Preparation of sorbents

Both banana (Musa paradisiaca L.) and durian (Durio zibethinus M.) were obtained from a local market in Brunei. The fruits were peeled, and the peels were chopped into small pieces (less than 2 × 2 cm). They were then put in aluminum trays and dried in an oven at 40°C. After drying, they were ground into a coarse powder using a blender (Zojirushi Mill BM-RT08-GA), then sieved through a 212-µm stainless steel metal mesh to ensure that that the BP and DP have a similar particle size distribution. To prevent atmospheric water adsorption, both adsorbents were kept in sealed plastic bags before use.

Characterization of adsorbents

The functional surface groups of BP and DP were identified based on their vibrational spectra obtained using an FTIR Prestige-21 spectrometer (Shimadzu). The spectra were recorded using the KBR pellet technique over the full frequency range of the spectrometer (4000−400 cm^-1) to test for the presence of all possible organic functional groups on the adsorbent surface. The surface morphology of the adsorbents was assessed by SEM imaging on a JSM-7600F microscope (JEOL) operating at 5.0 kV. Their Brunauer–Emmett–Teller (BET) surface area and Barrett-Joyner-Halenda (BJH) pore volume were determined based on multipoint adsorption-desorption of nitrogen gas using an ASAP 2020 V4.02 surface analyzer (Micromeritics). The adsorption and desorption isotherms were carried out at 77.4 K. The process included degassing and adsorption of N₂ from 10⁻⁵ atm to atmospheric pressure. Finally, in order to observe the desorption isotherm process, the pressure was reduced from atmospheric pressure to high vacuum. The surface charge of the adsorbents was characterized by monitoring the change in pH of 20 mL (0.1 mol L^-1) KNO₃ solutions containing 200 mg adsorbent. The initial pH of the KNO₃ solutions was adjusted to be in the range of 2 to 12 by adding HCl or NaOH, and their final pH was measured after shaking the suspensions for 24 hours.

AB25 adsorption

AB25 (1-amino-4-anilino-9,10-dioxoanthracene-2-sulfonate sodium salt; C₂₀H₁₃N₂NaO₅S; CAS# 6408-78-2; MW 416.38 g mol^-1) was purchased from Sigma-Aldrich Co. Various concentrations of AB25 in distilled water were prepared by successive dilution of its 1 g L^-1 stock solution. Prior to adsorption experiments, absorption spectra of the prepared AB25 solutions were measured in the spectral range of 200 to 800 nm in a 1 cm quartz cuvette cell using a UV-visible spectrophotometer (UV-1900, Shimadzu).

The adsorption of AB25 on BP and DP was carried out using the batch method; where 100 mg of the adsorbents was suspended in 10 mL AB25 solutions (2.40 × 10^-4 mol L^-1) in several conical flasks. The suspensions were shaken using an orbital water bath shaker at 100 rpm. After a desired contact time, the adsorption process was terminated, the mixture was vacuum filtered through a filter paper, and absorption spectrum of the supernatant was measured. The initial concentration (C₀) and remaining concentration (C_t) of AB25 in the solution were accurately determined using the Beer-Lambert law based on the molar extinction coefficient of the synthetic dye at 600 nm being 1.05 × 10⁴ L mol^-1 cm^-1 (Guiso et al., 2014). The experimental conditions were then carefully controlled, and the effects of contact time (0–240 minutes), initial AB25 concentration (0.24 × 10^-4–4.80 × 10^-4 mol L^-1), the pH of the medium (pH 3–10) and its temperature (20°C–45°C) were systematically evaluated. In addition, the spent adsorbents were collected and dried for SEM and FTIR analyses.

