellulose and lignin molecules or to the adsorbed water [32]. The bands at 2922 cm− 1 belong to the asymmetric C-H elongation of alkyl groups [46]. Bands at 1627 cm− 1 indicate C-O elongation of the carbonyl group [2]. The vibration frequencies in the region 1338 − 1058 cm− 1 represent the elongation of C-O belonging to alcohol, phenol, ethers, and carboxylic acid groups [33]. These findings suggest that the leaf powder contains a number of functional groups that are typical to lignocellulosic materials, which exhibit ion exchange capacity and general sorption properties derived from their polymeric components and structure. These polymers have sorption properties for a wide variety of metallic solutes and cations.
The surface micrograph of the material under different magnifications is shown in Fig. 1(III). In general, the samples have a morphology that consists of a smooth and fibrillar structure, without a defined geometry. In Figs. 1(III), C and D, the occurrence of cavities and pores that are favorable to adsorption, as they allow the penetration of dyes in the adsorbent structure can be observed. In the literature, it is common to find this type of morphology in plant materials in natural form [27, 47, 48].
3.2 Effects adsorbent dosage on removal of methylene blue
The effect of adsorbent dosage was studied ranging from 0.5 to 1.5 g.L− 1 (Fig. 2a). The first point to be analyzed is that increasing the dosage from 0.5 to 1.5 g.L− 1 caused a decrease in the adsorption capacity (qt) from 129.66 to 58.31 mg.g− 1. This is because, at higher doses, the adsorption sites aggregate, and the adsorbent is not fully utilized. Mathematically, this tendency can be easily explained by Eq. (1) [49]. Furthermore, the MB removal efficiency increased from 62.26 to 83.87% with increasing dosage, which can be attributed to the greater number of adsorption sites for MB [50]. Therefore, 0.8 g.L− 1 of foliar powder was selected as the optimal dosage, this being the intersection point of the curves in the graph.
3.3. pH effect and pHpzc on removal methylene blue adsorption
The effect of pH mainly depends on the ions available in the solution and the electrostatic interaction of the ions with the surface of the adsorbent influencing the binding sites of the adsorbents [51]. The point of zero charge (pHPZC) is the solution pH in which the material has zero surface charge. In this study, the adsorbent presented pHPZC = 5.9 (Fig. 2b). This indicates that at pH > pHPZC the adsorbent is in its ionic (negatively charged) form and, therefore, there is a better attraction for species with positive charges, for example MB. However, when pH < pHPZC the adsorbent is positively loaded, resulting in low affinity amid the adsorbent and the MB.
The influence of initial pH on MB removal was assessed varying the pH from 2 to 10 (Fig. 2c). A decrease in the adsorption rate was observed at lower pHs. This was because, at acidic pHs, ionic repulsion happens between the positively loaded adsorbent surface by H+ ions and the MB [52]. However, when the pH of the solution is neutral or alkaline, the MB removal rate is higher. As the pH increases, the adsorbent surface is deprotonated and negatively charged, resulting in electrostatic adsorption between the adsorbent and the MB, thus increasing the adsorption effect. [53]. Therefore, the chosen pH was 10 for a dosage of 0.8 g.L− 1 of adsorbent.
3.4 Adsorption kinetic profiles
Figure 3a presents the effect of contact time on the adsorption capacity (qt) of MB. The adsorption efficiency augmented fast in the initial minutes, starting to increase more slowly until reaching equilibrium before 60 min for practically all concentrations. This behavior that presents a fast initial rate followed by a slower rate is that in the initial stage there are more active sites available on the adsorbent surface, so the adsorption was easy and fast. After a period of time, the adsorption sites reduce and the MB concentration diminished. Then the adsorption efficiency gradually decline and dynamic equilibrium is reached [30]. Once the initial concentration augmented from 50 to 200 mg.L− 1, the adsorption capacity increased from 44.50 to 155.11 mg.g− 1. The higher initial adsorbate concentration increases the effective area of contact with the adsorbent and provides the driving force required to reduce the mass transfer resistance, which increases the adsorption capacity [54]. Similar behavior was reported by Neupane et al. (2014), on the adsorption of crystal violet powder from pineapple leaves.
