Moringa oleifera seed husks for methylene blue dye adsorption: kinetic, equilibrium, and thermodynamic analyses

M. ole·ifera se·ed hu·sks (H-M·O) w·ere u·sed as adsorbent for the removal of methylene blue (MB) dye from an aqueous solution. The ads ·orbent was char·acterized by SE·M, ED·X, X·RD, F·TIR, B·ET, and p·HPZC. To evaluate adsorption capacity, the influences of p ·H, kinetics, isotherms, and thermodynamic prop·erties were anal·yzed. Characteri·zation tech·niques indic·ated that H-MO has het·erogeneous morp·hological charac·teristics wi·th a spe·cific sur·face a·rea of 3.24 m g, mai·nly comp·osed of ce·llulose distr·ibuted in an amor·phous stru·cture. MB adsorption was favorable at pH values higher than pHPZC of 6.6, using an adsorbent dosage of 1.0 g L . Equilibrium was achieved in the first 240 min, and the pseudo-second-order model was suitable for describing the kinetic data. Freu ·ndlich was the mo ·st adeq·uate model for descri·bing the isoth·erm cur·ves, pred·icting a max·imum adsor·ption capa·city of 122.7 mg g at 24°C, hig·her tha·n oth·er natu·ral ad·sorbents. The ads·orption proc·ess w·as spo·ntaneous and exoth·ermic, indic·ating that lo·wer temper·atures fav·or the ad·sorption and th·at it is cont ·rolled by p·hysical fo·rces. The da·ta presen·ted indic·ate the M·B has t·he pot·ential to be succ·essfully trea·ted by the eco-fr·iendly and l·ow-cost ads·orbent M. oleif·era seed h·usks.


INTRODUCTION
In the last decades, a growing interest in the use of alternative, low-cost adsorbents (LCAs) has been observed. LCAs are classified as natural materials (e.g., wood, coal, clays), industrial wastes/by-products (e.g., fly ash, red mud), and agricultural wastes/by-products (e.g., seed, seed husk, corn cob waste) (Gupta et al., 2009). In this sense, the seed husk of Moringa oleifera is included as agricultural waste. This plant has several applications in the food, pharmaceutical, cosmetics, and food nutrition industries due to its properties (Ueda Yamaguchi et al., 2021). Also, for many years, several studies have been investigating the potential of its seeds in drinking water treatment (Ribeiro et al., 2019), and in domestic (Vega Andrade et al., 2021) and industrial wastewater treatment (Villaseñor-Basulto et al., 2018) due to its coagulant properties. In addition to its use as a coagulant, M. oleifera seed has also been investigated as an alternative in the treatment of drinking water and wastewater as an adsorbent, for example, in the removal of heavy metals (Araújo et al., 2013) and dyes (Reck et al., 2018). However, the use of the M. oleifera seed generates the seed's husk waste. The potential advancement in the use of this seed as a coagulant and adsorbent on a commercial scale requires better management of its waste (husk) through the evaluation of possible scientific and technological applications.
The use of M. oleifera seed husk has not proved to be efficient as a coagulant (Kansal and Kumari, 2014). However, recent studies have been conducted to evaluate the possibility of using the husk as an adsorbent to remove pollutants from water, such as atrazine herbicide , acetaminophen analgesic , diuron pesticide , trihalomethanes (Okoya et al., 2020), and cyanobacterial hepatotoxin microcystin-LR (Warhurst et al., 1997). The capacity of M. oleifera seed husk to remove those pollutants shows its potential application as a LCA in the treatment of drinking water and wastewater, and new studies for removal of different problematic pollutants are required, such as dyes from textile wastewaters.
Dye industry effluents, when not properly treated, can cause an oxygen imbalance in ecosystems by dissolved oxygen depletion, in addition to preventing sunlight penetration, thereby altering the photosynthetic activity of the environment, which results in water quality deterioration and, consequently, in adverse effects to the local fauna and flora (Yagub et al., 2014). Methylene blue (MB) dye is a synthetic and cationic dye, from the group of thiazines, soluble in water and widely used in the textile industry to color silk, cotton, wool, leather, and paper coating. This dye shows toxicity such as teratogenicity, mutagenicity, neurotoxicity, nucleic acid damage, etc. (Sabnis, 2010) An aggravating factor is that dyes are resistant to degradation, and this property becomes an obstacle to effluent treatment, since they are substances that have highly stable chemical structures, making their removal difficult whether by biological, chemical, or physical process (Yagub et al., 2014). Several methods are used for the removal of dyes from wastewater, such as adsorption, coagulation, advanced oxidation, and membrane separation (Gupta et al., 2009). Many textile industries use commercial activated carbon for the treatment of dye waste, as adsorption is one of the most effective processes of advanced wastewater treatment (Yagub et al., 2014). Several types of adsorbents have been evaluated for MB dye removal as a cost-effective alternative to commercial activated carbons, such as the agricultural wastes/by-products LCAs apple pomace , Brazilian berry seeds (E. uniflora) (Georgin et al., 2020), and sorghum husk (Adeogun et al., 2019).
In this context, considering the need for better use of M. oleifera seed husk as a low-cost adsorbent from agricultural waste and the risks associated with the discharge of MB dye from an industrial effluent, a systematic investigation, including equilibrium, thermodynamics, kinetics, and mechanism of the adsorption of MB dye by M. oleifera seed husk, is a subject of interest and is discussed in this paper.