Adsorption kinetics and isotherm

Based on the C₀ and Ct values of AB25, the adsorption capacity (Q_t) of the synthetic dye at different contact times was then calculated as;

(1)

Here, M is the mass of BP or DP in the suspensions, and V is the volume of AB25 solution. The adsorption kinetics were established by simulating the Q_t values with linear equations of the Lagergren pseudo-first order (Simonin, 2016) and pseudo-second order models (Robati, 2013), as given by

(2)

(3)

where k₁ and k₂ are the rate constants of the pseudo-first order and pseudo-second order kinetics, respectively, and Q_e is the adsorption capacity at equilibrium. In order to evaluate the diffusion mechanism and mass transfer processes, including intraparticle and film diffusions of AB25 onto BP and DP, the same Q_t data were simulated by the Weber–Morris intraparticle (Ho et al., 2000) and the Boyd diffusion models (Viegas et al., 2014) using;

(4)

(5)

where k_i represents the rate constant of intraparticle diffusion, C is the thickness of the boundary layer, and B_t is the Boyd constant.

In order to gain insight into the adsorption mechanism and the distribution of AB25 molecules adsorbed on the BP and DP surfaces, the Q_e values obtained at different initial AB25 concentrations were simulated using linear equations of the Langmuir (equation (6)), Freundlich (equation (7)), Temkin (equation (8)), and Dubinin-Radushkevich (equation (9)) isotherm models (Ghaffari et al., 2017), as given by

(6)

(7)

(8)

(9)

where, KL and KF are the Langmuir and Freundlich isotherm constants, n is the Freundlich exponent, K_T and b_t are the Temkin isotherm constants, β is the change in the Dubinin-Radushkevich mean free energy, R is the gas constant, and T is temperature.

The effect of temperature

The thermodynamics of the adsorption process were established based on the effect of temperature on the equilibrium. In this sense, the KL values obtained from the adsorption of AB25 at different temperatures were converted into dimensionless equilibrium constant ( ) values (Lima et al., 2019) using

(10)

where [AB25] is the concentration of AB25 solution which was in the order of 10^-4 mol L^-1, and the γ value is the activity coefficient which was assumed to be 1 because the solution is dilute.

Adsorbent regeneration

A 100 mL AB25 solution (2.40 × 10^-4 mol L^-1) containing suspensions of 1 g of BP or DP was shaken for 180 minutes at room temperature to generate adsorbent that is fully loaded with AB25. After filtration, the spent adsorbent was collected and dried. The spent adsorbent was then divided into three equal fractions, and each of them was soaked in different desorbing agents; 50 mL HCl (0.01 M, pH 2), 50 mL NaOH (0.01 M, pH 12), or 50 mL distilled water. These suspensions were then shaken for 120 minutes at room temperature. Next, the suspensions were vacuum filtrated, followed by drying. The regenerated adsorbents were then recycled and again used for the removal of AB25, following the procedures described in the previous Section. The adsorption-desorption process of AB25 was repeated and adsorption performance of the regenerated BP and DP were tested for several cycles.

Functional groups of BP and DP

Figure 1 shows the FTIR spectra of BP and DP before and after AB25 adsorption. For reference, the FTIR spectrum of AB25 is also shown. Before adsorption, the FTIR spectra of BP and DP had similar features, where the main bands were observed at 3400, 2922, 2849, 1733, and 1159 cm^-1 which have been assigned to the stretching vibrations of O–H, N–H, C–H, C=O, and C–O (alcohol) or C–N groups, respectively (Dinh et al., 2019; Jawad et al., 2018; Zaidi et al., 2018). There are several low intensity bands at 1648, 1433, and 1065 cm^-1 which could be attributed to non-aromatic or unconjugated C=C stretching vibration, C–H bending vibrations, and C–C vibrational modes of conjugated aromatic rings. The spectral bands at 1618 and 1381 cm^-1 have been assigned to the bending vibrations of aromatic C–H group and carboxylic acid O–H group (Dinh et al., 2019; Jawad et al., 2018; Zaidi et al., 2018).

Figure 1.

FTIR spectra of: (a) BP and (b) DP before () and after () AB25 adsorption, and (c) AB25 (Inset: the molecular structure of AB25).