To better understand and describe the adsorption of MB in the prepared leaf powder, the aforementioned kinetic models were used. The estimated parameters of each model are shown in Table 1. The choice of the best model was based on the statistical indicators and the physical meaning of the parameters. The highest correlation coefficient (R2 = 0.9907) and the lowest mean relative error (ARE = 2.72%) indicate that the Pseudo-second order model fitted better to the kinetic experimental data. In the literature, it is possible to find a multitude of works that present a good kinetic fit to this model, where many indicate that the material surface tends to be more heterogeneous [12, 55, 56]. Furthermore, the results show good agreement between the experimental data (qt) and the calculated values (qt) from the kinetic adjustment of MB by the pseudo-second-order model, with a maximum capacity of 150.96 mg.g− 1 at the highest concentration studied at 200 mg.L− 1. The Elovich model also presented with statistical indicators, being able to be used to represent the kinetic data. On the other hand, the pseudo-first order and Avrami models obtained the least statistical indices.
3.5 Equilibrium isotherms
The adsorption equilibrium isotherms were obtained at four temperatures (298-318K) as shown in Fig. 3b. Adsorption isotherms can be classified as L2 [57], as a plateau was formed, demonstrating the overload of adsorption sites. Instead, the adsorption capacity decreased with temperature increasing, reaching the maximum values (qe) at 298 K. For this same interval, the capacity decreased from 200 mg.g− 1 (298 K) to 174 mg g− 1 (328 K). Evidently, the adsorption capacity declines with the intensification in temperature, indicating an exothermic process. A similar trend was also found by other authors for MB removal [1, 52, 58–60].
Table 1
Kinetic parameters on the adsorptive removal of MB.
Models | C0 (mg L− 1) |
50 | 100 | 150 | 200 |
Pseudo-first order | | | | |
q1 (mg.g− 1) | 40.8833 | 94.1073 | 125.1999 | 141.5128 |
k1 (min− 1) | 0.2115 | 0.1805 | 0.1444 | 0.1649 |
R2 | 0.9715 | 0.9722 | 0.9590 | 0.9669 |
Adj. R2 | 0.9250 | 0.9269 | 0.8930 | 0.9131 |
ARE (%) | 5.2774 | 5.9366 | 7.3909 | 6.1702 |
MSE (mg.g− 1)² | 6.1029 | 31.8172 | 84.2856 | 86.7177 |
Pseudo-second order | | | | |
q2 (mg.g− 1) | 43.1257 | 99.9745 | 133.8046 | 151.0029 |
k2 (g.mg− 1.min− 1) | 0.0086 | 0.0030 | 0.0017 | 0.0018 |
R2 | 0.9947 | 0.9975 | 0.9913 | 0.9949 |
Adj. R2 | 0.9860 | 0.9934 | 0.9768 | 0.9864 |
ARE (%) | 2.2353 | 1.7596 | 3.0541 | 2.4209 |
MSE (mg.g− 1)² | 1.1256 | 2.8309 | 17.9925 | 13.3287 |
Elovich | | | | |
a (mg.g− 1.min− 1) | 2648.8677 | 1849.3463 | 729.0919 | 1286.1702 |
b (g.mg− 1) | 0.2652 | 0.1019 | 0.0664 | 0.0617 |
R2 | 0.9933 | 0.9889 | 0.9834 | 0.9922 |
Adj. R2 | 0.9822 | 0.9706 | 0.9562 | 0.9793 |
ARE (%) | 2.4542 | 3.2859 | 4.2631 | 3.0031 |
MSE (mg.g− 1)² | 1.4310 | 12.6897 | 34.0450 | 20.4252 |
Avrami | | | | |
qav (mg.g− 1) | 40.8933 | 94.03 | 125.06 | 141.44 |
kav (min− 1) | 0.4574 | 0.4281 | 0.3848 | 0.4084 |
nav | 0.4574 | 0.4281 | 0.3778 | 0.4084 |
R2 | 0.9715 | 0.9711 | 0.9564 | 0.9659 |
R2adj | 0.9620 | 0.9614 | 0.9419 | 0.9545 |
ARE (%) | 4.7105 | 5.2809 | 6.5832 | 6.3997 |
MSE (mg.g− 1)² | 4.7532 | 24.7843 | 65.5745 | 67.4978 |
qe (exp) (mg.g− 1) | 44.4964 | 99.4976 | 131.58 | 155.11 |
Table 2
Adsorption isotherm model parameter analysis.
