2.1. Biomass (pineapple stubble)

The pineapple stubble sample was provided by the Bioenergetic and Environmental Research Unit (UIBEA) of the Distance State University (UNED). It is a sample composed of several sub-samples of stubble collected in the cantons of San Carlos (10 ° 35'53.6 "N 84 ° 29'59.5" W) and Los Chiles (10 ° 49'52.7 "N 84 ° 42'33.1" W) from Costa Rica, between the months of January-December 2018. It was preserved at a moisture level lower than 10%, in hermetic packaging, and out of the reach of light. Particle size was smaller than 8 mm. Moisture verification was carried out at (105 ± 3) ° C, for 4 h (^{Sluiter et al., 2008}).

2.2. Preparation of activated carbon

To produce activated carbon, three factors were studied: activating agent [zinc chloride (Sigma-Aldrich, 98%) at 30% m/m and phosphoric acid (Quimar, 85%) at 30% v/v], activating agent (mL), and precursor (g) ratio (2:1 and 4:1) and the use or not of a drying stage prior to carbonization (dried for 24 h (103 ± 3) °C). Subsequently, the impregnated sample was placed in a crucible with a lid and its carbonization was carried out in a muffle (Thermolyne 1500) at an activation temperature of 600° C for 1.5 h. Once the sample is cooled, it is removed from the muffle, and washes were performed with approximately 900 mL of distilled water. Finally, the activated carbon was dried for 1 h (103 ± 3) °C, had its mass weighed, and was stored under normal laboratory conditions inside a plastic container for later use.

A 2³ factorial design was proposed, with triplicates, from the factors and levels mentioned above. Table 1 shows the 2³ factorial design matrix with the variables and their respective levels.

Activated carbon production yield was chosen as the response variable. After performing the statistical analysis using the Statsolver software, the following equation was proposed for the model:

And the production yield was obtained using the following equation:

Where:

%Y: activated carbon yield, %

m_ac: final weight of activated carbon produced, g

m_rast: initial weight of precursor, g

2.3. Infrared spectroscopy

The structure of the activated carbons obtained was characterized. Then, it was compared to the infrared spectrum of untreated pineapple stubble and the differences were observed. The equipment used was a Nicolet iS50 Thermo Scientific FT-IR infrared spectrophotometer, sampling with ATR accessory with diamond cell. For determinations, a range between 400 cm^-1 and 4000 cm^-1, with a resolution of 4 cm^-1, was used. (^{Mahamad et al., 2015}) Per analysis, 48 scans were done. This analysis was carried out in the specialized laboratory Instrumental Analysis Nucleus (NIA) of the UNED.

2.4. Moisture and ash determination

Pineapple stubble moisture and ash percentages were determined using ASTM E1756-08 and ASTM E1755-01, respectively. For the activated carbons produced, the ASTM D286 7-09 and ASTM D2866-11 standards were respectively used.

2.5. Elemental analysis

The CHNS content of the samples was determined using Thermo Scientific Flash Smart equipment from the Chemical Engineering Laboratory of the University of Costa Rica (UCR), using a 3-point calibration curve and BBOT as standard. The O content was calculated by difference. Combustion took place at 950 °C and helium was used as stripping gas. The amount of sample for each analysis was between 3-4 mg and 8-10 mg of vanadium pentoxide was added to the sample as a catalyst. Analysis of a blank sample was done prior to building the calibration curve and analyzing the samples. Samples and catalysts were placed in tin capsules.

2.6. X-ray fluorescence analysis

X-ray fluorescence analysis was performed to determine the chemical composition of activated carbons. 8 grams of the activated carbon sample were weighed and placed in the sample holder, which was introduced into the Bruker model S8Tiger WDXRF analyzer, where the X-rays were irradiated under a helium atmosphere.