The FTIR spectra therefore indicate that both BP and DP contain several functional groups with strong hydrogen bonding capabilities, such as COOH, OH, C=O, and NH₂ groups. These functional groups on the adsorbent surface can form dipolar and H-bonding interactions with C=O, S=O, and NH₂ groups of AB25, through which should make the adsorption of AB25 on the adsorbent surface very efficient. The similar spectral features of BP and DP suggested that these agricultural wastes contained the same functional groups, meaning that AB25 is probably adsorbed on both adsorbents in the same way.

The FTIR spectra of BP and DP were almost unchanged upon the adsorption of AB25, indicating that functional groups of the adsorbents remained intact. Nevertheless, several new minor peaks around 1200 to 1600 cm^-1 were clearly observed which could be assigned to either the vibrational bands of AB25 adsorbed on the adsorbent surfaces or the possible peak shift of the adsorbents upon AB25 adsorption. It is also noteworthy that, upon the adsorption of AB25, the OH vibrational band of BP at 3400 cm^-1 split into two bands at 3282 and 3582 cm^-1 due most likely to active participation of this functional group in AB25 adsorption. Overall, the FTIR spectra indicate that the negatively charged synthetic dye could easily bind to the adsorbents via hydrogen and dipolar bonding interactions.

Surface morphology and surface charge of BP and DP

SEM images (10,000× magnification) of BP and DP are shown in Figure 2. Due to differences in the roughness of the two adsorbents different working distances (WD) were used in order to control the depth of field so that the full field was imaged without aberration. However, the WD is related to the numerical aperture and therefore the resolution of the image. Therefore, for SEM images recorded at different WDs in the range of 0.1 to 0.6 mm (see Figure 2), the pore structures on the surface of BP and DP cannot be directly compared. Nevertheless, the SEM images provide the surface morphology before and after dye adsorption, as reflected by the line profile of a horizontal line-scans passing horizontally (x-axis) through the center of the images. Since the contrast in these images is obviously following the surface topography the line-scans are taken to indicate surface roughness. From these line-scans it is clearly observed that the surface roughness of BP was reduced after AB25 adsorption while for DP it increased after AB25 adsorption. These changes in the surface morphology were also revealed by their surface characteristics, where the BET surface area and BJH pore volume of BP was reduced whereas that of DP increased, as summarized in Table 1. The plot of N₂ adsorption and desorption isotherms, as shown in  Supplemental Figure S1, indicated that BP and DP contained macroporous structure.

Figure 2.

SEM images of (a) BP and (b) DP before (top row) and after AB25 adsorption (bottom row). All images were recorded at the same magnification (10,000×). The scale bars represent 3 µm. The line profiles on top of each image represent horizontal line scans from edge to edge passing through the center of the image.

Table 1.

The BET Surface Area and Pore Size of BP and DP Before and After AB25 Adsorption.

The surface charge characteristic of BP and DP was revealed by plotting their ΔpH versus the initial pH of KNO₃ solution, as shown in  Supplemental Figure S2. The x-intercept of the plots suggested that pH_pzc values of BP and DP were 6.32 and 6.18, respectively, which were within the values reported for a variety of agricultural wastes (Pathak et al., 2017). This finding supports the conclusion that the adsorbents derived from agricultural wastes, in general, have similar functional groups. With respect to the pH_pzc values, under the ambient experimental conditions (pH 6.7), the BP and DP surfaces will be negatively-charged due to the deprotonation of their carboxylic acid groups.

