Dried agroforestry residues (mesquite wastes) were mixed with H₃PO₄ solution at 40 % of concentration. The moisture of the residue mixture with the chemical agent was removed in an oven (EHOSELFC-3000 serials) at 105 °C for 12 h. In a muffle (NEY-M325) the dried material was thermally degraded at 400 °C for 2 h and cooled down under ambient conditions in a desiccator (NALGENE, 5312-0230). Subsequently, the obtained agroforestry-activated carbon (AAC) underwent several washes with deionized water to remove the remaining acid. Finally, the ACC was dried at 110 °C all day; the final product was used for characterization and adsorption essays. Physical parameters such as the moisture and ash content of ACC were characterized according to Yokout et al. (2015). The analysis for identifying the presence of the different chemical groups for ACC was determined using the Fourier Transformed Infrared Spectroscopy (FTIR), and the accessory used to read the samples was iD1 for diffuse reflectance. The Brunauer-Emmett-Teller (BET) method was followed to determine the superficial area through an advanced analyzer (Quantachrome Autosorb-iQ-C, USA). Commercial Granular Activated Carbon (GAC) and Powdered Activated Carbon (PAC) were also purchased commercially to compare a low-cost adsorbent with an existing one.

Aniline was obtained from the effluent of an anaerobic reactor that operated with wastewater from synthetic dyes and which concentration can reach 40 mg/L of AN. Table 1 shows the general characteristics of aniline.

Table 1

Parameter	Molecular Weight	λ max	Molecular Formula	Chemical Structure
Aniline	93.13	198	C6H5NH2	IMAGEN

Where: Molecular weight (mg/ mol); λ_max (nm)_.

The adsorption studies were carried out under different AN concentrations (1, 5, and 10 mg/L), contact times (1, 2, 6, 12, and 24 h), and pH (2, 4, 6, and 8). The pH solution was modified according to the necessity by dropping 0.1 N HCL. Each flask in the experiment had a working volume of 100 mL of AN sample that was contacted with 0.5 g of AAC with a stirring (50 rpm). A UV-visible spectrophotometer quantified AN at 198 nm. Equation 1 was used to estimate the removal efficiency.

A N   R e m   ( % )   = C o - C 1 C o x 100 (1)

Where: AN Rem (%) is the removal efficiency, C_o (mg/L) is the initial AN concentration, and C₁ (mg/L) is the final concentration of AN.

Several models were suggested for analyzing equilibrium experiments. One of the most important is adsorption isotherms: Langmuir, Freundlich, and Temkin isotherms.

Freundlich isotherm is explaining as an adsorption heterogeneous, considering that exist various sites of adsorption with different quantities of heat energy (Asencios et al., 2022). The linearized equation is:

L o g   q e = L o g   K F + 1 n F L o g C e (2)

Where q_e (mg/g) is the amount of adsorbate that is adsorbed in the adsorbent. C_e is the concentration (mg/L) of adsorbate in the equilibrium time. K_F is the constant related to the adsorption capacity (mg/g), and 1/n_F is a constant associated with adsorption intensity.

The Langmuir isotherm is a valid theoretical model for adsorption in a monolayer on a completely homogeneous surface. It has a finite number of identical and specific adsorption sites and negligible interaction between the molecules (Figueroa et al., 2015).

The linearized equation is:

C e q e = 1 q m a x C e + 1 K L q m a x (3)

Where q_max is the maximum monolayer adsorption capacity (mg/g), K_L is the Langmuir constant (L/mg), C_e equilibrium concentration (mg/L), and q_e is the adsorption capacity at equilibrium (mg/g).

The constant separation factor or equilibrium parameter, R_L, is a Langmuir isotherm parameter defined by equation 4 (Hall et al., 1996; Webber et al., 1974). The value of R_L suggests that adsorption can be unfavorable (R_L > 1), linear (R_L= 1), favorable (0 < R_L < 1), or irreversible (R_L = 0).

R L = 1 1 + K L C o (4)

Where C_o (mg/L) is the initial dye concentration and K_L (L/mg) is the Langmuir constant.

Temkin isotherm assumes the heat of adsorption of all the molecules in the layer would decrease proportionally with the increase in the coverage of the adsorbent (Hammed and Rahman, 2008). The linear form of Temkin isotherm is:

q e = R T b T I n K T + R T b T I n C e (5)

Where R (8.314 J/mol∙K) is the ideal gas constant, T is the temperature (K), b_T is the Temkin constant associated with the heat of adsorption (J/mol), K_T is the equilibrium binding constant (L/g).