2.7. Nitrogen adsorption analysis

The BET surface area, total pore volume, and average diameter for activated carbons were determined using the Autosorb iQ gas sorption analyzer, version 5.20.17081, from the Center for Electrochemistry and Chemical Energy (CELEQ) of the University of Costa Rica (UCR). Nitrogen adsorption-desorption isotherms were carried out at 77.350 K.

2.8. Scanning electron microscopy (SEM)

The morphology of the activated carbon produced was visualized with a Scanning Electron Microscope (SEM) model S-3700N, serial number 371238-03, with a voltage of 15,000 V at the Center for Research in Microscopic Structures (CIEMIC) of the UCR. The samples were mounted on a metal base, using a double contact tape which was used to adhere the sample to the metal base plate. The sample was coated prior to SEM analysis with a layer of gold.

2.9. Methylene blue adsorption

A standard solution of methylene blue (C₁₆H₁₈N₃SCl, Merck) of a concentration of 1000 mg/L was prepared with distilled water. The standard solution was stored in an amber bottle at room temperature before being used. Concentration standards between 0 and 12 ppm were prepared from the standard solution, and the calibration curve of absorbance vs. concentration was obtained by UV-VS spectrophotometry (Turner SP-870) at 660 nm with an R² value of 0.998. Adsorption tests were performed using an initial concentration of methylene blue of 400 mg/L, 0.5 g of activated carbon, and 50 mL of solution (^{Mahamad et al., 2015}). These tests were held in a 250 mL beaker, using a magnetic stirrer template (Cole Palmer). To define the adsorption time (1 h) and the stirring speed (500 rpm), an initial test of adsorption capacity was done using the activated carbon produced from zinc chloride, at a 4:1 ratio and without drying. From this early result, it was built a curve of adsorption capacity of methylene blue q_f (mg/g) vs time (min).

To determine if there is a statistical difference concerning the adsorption of methylene blue between the different factors and their interrelationships, a 2³ factorial statistical design, with triplicates, was proposed for each of the different activated carbon treatments shown in Table 1. The removal percentage of methylene blue was chosen as the response variable. The following equation was obtained for the model:

The percentage removal of methylene blue was calculated as follows:

Where:

% RAM: percentage removal of methylene blue, %

Cn_i: initial concentration of methylene blue, mg/L

Cn_f: final concentration of methylene blue after adsorption, mg/L

2.10. Activated carbon depletion

To determine the depletion of activated carbon (without regeneration process) activated carbon produced through treatment 3 (ZnCl₂, at a 4:1 ratio and without drying) was used. For such, different reuse cycles of activated carbon were performed during the adsorption of a solution containing 400 ppm of methylene blue as initial concentration. The final concentration of methylene blue was measured at the end of each cycle. A maximum concentration of 5 ppm was used as reference since this is the maximum limit allowed by the legislation on the discharge and reuse of wastewater in Costa Rica for methylene blue active substances (^{MINAE, 2007}). Adsorption time was 1 h, stirring speed was 500 rpm, and 50 mL of solution and the same 0.5 g of activated carbon were used for each of the cycles.

2.11. Surfactant adsorption

Surfactant concentration was determined using the methodology of ^{Maldonado (2008)}, which is an adaptation of the technique described by the ^{American Public Health Association (APHA) & American Water Works Association (AWWA) (2005)}, specifically the 5540C standard for anionic surfactants as methylene blue active substances. Activated carbon produced with zinc chloride, activating agent (mL), and precursor (g) of 4:1 and without previous drying was tested. Sodium dodecyl sulfate anionic surfactant solutions of 10 mg/L and 200 mg/L were prepared to test the adsorption of the activated carbon produced (^{Braga & Varesche, 2014}; ^{Wiel-Shafran et al., 2006}). Similarly, the adsorption of residual gray water from a washing machine was carried out, after washing clothes using a commercial detergent. In both cases, the same adsorption conditions previously used for the adsorption of methylene blue were used.

