To prepare the solutions, Merck brand Nickel Sulfate (NiSO4) and Potassium Dichromate (HCrO7) were used. To adjust the pH, solutions of 1M Sodium Hydroxide (NaOH) and Hydrochloric Acid (HCl) were used. The FTIR in a NICOLET 6700. The solutions were prepared using deionized water.

An experimental design known as central star composite response surface was used with the objective of not making replicas (Table I), in which the effect of three factors was analyzed: temperature, particle size and bed height; which were worked with five levels of variation each.

Tabla I .
Ranges And Levels Of Variation.

The plantain peel, collected in sufficient quantity to perform all the experiments, was washed, dried by sun for three days and at 90 °C in an oven for 24 hours [4]. It was then reduced in size and sized in a shaker-type sieve, selecting mesh sizes of 0.1351, 0.355, 0.6775, 1 and 1.2198 mm [5]. The prepared adsorbent was characterized by FTIR analysis to identify the groups involved in the adsorption process [4].

Adsorption tests of Cr (VI) and Ni (II) were performed at pH 2 and 6, respectively, and 100 ppm, following the proposed experimental design; a continuous system equipment with acrylic columns of 4.1 cm internal diameter and height of 15 cm was used, with a flow rate between 11-15 mL/min [5]. The variation and control of the temperature was made by means of an Arduino nano controller. Samples to measure the final ion concentration in the solution were taken at the bottom outlet of the column. The final concentration of Ni (II) was determined by atomic absorption and the Cr (VI) concentration by UV-VIS spectrophotometry using the 1.5-diphenylcarbazide colorimetric method at 540 nm in a UV-Vis spectrophotometer model BK-UV 1900.

The analysis of the results was carried out in the STATGRAPHICS Centurion 18.1.02 software, in order to optimize the process using the Response Surface Method (RSM) and to determine the incidence of the variables studied on the adsorption of Chromium (VI) and Nickel (II), using analysis of variance, Pareto diagram for each of the biomasses under study.

The breakage curve was made by placing Cr (VI) and Ni (II) contaminated solution in contact with the bioadsorbent, at the optimal conditions of temperature, particle diameter and bed height found by RSM for each contaminant. Adams-Bohart, Thomas, Yoon-Nelson and dose-response models were used to predict the column behavior, calculating the process constants and evaluating the adsorption capacity of the adsorbent.

Adams-Bohart model

The Adams-Bohart model is used only for the initial description of the breakage curve above the breakage point or saturation points; it only describes between 10-50% of the behavior. It assumes that the adsorption rate is proportional to the adsorption capacity and the concentration of the adsorbed species [6].

This model is described according to (1):

(1)

Where KAB is the kinetic constant of AB in (L/mg.min), NO is the maximum volumetric adsorption capacity in mg/L and v is the linear flow rate in (cm/min), L is the column bed depth in cm, C0 is the internal concentration mg/L, Ct is the effluent concentration and t is the residence time of the solution in the column [7].

Thomas model

Thomas' model is of a dynamic type and follows Langmuir's adsorption kinetics. In this model the axial dispersion is considered to be negligible in the bed, due to the fact that the driving force obeys a second-order kinetics with a reversible reaction [8].

Given by (2):

(2)

Where Kth, is the Thomas constant (mL/min-1 mg-1); qo, is the maximum solute concentration in the solid phase (mg/g); X, is the amount of adsorbent in the column (g), Q is the flow rate (mL/min) and V (L) is the volume of the effluent at the time of operation [9].

Dose-Response Model

The Dose-Response model has been commonly used in pharmacology to describe different types of processes, and is currently being applied to describe columnar bioadsorption [24]. The DR model can be represented by (3):

(3)

Where a is the model constant, q0, maximum solute concentration in the solid phase (mg/g), X, is the amount of adsorbent in the column (g) and Q is the flow rate (mL/min) [25].

The amount of metal removed for PP-Cr(VI) was resolved by calculating the area under the breakage curve using the following integral shown in (4) and (5):

(4)

And for PP-Ni(II) was used:

(5)

The numerical integration was done by the Simpson 1/3 method and the different integrals are done for the reason that the separation of all the points are not equidistant so it is done by intervals which certainly are. The total amount that passes through the column, the percentage retained, the biosorption capacity and the metal in solution at the end are calculated by (6) to (9):

(6)

(7)

(9)

The adsorption process is a complex multi-step phenomenon that involves many factors; among these, the surface chemistry and pore structure of the biomasses studied are influential. In Fig. 1a, it is shown the spectrogram obtained and the chemical composition in % of atomic weight for unsaturated PP, and after the removal of Chromium (VI) (Fig. 1b) and Nickel (II) (Fig. 1c). It was obtained that carbon and oxygen are the most present in all the materials studied, with 75.18 and 23.34%, respectively, which can be attributed to their organic nature; as well as elements in small proportions: Al, Cl and K [2], [26]. De igual forma, después del proceso de remoción, se evidencia la presencia de los iones metálicos removidos.

