Reactive Black 5 (RB5, 85% w/w from DyStar, Germany), was donated by Fabricato–Tejicondor S.A., from the city of Medellín, Colombia. Titanium dioxide (TiO₂, 91.5% w/w), NaOH, 100% w/w and HCl, 37% v/v were purchased from Merck, USA. Dulbecco's modified Eagle's medium (DMEM, reference D5030, reagent grade), dimethyl sulfoxide (DMSO, reactive grade), 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) (MTT, reactive grade) and doxorubicin (98% w/v), were obtained from Sigma-Aldrich, USA. Fetal bovine serum (FBS, reagent grade); Trypsin- EDTA (ethylenediaminetetraacetic acid, reagent grade, catalogue R001100), were provided from Invitrogen, USA.

HepG2 cell line (ATCC HB-8065), was purchased from the ATCC (American Type Cell Collection, USA) cell bank.

A Box–Behnken experimental design was selected in the Statgraphics Centurion XVI^® statistical program to study the effect of the combination of three factors on dye decolorization: TiO₂ concentration (0.1, 0.175, and 0.25 g L^-1), pH (3, 7, and 11), and RB5 concentration (50, 75, and 100 mg L^-1).

Fifteen tests were performed, with three central points (0.175 g L^-1 of TiO₂, 75 mg L^-1 of RB5, and pH 7), and five degrees of freedom for the experimental error (Table 1). Also, two control tests were performed, using two center point conditions: 1) without UV light and with TiO₂, and 2), with UV light and TiO₂ free. When fitting the model, residuals plot was checked, in searching of potential outliers.

Table 1. Box–Behnken experimental design.

Test	TiO2 (g L-1)	pH	RB5 (mg L-1)
1	0.100	3	75
2	0.175	11	50
3	0.175	7	75
4	0.175	7	75
5	0.175	7	75
6	0.175	11	100
7	0.250	7	100
8	0.100	7	100
9	0.250	11	75
10	0.175	3	50
11	0.175	3	100
12	0.100	11	75
13	0.250	3	75
14	0.100	7	50
15	0.250	7	50

The RB5 photocatalytic decolorization was performed in an aluminum photoreactor (R.A. Talleres), equipped with five lamps (Philips 30W, TL-D Actinic BL), with l = 365 nm. According to the experimental design (Table 1, Figure 1), 250 mL beakers were used, with 100 mL of the dye solution and TiO₂. The reaction mixtures were stirred at 100 rpm for 30 min; then, maintaining the stirring at 30 °C, the lamps were turned on for 14 h. The reaction was monitored every 2 h. The RB5 decolorization percentage was measured as the response variable. For the analysis, the reaction mixture was centrifuged at 3200 rpm for 15 min (Heraeus, D-37520 Osterode), and the supernatant was filtered through nitrocellulose filters (φ of 0.45 μm).

Degradation kinetics of the dye was determined by plotting the decolorization percentage of RB5 against the time of the photocatalytic reaction. As a part of the analysis of the results, the calculation of the area under the curve (AUC) was also taken into account, as it is a useful methodology when the progress of a process over time is tested. The total area was obtained by summing the area of the trapezoids under the curve, which is formed between pairs of decolorization percentage readings over time ( 19 ).

The RB5 decolorization was followed spectrophotometrically (Shimadzu UV-1800 spectrophotometer) at 598 nm. It was expressed in terms of decolorization percentage (D%) (Equation [1]) ( 20 ):

Where A₀ corresponds to the initial absorbance and A_t to the absorbance after the photocatalytic reaction.

Because the concentration of TiO₂ is a parameter that significantly affects the decolorization rate of the dye, it is essential to optimize it. TiO₂ particles are likely to aggregate at high concentrations, which reduces the transmission of UV light, and hence the speed of reaction ( 8 ).

To optimize the photocatalytic decolorization of RB5, the results obtained after applying the Box–Behnken experimental design were analyzed using the Statgraphics Centurion XVI^® statistical program. It was determined that the decolorization process could be improved using 0.25 g L^-1 of TiO₂, a pH of 3, and 50 mg L^-1 of RB5. However, to determine the influence of the TiO₂ concentration on the dye decolorization, two concentration values of TiO₂ were applied: 0.25 g L^-1 and 0.5 g L^-1, keeping a fixed pH of 3, and a concentration of 50 mg L^-1 of dye. Thereby, six additional trials (independent to the experimental design) were proposed, being stricter concerning to TiO₂ concentration (see tests 16–21, in Table 2).

The MTT reduction method was employed to establish the cytotoxicity of the reaction mixtures corresponding to the tests with the highest percentage of RB5 decolorization, utilizing the HepG2 cell line ( 21 ). The cells were seeded in 96-well culture dishes, using DMEM medium with 10% FBS. After 24 h, the mixtures from the photocatalytic reaction (100% w/v), were added to the culture dishes at serial dilutions of 50, 25, 12.5, 6.25, 3.13, and 1.56% w/v. The samples were incubated for 72 h at 37 ° C and 5% CO₂. After this period, the MTT was added to the culture dishes, and after additional 3 h of incubation at the same temperature, DMSO was added. Finally, the absorbance at 570 nm was determined and the minimum lethal concentration (LC₅₀) was found. The cells were observed trough an inverted microscope (Nikon ECLIPSE TS 100). The assays were run twice, in independent experiments, with two replicates per dilution. Doxorubicin was used as positive cytotoxicity control, as it is a cytotoxic agent intercalating the DNA; untreated cells were used as negative cytotoxicity control ( 22 ).