3.1. Production yield of activated carbon

The highest yields concerning the production of activated carbon, and using H₃PO_4, were achieved with treatments 2 and 6, with yields of 70.3% and 70.7% respectively; while using ZnCl₂ as activating agent, the highest yields were achieved with treatments 3 and 7, with corresponding yields of 55.3% and 52.5% respectively. Average production performance and standard deviation results are shown in Figure 1:

Figure 1 Results of the average production yield of activated carbon. 

Statistical analysis of variance of the 23 factorial design concluded that only the chemical activating agent used is significant for the production yield of activated carbon (p <0.05). The use of H₃PO₄ leads to a higher production yield, corroborating previous studies (^{Daffalla et al., 2012}; ^{Liou, 2010}).

The higher yield provided using phosphoric acid (H₃PO₄) is due to the dehydrating effect it has on cellulose, hemicellulose, and lignin present in stubble, which occurs during the heat treatment (carbonization). In this stage, according to the previous research of ^{Molina-Sabio and Rodríguez-Reinoso (2004)}, crosslinking reactions, which provide mechanical stability and rigidity, are predominant; in addition, there is a subsequent reduction in the output of volatile organic matter and tars, which favors the production yield of activated carbon. Moreover, the lower acidity of zinc chloride, when compared to phosphoric acid, as well as the presence of phosphates, catalyze the degradation of lignin and cellulose, therefore leading to a higher yield when using phosphoric acid (^{Molina-Sabio & Rodríguez-Reinoso, 2004}). Besides, the residual insoluble phosphorus that remains on the surface of the activated carbon, along with the cellulose that is not so easily degraded, act as a recalcitrant material to hydrolysis, which could also explain the higher yields when using H₃PO₄ as an activating agent (^{Zubir & Zaini, 2020}).

3.2. Methylene blue adsorption

Figure 2 shows the results for the removal percentage of methylene blue present in the water after adsorption using the activated carbons produced., From the statistical analysis, it could be observed that the effects A (chemical activating agent used), B (impregnation ratio activating agent (mL): precursor (g)), and the AB interaction are significant for the adsorption of methylene blue (p <0.05).

Figure 2 Results of the average percentage of methylene blue adsorption removal. 

From the significance of the AB interaction between the activating agent and the impregnation ratio used, it is possible to conclude that the activated carbon that provides higher adsorption under the studied conditions corresponds to the activated carbon produced using zinc chloride as activating agent at an impregnation ratio of 4:1 (treatments 3 and 7).

The investigated variable concerning drying prior to carbonization was not significant for adsorption of methylene blue. Therefore, to reduce costs associated with activated carbon production and the time it takes, treatment 3, in which activated carbon is produced without drying, was determined the best one. On the other hand, for activated carbon produced using phosphoric acid as activating agent, treatment 4 presented the best results. It favored the adsorption of methylene blue and save costs associated with the application of drying prior to carbonization. Besides no drying step, this treatment used an activating agent (mL): precursor (g) ratio of 4:1. Therefore, characterization analysis focused on activated carbons produced by treatments 3 and 4.

The adsorption of methylene blue on activated carbon can be assumed to follow the following four steps: 1) migration of the dye from the solution to the surface of the adsorbent, 2) diffusion of the dye through the boundary layer to the surface of the adsorbent, 3) adsorption of the dye to the active sites on the surface of the adsorbent and 4) intraparticle diffusion of the dye into the pores of the adsorbent (^{Hock et al., 2018}).

The electrostatic attractions between the adsorbate and the adsorbent surface are a factor that can also affect the adsorption of methylene blue since methylene blue dissociates in water as a cationic molecule. Therefore, a cationic adsorbate will be more attracted to the surface of activated carbon when this surface is negatively charged, as is the case with activated carbons produced with zinc chloride and phosphoric acid (^{Altenor et al., 2009}).

Figure 3a shows the methylene blue solution with a concentration of 400 ppm and Figure 3b shows the adsorption effect of the activated carbon produced by treatment 3 (ZnCl₂, a ratio of 4:1 and without drying prior to carbonization).