The results were expressed as mean ± standard deviation, and the isotherm equations were developed using Minitab Statistical 16 and Microsoft Excel software version 16. A regression adjustment was applied to find the best correlation coefficient close to 1.

The moisture and ash content parameters are important in adsorption efficiency (Tay et al., 2009). According to Castellar et al. (2019), the high ash percentage (≤ 30 %) would indicate an insufficient activation process. The AAC is in the range of other ACs (6.5 %). The moisture content of ACs varies widely ranging from 2 % to 90 %, and it is considered that values greater than 15 % decrease the adsorption capability. Finally, the yield parameter is fundamental in terms of costs since it implies the weight of activated carbon per weight of residues utilized for activation. The AAC showed a yield of 50 %, which is higher than others, such as those made from coffee (30 %) or bay leaf (29 %), but lower than that made from coconut (80 %) (Yunus et al., 2022).

In Figure 1, the main AAC functional groups are shown. The functional group regions show a broad peak around 3800-3600 cm^-1 due to -OH stretching vibrations (Shurvell et al., 2006). The signal at 2361.82 cm^-1 is attributed to the C-H stretching vibration (Varghese et al., 2024). The signals in a range of 1300 to 1200 cm^-1 were accredited to C=C aromatic stretching; according to the research of Tetteh et al. (2024), these bands also can appear in the range of 1600 to 1500 cm^-1. The C-C groups were identified at 1068.68 cm^-1. On the other hand, the bands at 700-900 cm^-1 indicated C-H bond vibrations and deformations in aromatic rings (Ghosh, 2020). Additionally, the characterization of the specific superficial area for the AAC resulted in 427.881 m²/g, which was favorable for the adsorption. Commercial activated carbons (PAC and GAC) had a surface area of 800 and 400 m²/g, respectively, according to Carbotecnia ®. The difference in the surface area values is attributed to the raw material and the use of chemical or physical activation methods (Munoz et al., 2003).

Figure 1

The Agroforestry Activated Carbon (AAC) shows a defined surface morphology represented by ordered fibers, which is presented in the image of SEM 1000 X (Figure 2). The arrangement of the fibers suggests an advantage for adsorption favored by the channels formed between fibers (Brasquet et al., 2020). According to Li et al. (2021), the benefits of this kind of arrangement include a large specific surface area, high adsorption rate and capacity, unique surface reactivity, and ease of synthesis. Additionally, Burchell et al. (2017), in their study, establish that the large micropore volume of fiber arrangement compared to GAC is the greatest advantage. This can be directly attributed to the fiber structure, which has a very low fraction of closed pores and a large fraction of open pores.

Figure 2

The pH is a fundamental parameter in the adsorption process (Kord Mostafapour et al., 2016). In Figure 3, this study obtained the maximum removal of AN at acidic pH (2 and 4). In contrast, preliminary analyses indicated that at pH ≥8 (results not shown), AN adsorption was not favored. In acidic conditions, the adsorption process increased its efficiency from 52 to ≥ 80 %), and this behavior was similar at each concentration studied, which was 1, 5, and 10 mg/L.

Figure 3

AN is a weak base of an anionic nature that can be transformed into an anilinium ion if it is positively charged under acidic conditions. In this regard, the PKa value of aniline is 4.6, so at acidic pH values, it is positively charged, which increases the H⁺concentration. Moreover, at alkaline pH, aniline adsorption reduces the competition of other anions, such as OH- (Kord Mostafapour et al., 2016). Sun et al. (2021) observed a similar behavior, performing adsorption tests with AC in a pH range between 2 to 12, noting that the adsorption capacity increased in the range of 2 to 5 units, while from pH 5, the capacity decreased by up to 10 %. On the other hand, Guo et al. (2009) conducted an AN adsorption test with starch-derived compounds; they observed that the adsorption capacities reached the highest value at about pH 4.

The experiment was performed with three different activated carbons: 1) AAC prepared during this study, 2) commercial granular activated carbon (GAC), and 3) powdered activated carbon (PAC). GAC and PAC were purchased to compare a low-cost adsorbent with an existing one on the market.

