Non-functionalized oil palm waste-derived reduced graphene oxide for methylene blue removal: Isotherm, kinetics, thermodynamics, and mass transfer mechanism

2.1 Precursor preparation and synthesis of rGO

OPEFB was washed with distilled water several times. The washed OPEFB was then placed in an oven, heated to 400 °C under nitrogen flow conditions, and crushed later into powder (Nasir et al., 2017). The rGO synthesis method was adapted (Model: Bruker D2 PHASER, Massachusetts, United States of America).

2.2 Adsorption studies

The batch studies were carried out at room temperature. Sev- eral parameters, including adsorbent dosage (0.01–0.10 g), pH (3–11), initial concentration (5–25 mg/L), and contact time (20–100 mins), were studied to determine the optimum condition for MB removal. The corresponding weighed rGO is performed in a set of 100 mL of tightly sealed volumetric flask. The flasks were placed on the shaker at a constant speed of 120 rpm for a given time. Upon completion of the shaking process, each solution was filtered using filter paper to separate the adsor- bent from the solution. Each solution concentration was evalu- ated using a UV–vis spectrometer with a wavelength of 664 nm (Ahmed and Dhedan, 2012).

2.3 Equilibrium study

The adsorption data at a different initial concentration (5–25 mg/L) at 1 h from the batch studies were taken for the ad- sorption equilibrium study. Through the nonlinear fitting, the experimental points were plotted using Langmuir, Freundlich, Sips, and n-BET isotherm. The respective isotherms were de- termined using the curve fitting tool in MATLAB R2022A.

2.4 Kinetic study

The data of adsorption at different contact times (20–100 mins) was taken from the batch studies. The nonlinear forms of Pseudo-first order (PFO), Pseudo-second order (PSO), and Bangham kinetic models were plotted using the curve fitting tool provided in MATLAB R2022A.

2.5 Thermodynamics study

Similar steps in adsorption studies were applied for the ther- modynamics study by varying the temperature from 30 to 40 °C. The thermodynamic parameters were determined by fitting Ar- rhenius Equation into linear and nonlinear forms in Microsoft Excel: from (Sa et al., 2018; Somanathan et al., 2015; Yu and Le, 2018). 12.5 g of oil palm empty fruit bunch (OPEFB) biochar powder was mixed with 2.5 g ferrocene in an aluminum con– tainer before being carbonized and oxidized. The mixing was placed in a furnace and gradually heated up to 300 °C for 20 mins under nitrogen flow conditions. The resulting sample was cooled down to room temperature and weighed.

2.6 Characterization

Fourier transform infrared spectroscopy (FTIR) was per- formed to analyze the functional groups in the OPEFB biochar and synthesized rGO. The surface morphology was examined using scanning electron microscopy (SEM) (Model: Hitachi TM3000, Tokyo, Japan). Energy dispersive X-ray spectroscopy (EDX) was used to identify the elemental composition of the samples. The crystallographic structure of the materials was determined using X-ray powder diffraction spectroscopy (XRD).

Where k is the rate constant (dimensionless), A is the pre- exponential factor, E_a is the activation energy (KJ/mol), and R is the (Universal gas constant = 8.314 J/K mol), and T is the absolute temperature (in Kelvin).

2.7 Mass transfer mechanism

The intraparticle diffusion model was plotted using MAT- LAB R2022A curve fitting tool through the linear fitting. The model was assessed with the PFO, PSO, and Bangham kinetic models to interpret the mass transfer mechanism. The intra- particle diffusion model was plotted using the following equa- tion:

where k_p is the intraparticle diffusion constant (mg g⁻¹ min^−0.5).

2.8 Error analysis

The margin of error between theoretical and experimental data of nonlinear fittings was analyzed through the inclusion of error functions: squared estimated of errors (SSE) and root mean square error (RMSE). Marquardt’s percentage standard deviation (MPSD) was used in the determination of isotherm fittings, while kinetic models utilize the use of sum squares er- rors (ERRSQ) as the number of parameters function,p is un- available in kinetic models. The model equations are given in the supplementary information.