Figure 3 (a) Solution before adsorption 400 ppm of methylene blue initial concentration. (b) Solution after adsorption with activated carbon produced by treatment 3 (ZnCl2, a ratio of 4:1 and without drying prior to carbonization). 

The difference of color present in the solution can be noticed, the initial concentration of methylene blue was 400 ppm, and after adsorption with the activated carbon produced by treatment 3 the methylene blue concentration was 0.179 ppm.

Table 2 presents a summary of the average percentages for production yield of activated carbon and methylene blue removal for each of the treatments performed.

Figure 4 shows the relationship obtained for the two main response variables evaluated in the experimental design: the production yield of activated carbon and methylene blue removal. Treatments corresponding to the use of H₃PO₄ as activating agent have a higher production yield, however, they resulted in a lower methylene blue removal performance. It can be seen, regarding the even treatments, that treatments 2 and 6 correspond to the use of H₃PO₄ as an activating agent and that an impregnation ratio of 2:1 promotes lower removal of methylene blue when compared to 4:1 ratios (treatments 4 and 8), using the same activating agent. Likewise, it is observed that the odd treatments, which correspond to the use of ZnCl₂ as activating agent, resulted in lower yields of activated carbon production; however, at the same time, they promoted higher methylene blue removal. Characterizations of the activated carbons, presented below, allowed a better understanding of the observed behaviors.

Figure 4 Correlation between the yield percentage and the percentage removal of methylene blue. 

Table 3 displays the adsorption results for the different activated carbon reuse cycles (depletion of activated carbon without regeneration) for treatment 3, with ZnCl₂ as activating agent, a ratio of 4:1, and without prior drying. Higher final concentrations of methylene blue were obtained each time by reusing activated carbon (after each adsorption cycle of 1 h) for adsorption of a new 400 ppm solution. This occurs due to more adsorption runs and an increase of competition between the methylene blue molecules, which leads to a decrease in the number of active sites (pores) available, thus hindering the methylene blue adsorption and, therefore, decreasing removal efficiency.

After three cycles, a final concentration of methylene blue of 3.68 ppm was obtained. It is below the value used as a reference which corresponds to the limit allowed by the legislation for active substances to methylene blue (5 ppm) in Costa Rica. After the fourth cycle, a concentration of 11.45 ppm was obtained, which exceeds the reference value of 5 ppm. It was observed a decrease in the adsorption capacity of activated carbon from 39.98 mg/g to 38.86 mg/g after 4 cycles of use without regeneration.

3.3. FTIR spectra for the produced activated carbon samples and the pineapple stubble

In Figure 5, the spectrum of raw pineapple stubble (purple-fuchsia color) is displayed, without any treatment. The other spectra correspond to the triplicates of treatment 1 (ZnCl₂ as activating agent, impregnation ratio of 2:1, and no drying prior to carbonization). It was observed that there is a notorious peak at 3309 cm^-1 for pineapple stubble, which corresponds to an O-H hydroxyl group. Close to that peak, another one can be seen, at 2917.15 cm^-1, which corresponds to a vibration of the C-H bond. It is noted that the spectrum of the produced activated carbons do not contain these peaks, indicating that the hydrogen present in the lignocellulosic matrix of the stubble is eliminated during activation and carbonization processes, a situation that was also observed in the research of ^{Mahamad et al. (2015)}.

Figure 5 FTIR spectrum of treatment 1 corresponding to ZnCl₂, a ratio of 2:1 and without drying. 