Table 2 shows that the GAC has removal efficiencies similar to commercial GAC, up to 5 mg/L of AN. On the other hand, one parameter that influenced the adsorption of AN was particle size. PAC used in the tests as a comparison was more efficient than GAC and AAC. At pH 2 and 4 (conditions at which the highest removal was achieved), these carbons showed a removal rate of about 89-90 %. Meanwhile, PAC achieved almost 94 % removal of AN. The above implies approximately 5 % higher. The same situation was repeated at a maximum concentration of 10 mg/L. Wu et al. (2013) carried out a study about the effect of particle size on AN adsorption with PAC in five size ranges from 6 to 220; these results showed that AN removal efficiency moderately increased as the particle size decrease, with PAC with powder characteristics obtaining better efficiencies (≥ 85 %) at the maximum concentration (10 mg/L).

Table 2

Adsorbent	pH	Co	C1	Removal (%)	Variation coefficient (%)
AAC	6	1.38	0.30	80.00±0.020	9.700
	4		0.65	52.99±0.12	17.76
	2		0.14	89.33 ±0.15	7.640
PAC	6	1.19	0.66	44.68 ± 0.09	13.08
	4		0.2	83.04 ± 0.13	63.98
	2		0.08	93.70 ± 0.01	8.49
GAC	6	1.52	0.5	66.84 ± 0.09	17.96
	4		0.32	79.05 ± 0.02	7.110
	2		0.15	89.79 ± 0.01	0.740
AAC	6	5.34	1.82	65.91 ± 0.091	10.10
	4		Undetectable	>90 ± 0.28	9.40
	2		Undetectable	>90 ± 0.028	12.10
PAC	6	5.91	0.99	82.82 ± 0.14	14.05
	4		Undetectable	>90 ± 0.03	2.7
	2		Undetectable	82.96 ± 0.03	2.7
GAC	6	5.86	1.01	82.76 ± 0.43	9.21
	4		Undetectable	>90 ± 0.52	7.21
	2		Undetectable	>90 ± 1.21	13.23
AAC	6	10.72	6.23	41.88 ± 1.05	18.21
	4		5.54	48.23 ± 1.01	10.54
	2		4.32	59.70 ± 0.98	7.52
PAC	6	9.96	2.70	72.82 ± 0.14	14.05
	4		1.69	82.96 ± 0.03	2.7
	2		0.98	89.96 ± 0.03	2.7
GAC	6	9.11	4.52	50.38 ± 0.52	6.21
	4		2.21	75.74 ±0.56	5.41
	2		1.33	85.40 ± 0.46	3.21

Where C_o is the initial concentration of AN (mg/L) and C₁ is the final concentration of AN (mg/L).

The Langmuir, Freundlich, and Temkin isotherms were studied to define the mechanisms of AN adsorption on AAC and commercial activated carbons (PAC and GAC) at equilibrium. Once the experimental conditions for the adsorption process were obtained, they were adjusted to the indicated models. This research shows the fit at AN concentration range of 1 to 10 mg/L and pH 2. Figure 4 shows the result of the determination coefficient as an indicator of the fit, where the value of the R² was 0.9901, 0.94, and 0.99 for PAC, AAC, and GAC, respectively.

Figure 4

Langmuir adsorption isotherm is described by the interaction forces between the adsorbed molecules. Once they occupy a site within the adsorbent, no further adsorption occurs. In this case, the GAC and PAC fit in at a percentage of 99 %, while AAC showed a slightly lower correlation (94 %). The above refers to a homogeneous adsorption process with no in-plane adsorbate transmigration to the surface (Kumar et al., 2007; Foo et al., 2010). The model explains that adsorption takes place in a monolayer and that all sites of the adsorbent material are equal or equivalent. The commercial origin of these carbons can explain this. It can be inferred that PAC and GAC may be structurally similar, so there is no interaction between molecules adsorbed on neighboring sites. When an AN molecule occupies a site in the AC, it is no longer occupied by another molecule of the same origin.

An important parameter for fitting the Langmuir model isotherms is the separation factor (R_L) (Webber et al., 1974). The R_L value is useful to define if the adsorption is favorable, linear, not favorable or irreversible. When R_L is 1, the adsorption is linear; R_L between 0 and 1 is favorable adsorption; irreversible is when R_L is equal to 0, while values of R_L higher than 1 are unfavorable adsorption. In this study, the following values for GAC, PAC, and AAC were of 0.1814, 0.2857, and -1.044, respectively. The above indicates that each studied adsorbent carried out favorable adsorption of AN.