Fig. 2a shows the scanning electron microscopy (SEM) images of the PP, which has a cylindrical agglomerated appearance.

Fig. 1.
EDS analysis of unsaturated plantain peel (a) and after the adsorption process of Chromium (b) and Nickel (c).

From SEM micrographs after the adsorption experiments it was found that Cr (VI) and Ni (II) ions were adsorbed on the plantain peel biomass surface according to an electrostatic attraction mechanism, suggesting that the process is dominated by chemisorption [26], [27].

Fig. 2
SEM microphotographs of unsaturated plantain peel (a) and after the adsorption process of Chromium (b) and Nickel (c).

FTIR analysis of biomass provides information regarding the organic functional groups present inside the structure of the lignocellulosic material, which have been reported as participants in ion exchange to capture the contaminant within the pores of the biosorbent [28]–[30]. Fig. 3 shows the IR spectrum of the PP before and after the Cr (VI) and Ni (II) removal process, and shows the representative peaks corresponding to the vibration and stretching of the OH, C-H, C=O, COOH groups [31], [32].

Fig. 3.
FTIR plantain peel (PP) before and after the adsorption process of Cr (VI) and Ni (II).

From the spectrum of Fig. 3, a broadband at 3387 cm-1 is identified due to the characteristic vibration of the O-H hydroxyl present in the hydrocarbon chains of cellulose, hemicellulose and lignin present in the material [33]–[35]. The fluctuation in 2931 cm-1 is due to the C-H bond or aliphatic group resulting from the organic compound structure in the material structure attributed to the possible presence of methyl, methylene or methoxy groups [36], [37]. The peak at 1652 cm-1 is an indication of carbonyl groups, possibly a result of the carboxylic groups of pectin, hemicellulose and lignin [38]. The frequency observed between 1010-1419 cm-1 can be attributed to the vibration of carboxylic acids and alcohols [30].

After the adsorption process a noticeable change in the spectrum of both biomaterial samples is observed, showing stretching of the bands at certain frequencies compared to the unused biomaterial. The vibrations of the hydroxyl, carbonyl, alcoholic, aliphatic and carboxyl groups are much more pronounced; these changes can be attributed to the metal presence, both Ni (II) and Cr (VI), in the structure of the biomaterial which is the result of the union of the metals to the different functional groups [33].

Table II presents the results obtained in each adsorption test, according to the experimental design previously raised. From the results, it is observed that the PP showed a good performance in the removal of Cr (VI), reaching percentages between 90 % - 98.7 %; good results were also obtained in the removal of Ni (II), reaching percentages of removal higher than 70 % below all the conditions evaluated. This is due to the presence of carbonate compounds in the structure of the biomaterial, which suggests high presence of lignin and cellulose. The high removal efficiencies are also attributed to the presence of H- ions at the pH worked, which could contribute to the selectivity of the active centers by the metal [31].

Tabla II.
Removal Percentages Of Cr (VI) And Ni (II) Using PP.

From Fig. 4a it can be said that none of the variables and their possible combinations in the experimental range established in the design of experiments reach 95% reliability, so they do not have a significant effect on the Cr (VI) removal, presenting the combination of temperature-particle size a negative effect on the system as well as the temperature and height of the bed, as evidenced in the effects graph (Fig. 5a). The Pareto diagram for the removal of Ni (II) (Fig. 4b), shows that the variable that has the greatest influence within the system is the particle size, which determines the internal surface area and the number of sites available in the bioadsorbent [11], with temperature being the variable that has the least influence on the process.

Fig. 4.
Standardized Pareto diagram for the removal process of (a) Cr(VI) and (b) Ni(II).

Fig. 5 relates the variables together with the removal percentage. This graph allows us to understand the behavior and effect that the controllable variables have on the system.

Fig. 5.
Effect of temperature, bed height and particle size on the % removal of (a) Cr (VI) and (b) Ni (II).

From Fig. 5, is established that increasing the solution temperature of the biomaterial throughout the evaluated range reduces the removal of Cr (VI), while for Ni (II) the decrease occurs from 40 to 50, which would imply a process of exothermic nature, which releases energy when the bonds between the contaminants with the functional groups are formed, and would indicate that the increase of energy in the system would cause a weakening of the interaction forces between the adsorbate and the adsorbent [39]–[41]. A temperature increase favors the desorption of the sorbate molecules over the biosorbent structure, because by gaining more energy than the bonds formed between the compounds support, the interaction forces are reduced, causing the release of what is not desired and this is a result of the temperature [42].

The increase of the bed height favors the adsorption due to the increase in the amount of active sites that allow the union of the ions with the functional groups, this effect is evident in the adsorption of Cr(IV) with bed heights between 30 and 56 mm; however for the heights of 56 to 100 mm when performing the experiment at high heights with low particle sizes the solution tended to seek the least tortuous path to descend along the bed, thus leaving areas without contact with the contaminated solution [43]. In contrast, the increase in the height of the bed favors proportionally the efficiency of NI (II) removal due to the fact that there is a greater biomass surface area because of the greater amount of particles; likewise, the diffusion phenomenon predominates over the axial dispersion phenomena in the mass transfer, causing an expansion of the mass transfer zone [44]. The good performance of Ni (II) removal can also be attributed to the availability of active contact sites between the metal and the particles, expanding the mass transfer zone by the higher mass transfer resistance and slower adsorption kinetics; in turn the increase in mass transfer resistance is caused by the repulsive forces between the adsorbed metals and the metals present in the solution bulk [45].