Furthermore, it is possible to observe characteristic peaks of lignocellulosic materials, such as the one at 1625.33 cm^-1, which is associated with aromatic skeletal vibrations mainly attributed to lignin. Peaks at approximately 1440 cm^-1, 1372.03 cm^-1, 1315 cm^-1, and others that are close to them are associated with torsional vibrations of the -CH₂, C-H, and C-O groups of the aromatic rings, respectively (^{Nicolaas et al., 2019}). The peak at 1244.72 cm^-1 belongs to the C - O stretching bond of lignin, while between 1030.08 cm^-1, and 1160 cm^-1, there are peaks corresponding to absorptions attributed to the deformation of the oscillating vibration C - H and to the stretching of the skeleton of the pyranose ring C-O-C (^{Nicolaas et al., 2019}). The high peak at 1030.08 cm^-1 can also be attributed to the C - O bonds, the C = C double bond, and the C - C-O stretch present in lignin, cellulose, and hemicellulose (^{Xu et al., 2013}).

It could be observed, from the spectra obtained for the replicas of this treatment, that almost all the previous peaks decreased. This indicates that the multiple bonds of the starting material were significantly eliminated. Therefore, the number of functional groups present in dried pineapple stubble is reduced after carbonization, due to the decomposition of the complex lignocellulosic matrix into simpler compounds or groups.

Similarly, to the spectrum observed for treatment 1, in Figure 6, which refers to treatment 2, it can be seen, for the replicas, that there is also a decrease in the peaks, compared to those for pineapple stubble. This implies a lower concentration of functional groups. Regions of multiple bonds and those corresponding to vibrations caused by sp² hybridization in C-H bonds were also eliminated. These results suggest that phosphoric acid serves well as an activating agent to produce activated carbon. When using both phosphoric acid and zinc chloride, decomposition of the lignocellulosic matrix was observed and, consequently, simpler carbonaceous structures were formed.

Figure 6 FTIR spectrum of treatment 2 corresponding to the use of H₃PO₄, a ratio of 2:1 and without drying. 

Since carbonization was held at 600 °C and for a longer time than in the previous investigation by ^{Mahamad et al. (2015)}, a decrease in the number of functional groups was already expected. This decomposition of the primary groups on the surface of activated carbon, at a high temperature, leads to the formation of new pores (^{Hock et al., 2018}).

3.4. Moisture and ash

Table 4 displays the moisture and ash results obtained for pineapple stubble and activated carbon (treatments 1 to 4) which are the treatments produced without drying prior to carbonization.

Activated carbons produced in the present study present high moisture levels when compared to other investigations, such as ^{Mahamad et al. (2015)}. This is mainly because, after carbonization and washing with distilled water, the activated carbons were left for drying for only one hour, at 103 °C. It is a short time, compared to other works. In these, after washing, activated carbon is left for drying for 6 hours or a whole day. However, it should be noted that this one-day drying time is costly, hence, to keep on working with the purpose of saving energy during the drying stage, time was set to only one hour and at 103 °C. However, the high levels of moisture did not affect the adsorption capacity of the activated carbon under the studied conditions.

For activated carbons produced with a lower impregnation ratio between activating agent (mL) and precursor (g) (2:1), higher amounts of ash were obtained, once compared to cases in which a ratio of 4:1 was used, demonstrating the effect of this ratio on the process. Higher volumes of the activating agent inhibit excessive degradation of the stubble since the stubble is impregnated to a greater extent. Therefore, carbonization is favored, resulting in less ash formation (^{Joseph et al., 2006}; ^{Yahya et al., 2015}).

It should be noted that treatment 2 (2:1 ratio using H₃PO₄) resulted in the highest levels of ash. Coincidentally, this treatment provided a higher production yield and a lower adsorption capacity. Other investigations have indicated that high levels of ash in the activated carbon decrease its adsorption: this occurs because the active sites are covered by the ashes, not allowing methylene blue molecules to adsorb on them (^{Ortega et al., 2017})

3.5. Elemental analysis

Table 5 displays the results of the elemental analysis for the activated carbons from treatment 3 (ZnCl₂, a ratio of 4:1 and without prior drying) and treatment 4 (H₃PO₄, a ratio of 4:1 and without prior drying). These treatments correspond to the produced activated carbons that presented the highest adsorption of methylene blue, for each of the activating agents used. Activated carbon produced by treatment 3 presents a higher carbon content than that of the precursor material. This indicates that after impregnation of the precursor with ZnCl₂, followed by activation, carbonization was favored.