The Freundlich isotherm model assumes that adsorption occurs in multilayer through a non-uniform adsorption distribution on a heterogeneous surface (Hossein et al., 2013). Historically, Freundlich isotherm has been applied for the adsorption of mineral carbon, showing that the percentage of adsorbate in a mass of adsorbent is not constant at different points. In this perspective, the amount adsorbed is the sum of adsorption at all sites (each with a different binding energy), being stronger at the binding sites that are occupied first, until the adsorption energy is exponential and reduces at the end of the adsorption process (Vazquez et al., 2023).

The following images show the fits to the Freundlich models (Figure 5), which were less than 94 % of R².

Figure 5

According to Figure 5, AAC, PAC, and GAC show a smaller fit to the Freundlich model compared with Langmuir, suggesting a relationship that describes adsorption as a process that is not restricted to forming a multilayer. The results of a similar study reported by Vijan and Neagu (2012), are comparable to those of this work.

The Temkin isothermal model suggests that with increasing surface coverage, the heat of adsorption of all molecules in the layer decreases linearly (Temkin, 1940). The plot of q_e versus ln C_e allowed the isothermal constants k_T and b_T to be determined from the intercept and slope, respectively. The Temkin model presented a good fit with an R² close to unity for the three carbons evaluated (Figure 6). Therefore, it was one of the two models that best fit the experimental data of this study, both for the system where AAC was used and the commercial ones.

Figure 6

Table 3 shows the parameters of the isotherm mathematic models Langmuir, Freundlich, and Temkin for AN adsorption in AAC, PAC, and GAC. The coefficient of determination (R²) resulting from the Langmuir model for PAC and GAC were similar, with values of 0.9897 and 0.9893, respectively. In the case of ACC, the value of R² (0.9462) was lower than the others mentioned before. Considering the model of Freundlich, the R² values for ACC, PAC, and GAC were 0.5406, 0.9129, and 0.9309, indicating that the adsorption was not related to a heterogeneous or multilayer way. The model of Temkin resulted in similar values of R² compared to Langmuir: 0.9786 for ACC, 0.9754 for PAC, and 0.9792 for GAC.

Table 3

Isotherm model	Parameters	AAC	PAC	GAC
Langmuir	qmax (mg/g)	1.1675	1.2070	0.6467
	kL (L/mg)	0.4510	0.2499	-0.2139
	R2	0.9462	0.9897	0.9893
Freundlich	kF (mg/g)	0.1414	0.2131	0.8019
	1/nF	0.7657	0.0372	1.0712
	R2	0.5406	0.9129	0.9309
Temkin	kT (L/g)	0.0039	0.0107	0.0493
	bT (J/mol)	-8326.6	-7743.9	-2298.2
	R2	0.9786	0.9754	0.9792

The parameter q_max indicates the maximum adsorption capacity for each type of activated carbon studied. It should be noted that PAC is the one that achieved the highest adsorption capacity of 1.2070 mg/g, then ACC with 1.1675, and finally, GAC with 0.6467 mg/g. This is in accordance with the AN removal studies, which show that PAC is the one that achieves the highest removal efficiency with 98 % confidence. Vijan and Neagu (2012) studied the influence of -OH on activated carbon adsorption; the results of the equilibrium adsorption show that the isotherms were Langmuir type. The same behavior was reported by Santos and Rodrigues (2014), who studied the adsorption equilibrium of aniline onto activated carbon and found that a Langmuir and Bi-Langmuir model was the best representation. It should be recognized that the Langmuir isotherm assumes monolayer adsorption which proposes that there is no interaction between the adsorbed molecules, all of this on a surface with defined adsorption sites and uniform energy.

The utility of the Temkin isotherm is that it allows evaluating the kind of thermic reaction during the process of adsorption. When the constant b_T results are positive an exothermic adsorption process was developed, by the contrast if a negative value is obtained, the process suggested as endothermic (Benjelloun et al., 2021). The k_T and b_T values are shown in Table 3. Similar results were previously reported for activated carbon from Raphia taedigera seeds, where an endothermic adsorption process for removing methylene blue was described (Olasehinde et al., 2020).

Figure 7 shows a proposal for aniline mechanism of adsorption on activated carbons. It was inferred that the aniline molecules were adsorbed on activated carbon in a monolayer, favoring the order of the occupied sites in the adsorbent; this fact is related to the Langmuir model with the homogeneous kind of adsorption. The possible mechanism of adsorption was an endothermic nature; it is deduced that the adsorption reaction needs heat to be carried out, which was suggested by the Temkin model.

Figure 7