Adsorbent preparation and characterization
Moringa oleifera seeds were purchased from Arbocenter, and the seeds were harvested in the city of Araçatuba (20°56′19.72″ S, 50°40′6.17″ W), São Paulo -Brazil. First, the seeds were manually peeled. Seed kernels were segregated for future studies related to oil extraction and further use as a coagulant. The husks were then washed with deionized water and dried in a SolidSteel's 42L SSDc oven at 60°C for 24 h. After drying, the husks were ground using a commercial blender and sieved to be standardized on a maximum particle size of 600 µm. The powder obtained was used as adsorbent, and hereinafter will be called "H-MO".
The surface functional groups of H-MO were determined by Fourier Transform Infrared (FTIR) Spectroscopy (Shimadzu IRAffinity-1), in which the samples were scanned in the region of 4000 -650 cm -1 , in transmittance mode. The structural properties of the powder samples were determined by X-ray diffractometer (Rigaku Ultima IV), equipped with CuKα radiation (λ = 0.154 nm) at 40 kV, 30 mA and 2θ scan ranging from 5° to 80° with 2° min -1 step. Surface morphology was evaluated by scanning electron microscopy (SEM) (Tescan Vega 3), with 15.0 kV acceleration voltage. The energy dispersive X-ray spectroscopy (EDX) provided the elemental analysis of the H-MO through the SEM coupled with the EDX detector (Oxford X-act). For this analysis, the samples were previously covered with gold at a thickness of approximately 25 nm. The point of zero charge (pHPZC) was measured by the salt addition method with 0.1M NaNO3 (purity ≥ 98%, Êxodo), proposed by Bakatula et al. (2018). To determine the specific surface area, the BET method was used by adsorption/desorption isotherms of N2 at -196°C in a Nova 4200e (Quantachrome). Before the adsorption of N2, the samples were dried for 24 h at 60°C and subjected to degassing for 3 h at 90°C.