Temperature (K) |
Model | 298 | 308 | 318 | 328 |
Langmuir | | | | |
qmL (mg.g− 1) | 225.8032 | 219.0591 | 209.6389 | 192.8882 |
kL (L.mg− 1) | 0.0266 | 0.0249 | 0.0235 | 0.0233 |
R2 | 0.9959 | 0.9944 | 0.9933 | 0.9957 |
R2adj | 0.9938 | 0.9916 | 0.9899 | 0.9936 |
ARE (%) | 5.0888 | 5.8622 | 6.2578 | 4.7974 |
MSR (mg.g− 1)² | 26.3937 | 33.3994 | 36.5577 | 19.7367 |
Freundlich | | | | |
kF ((mg.g− 1)(mg.L− 1)−1/nF) | 86.2242 | 82.9068 | 78.9530 | 73.4421 |
1/nF (dimensionless) | 0.1136 | 0.1136 | 0.1136 | 0.1136 |
R2 | 0.7580 | 0.7576 | 0.7618 | 0.7760 |
R2adj | 0.6369 | 0.6364 | 0.6427 | 0.6640 |
ARE (%) | 38.1269 | 37.5064 | 36.0013 | 33.3752 |
MSR (mg.g− 1)² | 1555.2233 | 1447.8469 | 1291.4651 | 1030.9600 |
Dubinin-Radushkevich | | | | |
qmDR (mg.g− 1) | 177.7004 | 170.3421 | 161.8381 | 148.6726 |
βx105 (kJ2.mol− 1) | 6.5791 | 6.2454 | 6.2444 | 5.7777 |
R2 | 0.9322 | 0.9363 | 0.9305 | 0.9288 |
R2adj | 0.8983 | 0.9045 | 0.8958 | 0.8931 |
ARE (%) | 19.6723 | 17.9983 | 17.7502 | 17.2337 |
MSR (mg.g− 1)² | 435.6814 | 380.4222 | 376.5513 | 327.8553 |
Tempkin | | | | |
a (mg.g− 1) | 52.9117 | 55.7229 | 59.8590 | 67.6834 |
b (L.mg− 1) | 0.2810 | 0.2560 | 0.2406 | 0.2447 |
R2 | 0.9657 | 0.9694 | 0.9757 | 0.9771 |
R2adj | 0.8870 | 0.8939 | 0.8842 | 0.8813 |
ARE (%) | 7.4658 | 7.6116 | 6.5362 | 6.0065 |
MSR (mg.g− 1)² | 112.9783 | 93.2060 | 66.1828 | 51.1229 |
To better understand the adsorption process, the isothermal data were fitted to the Langmuir, Freundlich, Dubinin-Radushkevich, and Tempkin models (Table 3). To determine the best model, the statistical coefficients were analyzed. In terms of best to smallest statistical fit, the Langmuir model was the one that best fitted the experimental data, followed by the Tempkin, Dubinin-Radushkevich, and Freundlich models. The Langmuir monolayer model showed high values of correlation coefficient (R2 > 0.9933), and adjusted coefficient of determination (R2adj > 0.9899) and lower values of ARE (< 6.2578%) and MSR (< 36.5577 mg.g− 1). On the other hand, the Freundlich model presented the lowest values of (R2 < 0.776), and (R2adj < 0.664) and the highest values of ARE (> 33.3752%) and MSR (> 1030.960 mg.g− 1). In addition, the Langmuir model presented capacity values close to those obtained experimentally, decreasing the capacity from 225.8032 to 192.8882 mg.g− 1, with the increase of the temperature in the system (298 K to 328 K). This model indicates that the accommodation of molecules on the surface of the adsorbate occurs through the formation of monolayers, distributed on surfaces with homogeneous sites[61].