The opposite occurred with activated carbon produced by treatment 4, which presents a lower carbon content after activation-carbonization. Such decrease might be due to the nature of the activating agent (H₃PO₄) and its interaction with the precursor material: the presence of heteroatoms in the biomass promotes the H₃PO₄ catalytic effect on the volatilization of the material and on the C-O-C and C-C bonds, resulting in the decrease of the carbon content (^{Ravichandran et al., 2018}). Decreases in the hydrogen and oxygen contents were also observed, after activation-carbonization of precursor material; this is associated with decreases in the number of functional groups and multiple bonds originally present in the complex lignocellulosic matrix of pineapple stubble and corroborates results obtained through the FTIR analysis.

3.6. X-ray fluorescence

Table 6 summarizes the surface chemical composition of the activated carbon in its oxide form. For activated carbon produced through treatment 3, ZnO predominates with 64.71%, followed by Cl with 8.11% and SiO₂ with 7.82%. Treatment 3 corresponds to the production of activated carbon using ZnCl₂ as activating agent, which explains why elements such as Zn and Cl predominate. Elements such as Al, Mg, P, Fe, and S in their oxide form are also present, in a range between 4.10 % to 3.36%, while traces of elements such as Ca, Ti, Cu, Mn, Cr, or Zr are also reported. The presence of these elements might be associated with the cultivation of pineapple, in which the quality and quantity of fertilizer used, influence the concentration of minerals absorbed by the plant. In addition, the way the soil is managed could also influence the presence of particles or minerals (^{Braga et al., 2015}).

Regarding activated carbon produced through treatment 4, it is observed that the predominant oxide is P₂O₅ (82.4%) followed by SiO₂ (11.2%). Since this activated carbon was produced using H₃PO₄, it was expected phosphorus, in its P₂O_5, to be the main component detected. Such high amount of P₂O₅ on the surface of the activated carbon could affect adsorption, resulting in a lower adsorption performance when compared to activated carbons produced using ZnCl₂. The research by ^{Zbair et al. (2018)} indicated that activated carbons with high phosphorus (P) content achieved lower adsorption yields of methylene blue than activated carbons with lower P content.

3.7. Nitrogen physisorption analysis

Nitrogen physisorption analysis was done for the corresponding samples of the treatments that provided the highest adsorption of methylene blue. These were treatments 3 and 4, both of which worked with an impregnation ratio of 4:1 and without drying prior to carbonization, the difference being the activating agent used: zinc chloride for treatment 3 and phosphoric acid for treatment 4.

Figure 7 shows the nitrogen adsorption isotherm obtained for the activated carbon produced by treatment 4 (H₃PO₄, a ratio of 4:1 and without drying). The red-colored curve represents adsorption, while the blue-colored curve, desorption. This isotherm tends to a type IV isotherm with a type H3 hysteresis cycle, according to the examples presented by IUPAC. This isotherm is characteristic of activated carbons with mesopores of disordered structure (^{Thommes et al., 2015}).

Figure 7 Nitrogen adsorption isotherm for the activated carbon sample produced with treatment 4 using H₃PO₄ as activating agent, at a ratio of 4:1. 

Figure 8 shows the isotherm for the activated carbon produced by treatment 3 (ZnCl₂, a ratio of 4:1 and without drying). This isotherm tends to a type IV isotherm with a type H4 hysteresis cycle, according to the examples presented by IUPAC (^{Thommes et al., 2015}). It is characteristic of activated carbons with micro and mesopore structures. It can be noticed, from the points close to a relative pressure of zero, that there is an initial volume for adsorption, which is indicative of the presence of micropores in the structure.