H-MO adsorption studies
The adsorption assays were conducted using methylene blue dye (MB) hydrate (purity ≥ 82%, Neon) as adsorbate. From a MB stock solution (1 g L -1 ), dilutions with demineralized water were made for the adsorption study. The experiments were conducted in triplicate in a batch mode on a thermostatic stirring bath (Dubnoff 304-TPA) at 100 rpm. After adsorption time, the solutions were then centrifuged (Excelsa II 206-BL) at 3,000 rpm for 5 min and the MB concentration was measured by UV-Vis spectrophotometer (Shimadzu UV 1800), operating at λmax = 664 nm wavelength. The percentage of dye removal (R, % -Equation 1), adsorption capacity at any time (qt, mg g −1 -Equation 2) and at equilibrium (qe, mg g −1 -Equation 3) were calculated as follows: Rev. Ambient. Água vol. 17 n. 1, e2812 -Taubaté 2022 Where C0 is the initial concentration of MB, Ct is the concentration of MB at any time, Ce is the equilibrium concentration of MB, V is the volume of solution, and m is the mass of the adsorbent.
The H-MO dosage effect was analyzed for the values of 0.2, 0.6, 1.0, 1.6, 2.0, 2.6, 3.0, and 4.0 g L -1 . For these assays, the pH was set at 7.0. The different H-MO dosages were put in contact with 50 mL of MB solution with an initial concentration of 50 mg L -1 . Since the equilibrium time was not determined yet, solutions were stirred during 1,200 min at 24°C. Based on the percentage of dye removal (R), the ideal dosage of adsorbent for the next tests was determined.
The pH effect test was conducted for the pH values of 5.0, 6.0, 7.0, 8.0, and 9.0, using 1.0 g L -1 of H-MO adsorbent, previously defined by the dosage effect. The pH adjustments were made with HNO3 0.1M (purity ≥ 66%, LS Chemicals) or NaOH 0.1M (purity 99.67%, Neon). H-MO adsorbent was added to 50 mL of MB solutions with initial concentration of 50 mg L -1 , and the solutions were stirred for 1,200 min at 24°C.
The kinetic adsorption experiments were conducted using the best conditions found in the pH (9.0) and dosage (1.0 g L -1 ) tests. H-MO adsorbent was added to 50 mL of MB solutions with an initial concentration of 50 mg L -1 , and the solutions were agitated at 24°C. Samples were collected at predefined times of 5, 20, 40, 60, 120, 240, 360, 480, 600, 720, and 1440 min. Through adsorption kinetics, the equilibrium time was determined, which was used in the adsorption isotherm test.

H-MO adsorbent characterization
The morphology H-MO's SEM images are shown in Figure 1, where non-uniform complex fiber matrix with no particular shape can be observed, in agreement with other authors (Araujo et al., 2018;Quesada et al., 2019). The overall H-MO chemical composition, as a result of semi-quantitative EDX spectrometry, presented a major presence of carbon (69.4% of weight) and oxygen (30.4% of weight), which are related to the natural organic structure of the lignocellulosic structure of the material, such as cellulose, hemicellulose, and lignin Garcia-Fayos et al., 2016). The N2 adsorption-desorption isotherm of H-MO adsorbent is presented in Figure 2, in which one can notice a typical behavior of a Type II isotherm with Type H3 hysteresis. Type II isotherms are given by the physisorption of most gases on nonporous or microporous adsorbents. The shape is the result of unrestricted monolayer-multilayer adsorption up to high P/P0. Since the curvature in Figure 2 is more gradual, this is an indication of a great amount of overlap of monolayer coverage and the onset of multilayer adsorption. When P/P0 = 1, the thickness of the adsorbed multilayer usually appears to increase without limit (Thommes et al., 2015). The open-loop presented in the N2 adsorption-desorption isotherm (Figure 2) is probably related to the pressure-dependent elastic deformation behavior, which is in accordance with low surface area adsorbents (Tang et al., 2017). The BET textural parameters of H-MO indicate a low specific surface area and total pore volume of 3.24 m 2 g -1 and 0.010 cm 3 g -1 , respectively. These results are expected, as the seed husks have not been activated, and are in accordance with the literature relating to in natura M. oleifera seed husks. Cusioli et al. (2019) reported 1.52 m 2 g -1 and 0.021 cm 3 g -1 , and Quesada et al. (2019) reported 1.2 m 2 g -1 and 0.002 cm 3 g -1 of specific surface area and total pore volume, respectively. The results are also in agreement with the behavior of a Type II isotherm, shown in Figure 2, of nonporous adsorbents.
The XRD pattern was conducted to determine the degree of crystalline or amorphous nature of the H-MO adsorbent and is presented in Figure 3(a). One can see, from the wide shape of the pattern, the material presents a predominance of an amorphous characteristic, suggesting a more disordered structure of the carbonaceous matrix. The main peak is detected at the 2θ value of around 22°, and the small peaks at around 16° and 35°, and they are attributed to crystalline cellulose (Barnette et al., 2012). Similar peaks have been also found for modified M. oleifera seed husk  and Brazilian berry seeds (Eugenia uniflora) (Georgin et al., 2020).
The FTIR spectrum of H-MO is shown in Figure 3(b), exhibiting characteristic bands of lignocellulosic materials (Li et al., 2018). The broad peak in 3370 cm -1 indicates hydroxyl bonds O-H. Smaller peak in 2922 cm -1 could describe symmetric and asymmetric C-H stretching vibrations. The 1638 cm -1 peak can appear due to the interactions between water and C=O or N-H stretching vibrations. Finally, the strong peak at 1032 cm -1 can be originated by C-O bonds. The FTIR spectra also agrees with the literature relating to in natura M. oleifera seed husks (Tavares et al., 2017;Garcia-Fayos et al. 2016). One can thus conclude that, according to the FTIR spectra found, the surface of the adsorbent possesses a wide variety of functional groups that favor the adsorptive process.