3.1 Characterization of OPEFB biochar and rGO

3.1.1

3.1.1 FTIR analysis

Fig. 1 indicates the spectrum peaks of OPEFB biochar and rGO. By comparing these spectrum peaks, three peaks have shifted and permanently exist for rGO at C⚌O stretching of amide (1556 cm⁻¹), C—N stretching of aliphatic amine (1030 cm⁻¹), and C—H bending of the alkyne (693 cm⁻¹). Untreated OPEFB biochar showed a much more complex spectrum sug-.

Close

Fig. 1

FTIR spectrum of OPEFB biochar and rGO.

Export to PPT

gesting the functional groups of lignocellulosic materials. A strong peak at peak 1740 cm⁻¹ indicated the C⚌O stretching of aldehyde. Moreover, peak 1576 cm⁻¹ showed the C⚌O stretch- ing of amide, while peak 1365 cm⁻¹ exhibited the existence of O—H stretching of hydroxyl phenol functional group. Peak 1216  cm⁻¹ indicates C—O stretching of ether. In addition, two peaks were observed at 1110 cm⁻¹ and 1020 cm⁻¹, indicating the ex- istence of the C—N stretching of aliphatic amine. The peak of 658 cm⁻¹ signals the presence of the alkyne group’s C—H bend- ing.

As the current study does not provide a comparison of peaks between GO and rGO, it is understandable for the material to have increased in the peak of the oxygen-containing functional group as the rGO undergoes the oxidation process and is re- duced by the ferrocene. The limited and low-intensity peaks compared to GO suggest a successful synthesis of rGO (Mahi- uddin and Ochiai, 2021; Sajab et al., 2018; Shen et al., 2012).

3.1.2

3.1.2 SEM analysis

Fig. 2 illustrates the OPEFB biochar and rGO morphol- ogy at 5,000x magnification. The porous structure does not diminish after carbonization, suggesting the pyrolysis possi- bly does not change the structure of the precursor much. The possible occurrence was the loss of water content and organic compounds through decomposition resulting in the mass loss (Abnisa et al., 2013). The crumpled thinner sheets in the rGO mainly caused by the reduction by ferrocene formed the dis- ordered structure of the rGO (Hidayah et al., 2017). The ob- served distorted structure on the surface prevents stacking of graphene layers by forming mesopores and macropores (Arias et al., 2020). Some impurities can be observed at the layers of the rGO, as confirmed through the EDX analysis (see Fig. 3).

Close

Fig. 2

SEM micrograph of (a) OPEFB biochar (b) rGO.

Export to PPT

Close

Fig. 3

EDX spectrum of (a) OPEFB biochar (b) rGO.

Export to PPT

The arrangement of the rGO can be seen as more parallel in arrangement compared to the biochar, indicating the pres- ence of stacking layers (Drewniak et al., 2016). The incom- plete stacking due to the nearby blocked pores, might result in restacking that eventually lead to the π-π interaction between rGO and the dye (Nasir et al., 2017).

3.1.3

3.1.3 EDX analysis

Fig. 4 illustrates the EDX analysis peaks. The carbon el- ement is higher for the rGO with up to 80 cps/eV (67 %) com- pared to the OPEFB biochar with around 70 cps/eV (66.2 %) due to the second carbonization, which increases the carbon content. The oxygen content is higher in the OPEFB biochar with 48 cps/eV (12.4 %) compared to 46 cps/eV (10.3 %) in the rGO, suggesting the reduction of oxygen-functioning groups.

Close

Fig. 4

XRD pattern of (a) OPEFB biochar (b) rGO.