The optimal operating conditions for the adsorption system were obtained using the Response Surface Methodology (RSM), and are shown in Table III:

Table III.
Selected Optimal Values For Ni(II) And Cr (VI) Adsorption

The predictions provided by the RSM for Cr (VI) are shown in the function graph as an estimated area (Fig. 6).

Fig. 6.
Estimated response area for Cr (VI).

The temperature selected will be the ambient temperature because the system's behavior is more efficient when it is not supplied with energy.

The predictions for Ni (II) are shown in Fig. 7:

Fig. 7.
Estimated response area for Ni (II).

The breakage curve of a bed makes it possible to establish its service life and saturation time [46]. Thus, the optimal bed height, particle size and temperature values, previously determined by the RSM, were used to achieve maximum removal in the adsorption tests. Fig. 8 shows the time vs C/Ci graph for Cr (VI) adsorption and Fig. 9 for Ni (II) adsorption, adjusted to Thomas, Response Dose, Yoon-Nelson, Adams-Bohart and BDST models [47].

The Cr (VI) breakage curve was made for approximately 17 hours, identifying the breakage time around of 300 min, where the output concentration was 7.3 ppm and therefore the removal percentage was 92.7%. The saturation point was determined at 1000 min with an output concentration of 69.4 ppm, which corresponds to 30% removal; and at 1020 min the system is considered to be in equilibrium, because the output concentration is 70 ppm. The behavior observed in the breaking curve, it can be said that the very low slope of the curve is due to the low concentration difference between the absorbent solution and the solute of the dissolution, creating a very low concentration gradient, resulting in a very weak driving force responsible for the transport of the contaminant from the fluid to the particle structure, which decreases the diffusion coefficient, causing the saturation time of the biomaterial to be much longer [48], [49].

Fig. 8.
Adjustment of Cr (VI) adsorption breakage curve to models.

The values of the system performance during the Cr (VI) and Ni (II) adsorption process are presented in Table IV, determining the amount of metal removed by calculating the area under the breakage curve using the x-equations, described in the methodology.

Tabla IV.
System performance.

Regarding the behavior of the Ni (II) breakage Curve, it can be seen that after 720 minutes of operation, it still presents a high removal capacity without reaching the breakage point. This could be due to the low volume of liquid that circulated through the column, which is represented by the value associated with the amount of metal fed into the column, which means that there would not be enough contaminant in the bed to come into contact with all the active sites of the biomass, causing the adsorption capacity of the biosorbent to be low in comparison with Cr (VI) [50], [51].

The adjustment parameters of the modeling of the breakage curves are presented in Table V.

Tabla V.
Setting parameters of Cr (VI) and Ni (II) breakage curve models.

Observing the previous table, the models of best adjustment to the experimental data of Cr (IV) adsorption are Yoon-Nelson and BDST with an R2 of 0.9843; nevertheless, the other models evaluated, also presented high R2 values which indicate us that they are not so far from predicting the behavior of the breakage curve. However, when observing Fig. 8, the Thomas model moves considerably away from the experimental values at the beginning of the test, so this model is discarded from the analysis. With regard to Ni (II), the models evaluated did not adequately fit the experimental data due to the behavior that occurred in the bed. This behavior suggests the need to develop our own equations to describe the process and perform a correct industrial scaling

Fig. 9.
Adjustment of Ni (II) adsorption breakage curve to models.

Thermodynamic parameters reflected the feasibility and spontaneous nature of the process; thus, free energy change, enthalpy change and surface entropy change were estimated using the change of the equilibrium constant with absolute temperature [28]. For this, the Van't hoff graphical method was used, which expresses the change in free energy as a function of the change in enthalpy and adsorption entropy with temperature, obtaining the regression line -ΔS° and intercepting ΔH° as expressed in the equation [39]. The thermodynamic parameters for Cr and NI(II) adsorption are shown in Table VI:

Tabla VI.
Thermodynamic parameters for adsorption of Cr(VI) and Ni(II).

It is observed that Gibbs' energy in the Cr (VI) removal process increased proportionally with temperature, which corroborates the non-spontaneous increase of the removal. The negative direction of Entalpy in the removal of Cr(VI) and Ni(II) indicates the exothermic nature of the process, which is consistent with the results found above. On the other hand, the negative value of the Entropy of adsorption of both metals indicates a low reversibility in the process and a low randomness in the liquid-solid interface in the adsorption process [52], [53]. The decrease in the free energy of Gibss during Ni(II) removal indicates that the spontaneity of the process is inversely proportional to the temperature increase, so the system does not need energy additions to react [54], [55].