AB25 adsorption on BP and DP

Although the adsorption capacity of AB25 in the presence of BP and DP should be affected by the pH of the medium, the adsorption kinetics and isotherm of the synthetic dye were evaluated at ambient pH 6.7. Figure 3 shows the absorption spectra of AB25 solution at different contact times from 0 to 240 minutes. The decrease in absorption intensity resulted from the adsorption of AB25, as demonstrated by the time dependence of Q_t values in  Supplemental Figure S3. It is clearly observed that the adsorption capacities of AB25 on BP and DP increased with contact time and saturated at contact times longer than 150 minutes. At equilibrium, the Q_e values of AB25 on BP and DP were 4.07 and 4.49 mg g^-1, respectively. Figure 4(a) and (b) show the adsorption kinetic data, represented by the Q_t values, simulated using the Lagergren pseudo-first order and pseudo-second order kinetic models. As summarized in Table 2, the regression coefficient (R₂) values of the two kinetics model were 0.519 and 0.999 for the adsorption of AB25 on BP, and they were 0.627 and 0.999 for the adsorption of AB25 on DP. These results suggest that adsorption kinetics of AB25 on both adsorbents was pseudo-second order with the rate constant of AB25 on BP (0.163 g mg^-1 minutes^-1) being faster compared to that on DP (0.092 g mg^-1 minutes^-1). The pseudo-second order kinetics implies that the adsorption of AB25 on BP and DP was controlled by chemical interactions (Hubbe et al., 2019), as was also suggested by the FTIR spectral analyses in the previous section.

Figure 3.

Absorption spectra of 2.40 × 10^-4 mol L^-1 AB25 solutions in the presence of (a) 100 mg of BP and (b) 100 mg of DP at different contact times.

Figure 4.

Adsorption kinetic data of 2.40 × 10^-4 mol L^-1 AB25 solutions in the presence of 100 mg of BP (○) and DP (•) simulated with (a) the Lagergren pseudo-first order, (b) pseudo-second order, (c) the Weber–Morris, and (d) the Boyd model.

Table 2.

The Adsorption Kinetics Parameters of Lagergren Pseudo-First Order and Pseudo-Second Order Kinetics Models on the Experimental Data of AB25 Adsorption on BP and DP.

Figure 4(c) shows the best simulation of the Weber–Morris intraparticle diffusion model on the adsorption kinetic data. The simulation plots showed two linear trends and one breakpoint, suggesting that the adsorption of AB25 occurred in a two-step intraparticle diffusion process. For BP, at early contact times, a fast intraparticle diffusion was observed with a rate of 0.246 mg g^-1 minutes^−1/2 due to external mass transfer of AB25 onto the BP. This was followed by slower pore-volume diffusion at the equilibrium phase of the reaction with the rate of 0.035 mg g^-1. Similarly, a two-step intraparticle diffusion was observed for DP, with the fast and slow diffusion rates being 0.368 and 0.046 mg g^-1 minutes^−1/2.

It is important to note that the intercept of the Weber–Morris plots deviates from the origin, suggesting the existence of boundary layer effect which could hinder the intraparticle diffusion (Fierro et al., 2008) and that the adsorption process was not solely controlled by the intraparticle diffusion (Dotto & Pinto, 2012). This conclusion was further supported by analyzing the adsorption kinetic data using the Boyd diffusion model in Figure 4(d), where the simulation plots revealed two linear trends with one breakpoint and the intercept of the simulation plots also deviated from the origin. Based on the Boyd diffusion criteria (El-Khaiary & Malash, 2011; Sharma & Das, 2013), these findings indicate that the adsorption of AB25 on BP and DP occurs with a combination of intraparticle and film diffusion mechanisms.

Adsorption isotherms

Figure 5 shows the adsorption isotherms of AB25 on BP and DP simulated using empirical isotherm models. The experimental data points were well simulated by both the Langmuir and Freundlich isotherm models (Figure 5(a) and (b)), whilst they deviated from the simulation plots of the Temkin and Dubinin–Radushkevich isotherm models (Figure 5(c) and (d)). As summarized in Table 3, the linear regression coefficient (R²) values of the fitting lines confirmed that the adsorption mechanism and distribution of AB25 molecules adsorbed on the BP and DP surfaces followed the Langmuir isotherm model. This suggests that AB25 molecules were adsorbed on active sites with homogeneous binding energy as a single monolayer on the adsorbent surface. Using the Dubinin–Radushkevich mean free energy, the average adsorption energy (E=(2β)−1/2) was estimated to be in the range of 3.78−3.80 kJ mol^-1. The low magnitude of the binding energy can be attributed to hydrogen bonding interactions dominating the process (Bhatt & Shah, 2015). This further supports the conclusions reached from assignments of the FTIR spectral bands in the previous section. The Langmuir isotherm model estimated the Q_m values of AB25 on BP and DP to be 70.0 and 89.7 mg g^-1, respectively.