Effect of adsorbent dosage on adsorption capacity
The effect of H-MO adsorbent dosage on the percentage removal of the adsorbate MB is shown in Figure 4. Initially, the percentage of MB removal increases sharply as the adsorbent dosage increases. Such an increase might be attributed to the higher number of available adsorbent active sites for adsorption (Shah et al., 2015). A maximum removal value of 98% is achieved at a dosage of 4 g L -1 . However, considering that in the dosage of 1 g L -1 an MB removal greater than 90% was achieved, and that the increase in this dosage did not bring significant increases in removal, the dosage of 1 g L -1 of adsorbent is justified for economic purposes and was chosen for this study.

Point of zero charge (pHPZC) and effect of the solution pH on adsorption capacity
Figure 5(a) shows the influence of pH value on the adsorption of MB by H-MO. One can observe that pH values increase from 5 to 9 to enhance the adsorption capacity. This effect is a result of the MB characteristics, as well as the H-MO surface properties when submitted to solutions with different pH values. Figure 5(b) shows that the pHPZC of H-MO is 6.6. The value of pHPZC is an important property to understand the adsorption process, since when the solution is at a pH value above pHPZC, the functional groups of the adsorbent are deprotonated, resulting in a negatively charged surface, and contributing to the favorable adsorption of the cationic adsorbates. MB has a pKa value of 2.6, so MB dye is in the cationic form at pH value greater than 2.6 (Sabnis, 2010). This effect enhances the electrostatic attraction between the negatively charged surface of H-MO and positively charged MB dye. Considering the results obtained for the pH effect and pHPZC, the maximum value of adsorption capacity on this study is reached at pH 9 (qe = 52.02 mg g -1 ), and this pH value was set for the other tests.

Adsorption kinetics and modeling
Adsorption kinetics were determined to evaluate the effect of contact time on MB dye adsorption on H-MO adsorbent. To better understand the MB dye adsorption process, the kinetic models of pseudo-first-order (Lagergren, 1898) and pseudo-second-order (Blanchard et al., 1984) were fitted to the experimental data using nonlinear method (Equations 4 and 5, respectively) and are represented in Figure 6.
Where qt is the amount of adsorbate uptake per mass of adsorbent at time t, qe is the amount of adsorbate uptake per mass of adsorbent at equilibrium, k1 is the rate constant of the pseudofirst-order equation, and k2 is the rate constant of the pseudo-second-order equation. The adsorption kinetics presented in Figure 6 shows that H-MO adsorption capacity increases sharply at the beginning of the process, and it can be associated with the large number of active sites available. But the adsorption capacity reaches stability with a contact time of 240 min (qe = 48.6 mg g -1 ). This stability of qe values indicates that equilibrium was achieved. This may be explained by the fact that almost all empty sites of the H-MO surface adsorbed the MB molecules available on the solution, making it difficult for new MB molecules to be adsorbed.
Correlation coefficients and parameters of the kinetic models for MB dye adsorption on H-MO are presented in Table 1. The pseudo-second-order was better fitted to the experimental data, since the coefficient of determination (R 2 ) value was higher (0.946); the chi-square (X 2 ) value, related to the error, was lower (0.326); and the adsorption capacity calculated (48.4 mg g -1 ) strongly agreed with that obtained experimentally (48.6 mg g -1 ). This means it is likely that both external and intraparticle diffusion steps control the total kinetics of the adsorption process (Ho and McKay, 1999). The pseudo-second-order model is a good fit to the kinetics data for most adsorption processes involving dyes (Yagub et al., 2014).

