The experiments with the synthetic effluent were carried out in a laboratory-scale reactor, operating in a batch system, consisting of a beaker with 250 mL of capacity and a magnetic stirrer to homogenize the solution, positioned inside a wooden box (80 cm x 40 cm x 60 cm) coated with aluminum foil to increase the radiation incidence in the solution and equipped with a UV source (high pressure Mercury vapor lamp without the bulb – 250W) fixed at the top about 20 cm from the solution. The inner temperature, after 20 min, was maintained at about 45 °C (Figure 1).

Figure 1.
Photo-Fenton process reactor.

Caption:

1 – Magnetic stirrer.

2 – Reactor.

3 – UV source.

4 – Wooden box coated with aluminum foil.

Samples were withdrawn at pre-established intervals, 0-20 min for red 4B dye, and 0-15 min for blue 5G dye, and immediately the remaining concentration in the solution was determined using the dual beam scanning UV/VIS molecular absorption spectrophotometer by the calibration curve.

In order to verify the influence of the H₂O₂ concentration, Fe²⁺ ions concentration and pH and to obtain an optimum discoloration condition, a central composite rotational design (CCRD) 2³ was elaborated with 2 replicates at the central point, using Statistica 8.0 software and the desirability function as tool. As the response variable in the statistical analysis the percentage of discoloration of the analytes (Red 4B and Blue 5G dyes) was used. The experimental data were adjusted to the linear and quadratic models it was evaluated using the Analysis of Variance (ANOVA) at the 95 % confidence level.

In the study of kinetics of dye discoloration, solutions were used with concentration of 100 mg L^-1 for red 4B and 50 mg L^-1 for blue 5G in 250 mL beakers. Aliquots were taken in the time intervals of 0; 1; 2; 4; 8; 10; 12; 15 and 20 min to the red 4B dye, and 0; 1; 2; 4; 8; 10; 12 and 15 min to the blue 5G dye, with determination based on the previously determined maximum wavelength.

The discoloration was evaluated using the photo-Fenton process (UV/H₂O₂/Fe⁺²) by kinetic models. The order of the reaction is the dependence of the speed of the reaction with the concentration. With C₀ being the initial concentration of the reagent, and C the concentration of the reagent elapsed at time t of reaction and n the order of the reaction. When n = 1 the reaction is of pseudo-first order (Equation 1) and n = 2 of pseudo-second order (Equation 2).

(1)

(2)

The experimental data obtained in the discoloration assays were fitted to these models in order to evaluate their decay during the experiment time, as well as to determine the half-life of each analyte according to the equation of pseudo-first order (Equation 3) and of pseudo-second order (Equation 4).

(3)

(4)

In order to determine the amount of dye removed, the optimum wavelength definition for the red 4B and blue 5G dyes reagent was first determined. This definition was obtained by double beam spectrophotometer scanning, where the molecular absorption spectrum was obtained, indicating the maximum absorbance band at 536 nm for red 4B dye and at 591 nm for blue 5G dye (Figure 2).

Figura 2
a) UV absorption spectrum in aqueous solution at 50 mg L^-1 concentration for red 4B dye and b) blue 5G dye.

The experimental matrix for the 2³ central composite rotational design (CCRD) delineation is presented in Table 1, with the levels of each factor, the actual values and the response variable obtained for the discoloration of red 4B and blue 5G dye in the photo-Fenton process (UV/H₂O₂/Fe⁺²) in the course of 16 runs performed randomly.

Table 1.
CCDR 2³ planning matrix with the factors coded (and real) and responses regarding the discoloration efficiency of 4B red and 5G blue dyes.

More effective discolorations were observed in assays 5, 7, 10, 14, 15 and 16 for both dyes, where H₂O₂ concentrations were applied at higher levels and pH was more acidic.

With the experimental results obtained in the discoloration of the dyes by photo-Fenton from CCDR 2³, the values of the estimated effects of each parameter (concentration of H₂O₂ and Fe²⁺ ions, and pH) on the response variables presented in Table 1 were obtained, values that presented p-value less than 0.05 were considered significant for the 95 % confidence interval. From the significant values, the mathematical equation of the quadratic regression model and its respective determination coefficients (R²) were obtained (Table 2).

Table 2.
Mathematical models and determination coefficients (R²) of the models adjusted for discoloration of the red 4B and blue 5G dyes.

For the discoloration of the red 4B and blue 5G dyes respectively the R² values show that 88.56 % and 87.42 % of the responses were explained by the models, and the linear effects of pH and concentrations of [H₂O₂] (mg L^-1) and [Fe²⁺] (mg L^-1) were the most important to describe the behavior of discoloration of the analytes by the photo-Fenton process.

To verify the fit quality of the model, the analysis of variance (ANOVA) was used, evaluating the determination coefficients (R²) and the F test for both discolorations (Table 3).

Table 3.
ANOVA of the quadratic model to remove red 4B and blue 5G dyes.