Table S2 shows the comparison of the maximum adsorption capacity of the MB dye in several adsorbents present in the literature. The foliar powder of the species Sansevieria trifasciata showed an adsorption capacity close to that found in the literature. However, it is worth mentioning the great advantage of this material, because it does not present any chemical modification, and it comes from residual biomass.
3.6 Thermodynamic parameters
The thermodynamics parameters, standard change Gibbs (ΔG0, kJ.mol− 1), standard change enthalpy (ΔH0, kJ mol− 1), and change standard entropy ΔS0 (kJ.mol− 1.K− 1) were estimated from the equilibrium constants, and are displayed in Table S3. The values of ΔG0 and ΔH0 < 0 show that the adsorption process is spontaneous and exothermic, respectively (Mohammadi; Abdolvand, 2021). The magnitude of the ΔH0 value confirms the physisorption nature of the adsorption process (Pal et al., 2018). Furthermore, the positive values of ΔS0 indicate increased randomness in the solid-liquid interphase throughout the adsorption period. (Kundu, Mondal, 2019).
3.7 Simulation of dyeing effluent treatment
The leaf powder was analyzed in a synthetic mixture containing various dyes and salts (Table 1). In this study, 0.8 g.L− 1 of adsorbent was added to 50 mL of effluent under stirring at 150 rpm for 2 h at room temperature (298 K). The visible spectra of the effluent before and after the treatments are shown in Fig. 4. It is possible to verify that the spectra of the effluent after the treatment was strongly amortized, indicating an efficient color removal. The percentage of color removal from the simulated effluent was 51.5%. Therefore, it is possible to conclude that the foliar residues of the species studied are a low-cost material with great potential to be applied in the treatment of textile dye effluents.
3.8 Fixed bed adsorption performance
Seeking to analyze the behavior of the adsorbent in a fixed bed column, aiming its possible future application in real scale, continuous experiments were carried out. The results are shown in Fig. 10. The column can operate for 1430, 1130, and 525 min for concentrations of 100, 200, and 500 mg.L− 1, respectively, without adsorbent regeneration. In this case, until the break-up time, the fluid exiting the column had a clear color. After the break-up time, the color began to appear, until reaching equilibrium, in the region of the exhaustion time for each concentration. The relation between the increase of the breakthrough time with the diminishing of the initial concentration is related to the mass transfer effects on the fixed bed. Lower initial concentrations tend to generate lower driving force and low mass transfer rate, thus, requiring more time to achieve the equilibrium and leading to higher adsorption capacity [62].
The parameters for the dynamic models are showed in Table 3. The good fit with the experimental data is confimed by R2 of > 0.99829 and MSE of < 0.0106×10− 5. Thomas model forecast the adsorption capacity with an accuracy higher than 0.99829%. Also, the Yoon–Nelson model achieved an accuracy higher than 0.9983 regarding the τ. This shows that models can be used to describe the fixed bed adsorption of MB on the adsorbent. These results revealed that the waste from Sansevieria trifasciata is auspicious to treat dyed effluents in continuous operation mode.
Table 3
Parameters for the dynamic models.
Models | Initial concentration (mg.L-1) |
100 | 200 | 500 |
Thomas | | | |
kth (mL.mg-1.min-1) | 0.10918 | 0.0603 | 0.0604 |
qeq (mg.g-1) | 61.10165 | 188.3176 | 325.7264 |
qeq.exp (mg.g-1) | 60.48261 | 187.0147 | 322.6535 |
R2 | 0.99829 | 0.9995 | 0.9992 |
MSE | 0.00026 | 0.0001 | 0.0002 |
Yoon-Nelson | | | |
kyn (min-1) | 0.0109 | 0.0121 | 0.0302 |
τ (min) | 1222.0329 | 941.5880 | 434.3019 |
τexp (min) | 830.0000 | 780 | 380 |
R2 | 0.9983 | 0.9995 | 0.9992 |
MSE | 0.0106 | 0.0003 | 0.0007 |