Figure 8 Nitrogen adsorption isotherm for the activated carbon sample produced with treatment 3 using ZnCl₂ as activating agent at a ratio of 4:1.  

The highest surface area, as well as the highest pore volume amongst the activated carbon produced, was achieved using zinc chloride as activating agent (Table 7). This indicates that, for this type of activated carbon, there is a high number of active sites in which methylene blue is retained. And it is the reason why activated carbons produced with zinc chloride provided better adsorption results than those produced using phosphoric acid.

The difference between the surface areas for each activated carbon produced using the activating agents H₃PO₄ and ZnCl₂ is related to the different activation mechanisms. It has been proposed that the activation of phosphoric acid is carried out through the formation of cross-linked structures in the form of phosphate esters, which are directly related to the development of porosity (^{Jagtoyen & Derbyshire, 1998}). At a temperature of 600 °C occurs a breakdown of these cross-linked structures, causing contraction and reduction in the porosity development (^{Donald et al., 2011}). In contrast, the activation mechanism of zinc chloride is believed to be more related to the "swelling" effect of the lignocellulosic matrix of pineapple stubble caused by ZnCl₂. This results in a breakdown of the lateral bonds of the molecules of cellulose and consequent creation of interspaces, allowing easier access of the activating agent to the interior of the biomass structure (^{Donald et al., 2011}). In other words, the activating agent acts as a template for the subsequent formation of microporosity, later obtained through the activated carbon washing, in which remnants of zinc chloride are eliminated, thus creating microporosity (^{Molina-Sabio & Rodríguez-Reinoso, 2004}).

This microporosity on the sample prepared with zinc chloride is related to the size of this salt molecule. However, this is not the case for phosphoric acid, since there is no specific molecule of H₃PO₄, but a mixture of molecules from H₃PO₄ and H₄P₂O₅ to H₁₃P₁₁O₃₄, in the proportion predicted by the liquid-vapor phase diagram for P₂O₅-H₂O mixtures, resulting in heterogeneity in terms of pore size (^{Molina-Sabio & Rodríguez-Reinoso, 2004}). Additionally, since phosphoric acid is a stronger acid than zinc chloride and considering that a 4:1 ratio is high, a more intense chemical attack occurs on the precursor, which contributes to the average diameter of the pores of this activated carbon: 8 nm, notably larger than that of the activated carbon sample prepared with zinc chloride (3.1 nm.).

Therefore, according to the results obtained in this analysis and in agreement with ^{Donald et al. (2011)}, it can be indicated that zinc chloride is more effective in the development of meso and microporous structures, which promote higher surface areas. Phosphoric acid, on the other hand, is more effective in the development of mesopores (disordered and heterogeneous), which lead to larger average pore diameters.

Table 8 compares the surface area results of the activated carbon produced in this investigation with predecessor work.

3.8. Surfactant adsorption

Table 9 displays the results for the adsorption of surfactants in the activated carbon produced with treatment 3 (ZnCl₂ as activating agent, a ratio of 4:1 and without drying prior to carbonization).

The produced activated carbon presents efficient adsorption of sodium dodecyl sulfate solutions. For the same conditions used in the adsorption of methylene blue, at both scenarios a final concentration of methylene blue active substances lower than 1 mg/L was obtained. In Costa Rica, the maximum discharge limit of methylene blue active substances in sanitary sewers and receptor bodies is 5 mg/L (^{MINAE, 2007}). Hence, the produced and evaluated activated carbon provided remarkably interesting results, within legislation limits.

Activated carbon proved to be effective in the adsorption of sodium dodecyl sulfate solutions. For both scenarios, a removal efficiency higher than 98 % was achieved. Similar results were obtained in the adsorption of gray water: a final concentration of 1.88 mg/L of methylene blue active substances was achieved, a value that is also below the allowed limit (5 mg/L) (^{MINAE, 2007}). A removal percentage of 98.6 % was obtained, thus indicating effective adsorption of the anionic surfactants present in the residual gray water.