It is observed in Table 3 that the ratio of F_cal to F_tab for the regression presented a statistically significant value for the equation of discoloration of both dyes, a fact evidenced by a value higher than 1. Moreover, for the blue 5G dye, it was not observed (Figure 2). In this study, the F_cal/F_tab ratio for the lack of fit was lower than 1⁵.

On the other hand, for the discoloration of red 4B dye the regression model generated was significant (p ≤ 0.05) because F_cal = 5.10 was higher than the F_tab = 4.10. However, the lack of fit was also significant (F_cal = 786.83 > F_tab = 230), although the ideal was a value of F_cal < F_tab, and therefore not significant. However, since the means at the central points were very close and the pure error very low (reasons for the F of the lack of high fit), the model was considered valid for predictive purposes⁵.

The Figure 3 shows the behavior of the process regarding the discoloration efficiency of red 4B according to the relation of the dependent variables by the surface response graph.

Figure 3.
Response surface for discoloration efficiency red 4B dye by a) Fe²⁺ (mg L^-1) and H₂O₂ (mg L^-1); b) pH and H₂O₂ (mg L^-1); c) Fe²⁺ (mg L^-1) and pH.

In Figure 3, it can be observed that the discoloration efficiency of red 4B dye presented mean values between 65.04 % and 100.00 %. The best discoloration efficiency values, 98.75 % and 100.00 %, were obtained with pH 3, H₂O₂ concentration of 75 mg L^-1, Fe⁺² concentration of 5 and 15 mg L^-1, respectively.

The Figure 4 shows the behavior of the process regarding the discoloration efficiency of the blue 5G dye according to the relation of the dependent variables by the surface response graph.

Figure 4.
Response surface for blue 5G dye discoloration efficiency by a) Fe²⁺ ions (mg L^-1) and H₂O₂ (mg L^-1); b) pH and H₂O₂ (mg L^-1); c) Fe²⁺ ions (mg L^-1) and pH.

In Figure 4 it is observed that the discoloration efficiency of the blue 5G dye presented average values between 16.24 % and 100.00 %. The best discoloration efficiency values, both at 100.00 %, were obtained with pH 3 and 1.6, H₂O₂ concentration of 75 mg L^-1 and 50 mg L^-1, and Fe⁺² concentration of 5 and 10 mg L^-1.

Analyzing Figures 3 and 4 it is possible to verify that the concentration of H₂O₂ and Fe²⁺ ions showed a positive effect on the discoloration efficiency of red 4B and blue 5G dyes and that the pH variable had a negative effect on the discoloration efficiency, indicating that the highest discolorations were achieved with the decrease of this variable.

Škodič et al. demonstrated that the additions of moderate concentrations of H₂O₂ and Fe⁺² ions catalyst during the AOPs obviously increased the decolorization efficiencies within the first few minutes of the processing time (5–10 min) for Reactive Blue 4 and Reactive Blue 268 dyes⁴.

As a more favorable response to the discoloration of dyes in aqueous solution, the optimization tool employing the desirability function indicated a H₂O₂ concentration of 66.80 mg L^-1, Fe²⁺ ions of 9.66 mg L^-1 and pH 5.81 with a percentage of 100.01 % discoloration to red 4B dye. For the blue 5G dye, a H₂O₂ concentration of 55.04 mg L^-1, Fe²⁺ ions of 10.34 mg L^-1 and pH 2.59 with a predicted percent discoloration of 100.56 %.

The optimum conditions were used in the kinetic study for both dyes. The pseudo-first order and pseudo-second order models fitted to the experimental data are shown in Figure 5.

Figure 5.
Adjustment of the experimental data of dye discoloration to the kinetic models of Pseudo-first order and Pseudo-second order. a) Red 4B; b) Blue 5G.

By means of this adjustment, it was possible to obtain the kinetic constant of pseudo-first order (k₁) and pseudo-second order (k₂), correlation coefficient (R²) and half-life time (t_1/2), indicated in Table 4, for the discoloration of the red 4B and blue 5G dyes.

Santana et al. also obtained a good kinetic adjustment for the pseudo-first order for the removal of reactive blue dyes 5G and remazol red RB 133 % by photo-Fenton process6.

Table 4.
Kinetic data for the photo-Fenton process (UV/H₂O₂/Fe₊₂) of the red 4B and blue 5G dyes.

Half-life times of 1.16 min for red 4B and 4.60 min for blue 5G dyes were observed. It was also verified that there was a greater speed of dye discoloration in the pseudo-first order model, besides a better adjustment in both treatments.

According to Núñes et al., taking into account that the concentration of the reactive species needs to reach a steady state during the process, and once the oxidant concentration can be considered constant, the kinetics of the discoloration process can be treated as pseudo-first order, in terms of the consumption of the organic compound, in this case, the dye⁷. Generally, advanced discoloration processes obey the pseudo-first order kinetics8.