Export to PPT

The presence of N, K, Fe, and Si are attributed to the nu- trients in the biochar through devolatilization via the pyroly- sis process (Claoston et al., 2014). The impurities are possibly present due to the reduction process as well (Narayanan et al., 2022). However, the presence of impurities does not interfere with the performance of rGO as the elements are completely liberated from the carbonaceous components in the rGO (Peng et al., 2016). (Sadhukhan et al., 2016) on their study also found several impurities in the rGO while obtaining high carbon ele- ment content with reduction lowering the oxygen content. The impurity may increase due to the trapping of atoms of elements such as potassium and sulfur due to the heating with nitrogen atmosphere (Nugraheni et al., 2015) that occurred both in py- rolysis and carbonization-oxidation process.

3.2 Effect of adsorbent dosage

In controlled conditions, Fig. 5 shows the MB removal percentage increased from 93.7 % to 98.78 % before deteriorat-ing to 79.89 %, with the increase in dosage from 0.01 g to 0.10 g. As the optimum dosage is 0.05 g with the highest MB removed; therefore, the dosage was selected for other parameters.

Close

Fig. 5

Effect of adsorbent dosage on the removal efficiency of MB (Initial dye concentration: 5 mg/L, Volume = 20 mL, T = 25 °C.

Export to PPT

The decreasing pattern after 0.05 g probably because when the dose of adsorbent exceeds the optimal value, the dye molecules cannot access the adsorption sites (Rajumon et al., 2019). Additionally, the aggregation of the adsorbent particles was induced by increasing the adsorption dosage, and the in- crease in screening effect might reduce the surface area and thus removal efficiency (Huo et al., 2015; Singh et al., 2021; Wang et al., 2014; Zhang et al., 2011).

Interestingly, several studies (Gilbert et al., 2011; Kumar et al., 2008; Pal and Maiti, 2020; Yeddou-Mezenner, 2010) have revealed that the adsorption efficiency decreased with an in- crease in the dosage of biosorbents. In correlation, rGO pro- duced from OPEFB in this study can be categorized as biosor- bent as it is produced from biomaterial. The particle aggrega- tion occurred due to the high dosage may result in aggregation of particles and inferential or repulsive forces between the bind- ing sites, reducing the interaction between the dye and biomass and reducing the adsorbent’s surface area.

3.3 Effect of initial concentration

Fig. 6 shows the intercorrelations between the effect of the initial concentration on the adsorption capacity and dye removal efficiency. As can be seen from the graph, the dye removal efficiency was demonstrated to reach the maximum value of 99.27 % at the lowest MB concentration of 5 mg/L and is the lowest with the value of 81.02 % at the highest concentration of 45 mg/L. It is hypothesized that the increase in MB concentra- tion decreases the removal efficiency due to the saturation of the active sites of the reduced graphene oxide (rGO) despite the in- crease in adsorption capacity, q_e (Azmi et al., 2016; Mahiuddin and Ochiai, 2021).

Close

Fig. 6

Effect of initial concentration on the removal of MB dye by rGO (rGO dosage: 0.05 g / 20 mL, T = 25 °C).

Export to PPT

At 100 mins mark, it can be assumed that all the dyes that the rGO could adsorb have been adsorbed (Gupta and Khatri, 2017). It happened due to the saturation at a higher concen- tration of MB. Therefore, as the initial concentration increases, the binding sites on the rGO are more occupied, leaving fewer unoccupied sites. The initial concentration provides the nec- essary force to overcome the resistance of interchanging dye molecules between the aqueous and solid phases (Azmi et al., 2016).

3.4 Effect of contact time

Data in Fig. 7 shows that the adsorption capacity at a given time increases rapidly with the increase in time at the first 20 min and increases slowly for the next few minutes before reaching the plateau. Interestingly, no significant adsorption occurred after 20 min for most initial concentrations, indi- cating the adsorption had reached equilibrium. Due to the early reach of the equilibrium due to the saturation limit, the adsorp- tion can be carried very fast in 20 min to obtain the maxi- mum adsorption. In correlation, as contact time increases, the equilibrium was approached, resulting in the adsorption process becoming less efficient, thus explaining the decrease in removal percentage.