3.9. SEM analysis

SEM analysis is shown in Figure 9. In Figure 9(a) it can be seen the different pores that compose the carbonaceous structure of the activated carbon produced using ZnCl₂ as activating agent (treatment 3). The structure is similar to that obtained in the work of ^{Mahamad et al. (2015)}, which also used pineapple stubble and ZnCl₂. Porous structures can be attributed to the dehydration effect of ZnCl₂, the oxidation of organic compounds in the carbonization stage, and the fact that zinc chloride acts as a template for porosity formation. The SEM analysis confirms the results found in the BET analysis of surface area and porosity, from which it was concluded that the activated carbon produced with ZnCl₂ contains micro and mesopore structures.

Figure 9 SEM of activated carbons: a) AC treatment 3 (ZnCl₂, 4:1 ratio), b) AC treatment 4 (H₃PO₄, 4:1 ratio), c) AC treatment 3 after the adsorption cycles of methylene blue, d) AC treatment 3 after the adsorption of sodium dodecyl sulfate 

In Figure 9(b) (treatment 4, H₃PO₄ as activating agent and a 4:1 ratio), a clear difference in morphology can be noted. This activated carbon presents a more irregular porous structure and less porosity when compared to activated carbon produced by treatment 3 (Figure 9(a)). It can also be seen that porous and cavities are not as defined or ordered. This corroborates results previously obtained in the BET analysis, in which it was indicated that the surface area and the total pore volume were lower for this activated carbon.

Activated carbon shown in Figure 9(c) corresponds to the same treatment as that shown in (a), but after the adsorption process of methylene blue. There is a clear difference in the surface observed for this activated carbon in comparison with the observed for the same activated carbon without adsorption. A uniform layer that covers the carbonaceous structure after adsorption can now be seen. Also, pores are not as clearly distinguished as in Figure 9(a). These results are similar to those reported by ^{Mahamad et al. (2015)} for the SEM analysis of activated carbon after adsorption of methylene blue. And suggest that methylene blue sticks to the surface of the activated carbon, thus saturating the pores.

The morphology of the activated carbon plays an important role in its adsorption capacity since characteristics such as the surface area or the pore size present determine whether the activated carbon is efficient or not for the adsorption of a certain pollutant. The greater the surface area present in the activated carbon, the greater the number of active sites available for adsorption. In the present work, it was demonstrated that the activated carbon produced with a higher surface area (treatment 3, ZnCl₂, ratio 4:1 and without prior drying) of 685.6 m²/g, which also has a greater total pore volume had better adsorption than activated carbon produced by treatment 4 (H₃PO₄).

Another important factor to consider in the morphology of activated carbon that influences the adsorption of activated carbon is the size of the pore present since the presence of microporosity favors the adsorption of adsorbates such as methylene blue. This occurs because micropores have molecular dimensions (^{Sann & Adeeyo, 2016}), therefore a small adsorbate molecule such as methylene blue will easily penetrate a pore that has a particular size. According to the physisorption analysis carried out, it was determined that the activated carbon produced by treatment 3 has the presence of micropores, while in the activated carbon produced by treatment 4 the presence of mesopores predominates, therefore it is not surprising that activated carbon produced by treatment 3 showed better adsorption of methylene blue.

SEM analysis was also carried out for the activated carbon used in the adsorption of the surfactant sodium dodecyl sulfate. Results are shown in Figure 9(d). On the surface of the carbonaceous structure, some particles can be observed, which could be remnants of zinc chloride or sodium dodecyl sulfate. However, a uniform layer is not observed on the surface of this activated carbon, as it was in Figure 9(c). This could be attributed to the fact that 4 reuse cycles of activated carbon were performed during the adsorption of methylene blue, while activated carbon shown in Figure 9(d) was used for surfactant adsorption only once.