Figure 5.

Adsorption isotherm data of AB25 solutions with the initial concentrations of 0.24 × 10^-4−4.80 × 10^-4 mol L^-1 in the presence of 100 mg of BP (○) and DP (•), simulated with: (a) Langmuir, (b) Freundlich, (c) Temkin, and (d) Dubinin-Radushkevich isotherm models. The solid line represents the best fits of the respective isotherm model.

Table 3.

The Parameters of Simulation Plots of Various Empirical Isotherm Models With Two Parameters on the Experimental Data.

Comparison of Q_m values for adsorption of AB25 on different adsorbents

As the adsorption behavior of AB25 should depend on the functional groups, surface area, and pore size of the adsorbents, it is interesting to compare the Q_M values of AB25 on different adsorbents. As listed in Table 4, the Q_m values of AB25 on BP and DP were much lower compared to those on modified activated carbon derived from egg shell (109.8 mg g^-1) (Tovar-Gómez et al., 2012), lychee peel (200.0 mg g^-1) (Bhatnagar & Minocha, 2010), tea waste activated carbon (203.3 mg g^-1) (Auta & Hameed, 2011), Penaeus indicus shell (415.3 mg g^-1) (Kousha et al., 2015), and pectin (719.4 mg g^-1) (Shahrin et al., 2021). However, they were comparable or slightly higher with those on several agricultural wastes such as peach seed (95.2 mg g^-1) (Kul et al., 2019), modified banana peel (89.5 mg g^-1) (Guiso et al., 2014), hazelnut shells (60.2 mg g^-1) (Ferrero, 2007), and Ficus rasemosa leaf powder (83.3 mg g^-1) (Jain & Gogate, 2017), and were much higher than adsorbents derived from other agricultural wastes as such as Azolla pinnata (50.5 mg g^-1) (Kooh et al., 2016), soybean waste (38.3 mg g^-1) (Kooh et al., 2016), walnut sawdust (37.0 mg g^-1) (Ferrero, 2007), rubber leaf powder (28.1 mg g^-1) (Khalid et al., 2015), Shorea dasyphylla sawdust (24.4 mg g^-1) (Hanafiah et al., 2012), cempedak durian peel (26.6 mg g^-1) (Dahri et al., 2016), pine sawdust (26.2) (Ferrero, 2007), oak sawdust (27.9 mg g^-1) (Ferrero, 2007), and rambutan seed (25.6 mg g^-1) (Lakkaboyana et al., 2018). These findings indicate that BP and DP were among the best of the potential agricultural wastes to remove negatively charged dyes from an aqueous solution. In addition, it is also interesting to compare the Q_m values of AB25 with those of other acidic synthetic dyes on BP and DP. It was found that the Q_m values of AB25 are higher than those of AG25 on DP (63.3 mg g^-1) (Hameed & Hakimi, 2008), AV54 on BP (36.5 mg g^-1) (Kumar et al., 2010), and CR on BP (1.7 mg g^-1) (Mondal & Kar, 2018). The higher adsorption efficiency of AB25 could be ascribed to its smaller molecular size compared to AG25, AV54, and CR, thus it experiences less steric hindrance to being adsorbed on the surface of the agricultural wastes.

Table 4.

A Comparison of Q_m Values of Various Types of Adsorbents in the Removal of AB25 From Aqueous Solution.