Close

Fig. 7

Effect of contact time on the removal of MB dye by rGO (initial dye concentration: 5, 10, 15, 20, and 25 mg/L, rGO dosage: 0.05 g / 20 mL, T = 25 °C).

Export to PPT

3.5 Effect of pH

Fig. 8 shows the effect of pH of the initial MB solution for the adsorption of MB on rGO. In Figure 4.9, there is a clear trend of increasing removal efficiency with the increase in pH, with the highest percentage of removal at pH 11 with 95.41 % and the lowest at pH 3 with 78.85 %. The adsorption of cationic dye with rGO showed that adsorption occurred faster at higher pH, suggesting adsorption of MB on rGO occurs faster at neutral and alkaline pH (Gupta and Khatri, 2017). It might be that by decreasing the pH, more H⁺ ions need to compete with the cationic dye to be adsorbed by the rGO, decreasing ef- ficiency. At neutral and alkaline pH, the electrostatic attraction between the negatively charged rGO and the dye increase due to the fewer H + ions available, thus increasing the adsorption (Adel et al., 2021b).

Close

Fig. 8

Effect of pH on the removal of MB dye by rGO (initial dye concen- tration: 5 mg/L, rGO dosage: 0.05 mg / 20 mL, T = 25 °C).

Export to PPT

As MB is a basic thiazine dye with pKa of 2.6, MB ac- cepts protons under highly acidic conditions, making the MB di-cationic, thus increasing the MB’s polarity, leading to the low affinity between the di-cationic MB with the hydrophobic rGO surface, decreasing the efficiency. On the other hand, in the case of higher pH in neutral and alkaline conditions, the stable mono-cationic MB with moderate amphiphilicity will enhance the adsorption (Mahiuddin and Ochiai, 2021).

3.6 Adsorption isotherm

Adsorption isotherm models were used to predict the mech- anisms of adsorption processes in this study. Fig. 9 indicate the fittings, and Table 1 shows the Langmuir, Freundlich, Sips, and n-BET adsorption isotherm models with their respective parameters and error function values.

Close

Fig. 9

Fitting of adsorption isotherms models to experimental data.

Export to PPT

Through the fittings, the four-parameter n-BET model fit the adsorption experimental data from the batch studies, obtaining the highest value of R² and the lowest value of SSE, RMSE, and MPSD. The value indicates that the adsorption of MB can be described through multilayer adsorption with the assump- tion of five layers based on the n_BET value of 5.502. (Ahmad et al., 2021). The n_BET value of higher than 1.5 emphasizes that the adsorption occurred on the multilayer surface of rGO (Ahmad et al., 2020). The Sips value of n_s more than 1.0 de- scribes the heterogeneity of the rGO surfaces (Nethaji et al., 2013), thus, describing Freundlich rather than Langmuir, ex- plaining the higher R² and lower other error functions value of Freundlich compared to Langmuir. In addition, with the Sips favoring the Freundlich, the finding supports the multilayer ad- sorption assumed by the n-BET isotherm. Q_s is used as the maximum adsorption capacity with the value of 50.07 mg/g as Q_mBET value only presents the maximum adsorption capacity per layer. Table 2 provides the comparison of maximum ad- sorption capacity in this study with previous literature.

3.7 Adsorption kinetics

Table 3 shows the value of error functions and parameters for each kinetic model. The plots of PFO, PSO, and Bangham indicate that PSO fits the adsorption kinetics with the highest R² and the lowest SSE, RMSE, and ERRSQ values, closely fol- lowed by Bangham and PFO. However, to conclude that the adsorption is chemisorption through the determination of high fitting of PSO is misleading as the diffusion models and ther- modynamics parameters can better describe the physicochem- ical adsorption (Kajjumba et al., 2019). The good PFO fitting might be attributed to the irregular stacking of the rGO that is responsible for the physisorption. The good fitting of the Bang- ham kinetic model defines that the pores formed caused by the distorted structure through the reduction of oxygen functional groups play a major role in the adsorption process (Mezenner and Bensmaili, 2009).