Adsorption isotherms
The adsorption isotherms were determined to better understand the interaction mechanisms between MB dye and the H-MO adsorbent. The isotherm models of Langmuir (Langmuir, 1918) and Freundlich (Freundlich, 1906) were fitted to the experimental data using nonlinear methods (Equations 6 and 7, respectively) and are represented in Figure 7.
Where qe is the amount of adsorbate uptake at equilibrium, Ce is the equilibrium concentration of adsorbate, qmax is the maximum adsorption capacity, KL is adsorbent/adsorbate interaction constant, KF is Freundlich adsorption capacity constant, and 1/n is the Freundlich intensity parameter, which indicates the magnitude of the adsorption driving force or the surface heterogeneity.  The calculated parameters for such isotherms are shown in Table 2. The separation factor or equilibrium parameter RL is derived from the Langmuir model and can be calculated according to Equation 8 (Hall et al., 1966): Where KL is the Langmuir equilibrium constant, and C0 is the initial adsorbate concentration. The calculated values of RL are in the range of 0.06 to 0.56 (24°C), 0.07 to 0.59 (34°C), and 0.11 to 0.72 (44°C).
According to Figure 7, the adsorption capacity of MB dye increased with increasing equilibrium concentration (Ce), with the adsorption isotherms showing a typical concave shape. In addition to that, 0 < RL < 1 and 0 < 1/n < 1 ( Table 2). All these parameters indicate the adsorption is favorable for the studied temperatures (24°C, 34°C and 44°C).
The correlation coefficients and parameters of the adsorption models for MB dye adsorption on H-MO presented in Table 2 indicate that both models were well adjusted for the adsorption processes in the studied temperatures, with a coefficient of determination R 2 ranging from 0.97 to 0.99. However, the chi-square ( 2 ) values were lower from the Freundlich model (6.25 to 13.23) when compared to Langmuir model (11.29 to 21.98) for the three temperatures studied, indicating that the Freundlich model was better fitted to the experimental data. The Freundlich model assumes the adsorbent surface energy is heterogeneous with the strongest binding sites being occupied first and the binding force decreasing with the increasing degree of site occupancy (Febrianto et al., 2009). A better suitability of the Freundlich model compared to the Langmuir model for the adsorption of MB dye was also reported by Tuli et al. (2020) using activated tea waste carbon, and by Khodaie et al. (2013) using activated corn husk carbon.
From the temperature effect on the equilibrium curves shown in Figure 7 and Table 2, one may notice the temperature increase provided a little decrease in adsorption capacity (qmax), from 122.7 (24°C) to 119.8 mg g −1 (44°C).
The adsorption capacity of various adsorbents is commonly compared by the parameter qmax. Table 3 shows the maximum adsorption capacity of different adsorbents used for MB dye removal. The value obtained for qmax in this study is higher than those of other adsorbents produced with agricultural wastes. M. oleifera seed husk is shown to be a good adsorbent for MB dye, although the result for qmax is quite lower than those obtained with activated carbon. However, the adsorbent used in this study does not require high temperatures during its processing or activation steps, meaning that this material has a lower operating cost than activated carbon.