The effect of pH

In real application, the presence of other chemicals could change the pH of wastewater which would potentially affect the adsorption capacity of adsorbent for synthetic dyes. To simulate this condition, the effect of the pH of the medium on the adsorption of AB25 was evaluated in the range of pH 3 to 10. As shown in Figure 6, at pH 3, the Q_e values of AB25 on BP and DP were 4.69 and 5.03 mg g^-1, respectively, and the removal of the acidic synthetic dye decreased gradually with pH. It can be rationalized that, at pH level below pH_pzc, the functional groups on the adsorbent surface were either protonated or positively charged, while AB25 existed in its anionic form. In this sense, the adsorbent surface facilitated electrostatic and hydrogen bonding interactions. This generated better chemical anchors for the adsorption of AB25. As the medium pH was increased above pH_pzc, the Q_e values continued to decrease until it had decreased by approximately 21% at pH > 9. This is most likely due to changes in electronic state of AB25. The relatively small effect of medium pH on the dye removal efficiency suggests that hydrogen bonding was the predominant interaction responsible for the adsorption of AB25 on BP and DP. This is schematically illustrated in Figure 7. A similar adsorption mechanism has also been proposed for the adsorption of AB25 on adsorbents derived from different agricultural wastes, such as cempedak durian peel (Dahri et al., 2016), soybean waste (Kooh et al., 2016), and Azolla pinnata (Kooh et al., 2016).

Figure 6.

The effect of pH on the adsorption capacity (Q_e) value of 2.40 × 10^-4 mol L^-1 AB25 solutions in the presence of 100 mg of BP (○) and DP (•) after a contact time of 240 minutes.

Figure 7.

A schematic illustration of the proposed adsorption of AB25 on agricultural wastes.

Thermodynamics of the adsorption of AB25 on BP and DP

The adsorption of AB25 on BP and DP at different temperatures revealed that the K_L values, and hence values, increased with temperature. This finding implies that the adsorption process was more favorable at higher temperatures. The Gibbs free energy (ΔGads) was then estimated based on the Van’t Hoff equation;

(11)

It was found that ΔGads values were in the range of −14.66 to −14.74 kJ mol^-1 and −13.62 to −13.64 kJ mol^-1 for the adsorption of AB25 on BP and DP, respectively. The negative ΔGads values confirm that the adsorption process is spontaneous. From the linear plot of ΔGads values against Q_e, as shown  Supplemental Figure S4, the enthalpy (ΔH) values of the adsorption of AB25 on BP and DP were estimated to be −12.1 and −13.4 kJ mol^-1, while the entropy (ΔS) values of the adsorption process were found to be 8.02 and 0.82 kJ mol^-1 K^-1, respectively. These thermodynamic parameters show that the adsorption process is exothermic, and that irregularity on the surfaces of the adsorbents increased upon the adsorption of AB25 (Ghosh & Goswami, 2006; Jiménez-ángeles et al., 2017).

Regeneration of BP and DP

Desorption of AB25 from the spent BP and DP was achieved by washing them in acidic, basic, and aqueous solutions. The regenerated adsorbents were recycled for the removal of AB25 for a few adsorption-desorption cycles. Figure 8 shows that washing the spent adsorbents with acid was the most effective method to regenerate BP and DP, where the Q_m values of AB25 decreased by less than 25% after the sixth adsorption-desorption cycle. In comparison, washing the spent adsorbents with alkali solution abruptly decreased the Q_m value by 50% after the first cycle, and it continued to gradually decrease after subsequent cycles. Rinsing the spent adsorbents with water resulted in poorer adsorptive performance recovery, as the Q_e value continuously decreased by more than 85% after the sixth cycle. This result indicates that acid treatment efficiently desorbs AB25 from the spent BP and DP and revives the active sites on the adsorbent surfaces. It might be interesting to note that acid treatment has also been revealed to regenerate adsorbents derived from different types of agricultural wastes (Zaidi et al., 2018). We therefore conclude that acid treatment is the most promising method to refresh functional groups on the active sites of agricultural wastes after they have been occupied by synthetic dyes through hydrogen bonding interactions.

Figure 8.

The adsorption capacity (Q_e) value of 2.40 × 10^-4 mol L^-1 AB25 solutions in the presence of: (a) 100 mg of BP and (b) 100 mg of DP regenerated by different desorbing agents; that is, 0.01 M HCl (black bars), 0.01 M NaOH (white bars), and H₂O (gray bars), after a contact time of 240 minutes.