3.8 Thermodynamics

Table 4 shows the thermodynamic parameters of the adsorp- tion. To analyze the energy changes due to the adsorption pro- cess, thermodynamic parameters, including Gibbs free energy change (ΔG^o), enthalpy change (ΔH^o), entropy change (ΔS^o), and activation energy (E_a^o) were determined. There is no real difference between results obtained from linear and nonlinear fitting, as Microsoft Excel could not fit the model data points into the value of E^o using the add-in solver, probably due to the restricted number of data points. The linear plotting was adapted from Van’t Hoff plots (Shikuku et al., 2018) by using 1/T (K) for the X-axis and ln k for the Y-axis, while for nonlin- ear fitting by plotting T (^oC) against rate constant, k.

The negative value of ΔG^o at each different temperature in- dicated the spontaneous nature of the adsorption of MB onto rGO. The values obtained are 0 to −20 KJ/mol suggesting the physisorption process (Arias et al., 2020; Shikuku et al., 2018).

The result further validated the Freundlich isotherm and PFO, which predicted the multilayer physisorption. The ΔG^o value increase with temperature rises, highlighting the exothermic na- ture of the reaction, which is unfavorable with temperature rise. The positive value of the entropy change (ΔS^o) emphasizes the increase in molecular randomness at the solid/liquid interface, thus supporting the alteration of the structure of the rGO dur- ing reduction (Arias et al., 2020; Rajumon et al., 2019; Shikuku et al., 2018). The negative value of ΔH^o supports the deter- mined ΔG^o value by confirming the exothermic reaction with negative enthalpy of −1.306 kJ/mol.

For the determination of E_a^o through Arrhenius plotting, the slope value determined through the linear plotting was mul- tiplied by the ideal gas constant value of 8.314 J/K mol to obtain the −1.306 kJ/mol. The obtained value lower than 40 kJ/mol suggests the physisorption process (Inglezakis and Zor- pas, 2012). The value of activation energy results in a nega- tive value because the breaking of bonds only takes place in the chemisorption process, as the activation energy is needed to break bonds that require higher temperatures, which is negligi- ble in the case of physisorption.

3.9 Mass transfer mechanism

Table 5 shows the intraparticle diffusion parameters. The in- traparticle diffusion low R² value does not qualify as the kinetic model to interpret the adsorption in this study. However, the obtained parameter C value of 0.916 (which is not zero) in intraparticle diffusion fitting is useful in indicating that the adsorption kinetics are not only controlled by intraparticle dif- fusion but involve complex mass transfer mechanism pathways (Ramesha et al., 2011). The increase in contact time of the ad- sorption contributed to the increase of the adsorption capacity of MB onto the rGO (Calimli et al., 2020). The K_p value of 0.5 > K greater than 0.1 interpret that the adsorption occurs strongly instan- taneously (Wu et al., 2009), in agreement with the result from batch studies.

The fitting of PFO signaled that the adsorption occurred due to a concentration difference attributed to the external coeffi- cient mass transfer. In addition, the negative activation energy indicates MB’s diffusion onto the surface of the rGO before mi- grating into the pores (Al-Ghouti et al., 2005), agreeing with Bangham kinetic model. The lower activation energy needed for the adsorption to happen instantaneously is supported by the parameter K_p value of the intraparticle diffusion model.

The physical adsorption behavior can be supported through the lower availability of H⁺ ions that directly enhance the ad- sorption by increasing the electrostatic interaction in the higher pH, thus proving that the rate-limiting step is diffusion con– trolled instead of chemical reactions (Adel et al., 2021a). The SEM characterization further suggested the involvement of Van der Waals (π-π) interactions between the rGO and dye due to the irregular stacking layers. Therefore, it can be assumed that it is an electrostatic and Van der Waals interaction, with no hint of hydrogen bonding (Wu et al., 2005), thus explaining the hy- drophobicity nature of rGO. Hence, these findings dismissed the significant role of functional groups on the rGO in the ad- sorption process.