Thermodynamic parameters
Thermodynamic adsorption plays a key role in predicting viability, spontaneity, and adsorptive mechanisms (Tran et al., 2016). The thermodynamic parameters can be computed according to the laws of thermodynamics, using the following Equation 9: The relationship between the three thermodynamic parameters, Gibbs energy (∆°), enthalpy (ΔH°), and entropy (ΔS°) is described by Equation 10: Where R is the universal constant (8.314 J mol -1 K -1 ) and T is the solution temperature (K). Enthalpy (ΔH°) and entropy (ΔS°) changes were determined from the slope and intercept, respectively, of a plot of lnKC versus 1/T (Equation 11). Gibbs energy change (ΔG°) was directly calculated from Equation 9.
The dimensionless thermodynamic equilibrium constant KC in Equations 9 and 11 was calculated from the Freundlich adsorption isotherm constant KF, since this model was better fitted to the experimental data, through Equation 12 (Tran et al., 2016;Ghosal and Gupta, 2015).
Where is the density of pure water (assumed as ~ 1 g mL -1 ) and 1/n is the Freundlich intensity parameter.
The MB adsorption thermodynamic parameters onto H-MO are shown in Table 4. Evidence that suggests the appropriate application of the Freundlich constant KF in estimating the thermodynamic parameters is the high determination coefficient of the van't Hoff equation (R 2 = 0.991). A critical factor for determining the degree of spontaneity of the adsorption process is the Gibbs energy variation: the negative values of Gibbs energy (ΔG°) shown in Table 4 denote that the adsorption is energetically favorable; in other words, the process tends to be spontaneous. In addition, the decreasing of the magnitude of ΔG° with temperature means the process tends to a less thermodynamically favorable process at higher temperatures, which can be confirmed by the enthalpy value. Similar results were found for the adsorption of MB by apple pomace . The negative value of the enthalpy parameter (ΔH°) shown in Table 4 indicates the adsorption process is exothermic, which means the decrease in temperature results in an increase in adsorption capacity (qe) ( Table 2) and in the equilibrium constant (KC) (Table 4). Furthermore, the type of interaction can be assigned, to a certain extent, by the magnitude of enthalpy change. Physisorption, such as van der Waals interactions, is usually lower than 20 kJ mol -1 . Electrostatic interaction ranges from 20-80 kJ mol -1 and they are also frequently classified as physisorption, while a value in the range of 80-450 kJ mol -1 indicates a chemisorption (Alencar et al., 2012;Ghosh et al., 2021). Therefore, the typical physisorption enthalpy value for this study (ΔH° = -45.8 kJ mol -1 ) may be related to the sum of different molecular interactions, such as hydrogen bonding and electrostatic interactions between active sites on the adsorbent and the cationic dye. An exothermic adsorption process was also reported elsewhere (Ghosh et al., 2021). Finally, the negative sign in entropy (ΔS°) indicates the organization of the adsorbate at the solid/solution interface during the adsorption process becomes less random, also implying the adsorption phenomenon involves an associative mechanism. Figure 8 shows the proposed interaction between the MB and the H-MO. Based on the dataset discussed in this study, one could show that MB dye removal by H-MO occurs mainly due to electrostatic interactions, which is evidenced by the adsorption capacity enhancement in pH values above pHPZC. Similar conclusions were reported by other authors who investigated the adsorption of MB by agricultural wastes Ghosh et al., 2021). Additionally, the value found for ΔH° (-45.8 kJ mol -1 ) reinforces that these interactions are in the range of electrostatic interaction, also classified as physisorption, with no sharing or exchange of electrons. Moreover, H-MO composition presents several oxygenated functional groups, as presented by crystalline cellulose shown in XRD pattern (Figure 3(a)), and by O-H, C=O, N-H and C-O groups by FTIR spectrum (Figure 3(b)), which can also interact with MB molecules through dipole-dipole hydrogen bonds (Tran et al. 2017). The incidence of aromatic rings, both in the chemical structure of H-MO (cellulose) and in MB dye molecules, can also favor π-π interactions, contributing to the adsorption process (Cemin et al., 2021).

CONCLUSIONS
This study evaluated the adsorption capacity of methylene blue (MB) dye using Moringa oleifera seed husks (H-MO). The kinetic model that showed the best fit was the pseudo-secondorder model, and the isotherm that presented the best fit was a Freundlich model, indicating a favorable adsorption process (1/n from 0.60 to 0.63) with a maximum adsorption capacity of MB by H-MO of 122.7 mg g -1 at 24°C, higher than some recently published studies in the literature. The thermodynamic data indicated that the adsorption process is spontaneous (ΔG° from −3.88 to −1.06 kJ mol −1 ) and exothermic (ΔH° of -45.80 KJ mol −1 ), showing that lower temperatures favor MB adsorption by H-MO and that adsorption is controlled by physical forces. The disadvantage of the adsorbent preparation method of presenting low surface area (3.24 m 2 g -1 ) and total pore volume (0.010 cm 3 g -1 ) were expected, as the seed husks have not been previously activated, but were compensated by the electrostatic attraction at pH values above 6.6 (pHPZC) and by dipole-dipole hydrogen bonds and π-π interactions due to its surface composition, at any pH, resulting in a favorable adsorption process. Thus, from the results obtained, one can conclude that H-MO has promise for use in the adsorption process to remove MB in aqueous solution. Considering that seed husks are an agricultural waste, their use as