The effect of TiO₂, pH and RB5 on decolorization percentage was evaluated, using a Box-Behnken design. Under the conditions of the 10^th (0.175 g L^-1 TiO₂, 50 mg L^-1 RB5, and pH 3), 13^th (0.25 g L^-1 TiO₂, 75 mg L^-1 of RB5, and pH 3), and 15^th (0.25 g L^-1 of TiO₂, 50 mg L^-1 of RB5, and pH 7) tests, 100%, 99.40%, and 99.6% of the RB5 were removed after 14 h, respectively (Figure 2. For best interpretation, see Table 3).

Table 3. Decolorization percentage of RB5 for 14 h. Results obtained after applying the Box–Behnken experimental design.

time (h)	Test
	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15
0	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00
2	29.61	63.37	46.11	49.72	47.04	28.50	42.23	11.01	46.41	65.93	48.39	46.78	56.76	56.83	60.31
4	38.97	67.15	52.22	50.00	54.93	28.50	45.34	17.30	46.71	67.26	51.83	50.58	68.47	64.94	77.82
6	52.87	81.10	63.06	63.35	68.17	45.80	51.04	22.33	56.59	82.74	57.57	52.05	75.08	69.37	88.33
8	52.87	81.40	67.50	67.90	76.06	50.13	65.80	27.36	64.97	91.59	72.25	57.60	84.08	78.60	95.72
10	58.91	83.43	73.61	71.88	75.21	55.47	72.54	24.21	70.06	94.69	73.85	59.36	87.99	84.13	98.44
12	68.28	91.86	82.78	82.39	87.32	63.36	68.13	25.47	74.25	96.46	74.54	62.87	90.39	86.72	99.61
14	73.72	93.31	88.33	84.38	87.61	66.67	82.38	35.22	83.23	100.00	87.16	66.96	99.40	94.10	99.61

Particularly, for the 15^th test, the degradation was almost complete (98.44%) after 10 h; hence, the process could be completed in this period with a representative removal percentage. Also, it was observed that under such conditions (15^th test), the highest dye removal rate was achieved following first-order kinetics (R² = 0.9819, ln C vs t, Figure 3), with a rate constant of k = 0.4223 h^-1 (Equation [2]).

Where Co corresponds to the RB5 initial concentration and C relates to the RB5 concentration at time t.

The results obtained with the experimental data are consistent with those previously reported for different azoic dyes, such as Remazol Red RR, Remazol Yellow RR, Procion Crimson H-exl, and Procion Yellow H-ex, for which removal percentages ˃ 90% were achieved ( 4 ), ( 23 ). Additionally, other authors have reported that the degradation of azoic dyes and other substances by heterogeneous photocatalysis, using TiO₂ and UV, light can be followed by a first-order kinetics process ( 19 , 24 ). Furthermore, it has been reported previously that the degradation of reactive orange 16 (an azo dye like RB5), by heterogeneous photocatalysis with UV/TiO₂ follows first-order kinetics ( 25 ).

Since the kinetics were evaluated only by the RB5 concentration, it is necessary to take into account that the evaluation of other involved species may be necessary, if the overall kinetic parameters are required ( 19 ). These species (e.g., intermediates) might include •OH, H₂O₂, and O₂ ^•−, as indicated by the following oxidative-reductive reactions (Eqs. [3]-[9]) ( 12 , 26 ):

In general, RB5 degradation was favored with low contaminant loads (50 mg L^-1, see tests 2, 10, 14, and 15). However, removal percentages higher than 67% were achieved for an RB5 concentration of 100 mg L^-1 (trials 6, 7, 9, and 11, Figure 2). These results were excellent compared to those reported by Puentes-Cárdenas et al. ( 27 ), who achieved removal of 17% of dye when treated 80 mg L^-1 of RB5 by photocatalysis with TiO₂ particles, deposited on borosilicate. However, they also indicated that at low RB5 concentrations (10, 50, and 70 mg L^-1), the decolorization was ˃ 80% ( 27 ). Furthermore, it was observed that the degradation of 50 mg L^-1 of RB5 using 0.25 g L^-1 TiO₂ and UV light was effective for the dye degradation, indicating that under such conditions the surface of the catalyst was not saturated and therefore TiO₂ was active in the photocatalytic process.

Also, TiO₂ can be reutilized up to four times and still degrades the RB5 (˃ 90%), which demonstrates its stability. This characteristic of TiO₂ would allow treating several loads of RB5 using the same catalyst, and therefore a system of continuous degradation could be implemented.

Due to the fact that pH can affect the adsorption of dye on the catalyst surface, the point of zero charge (pzc) of the TiO₂ must be taken into account; the pzc corresponds to a pH value of 6.9 ( 12 ). In acid (pH < 6.9) or alkaline (pH ˃ 6.9) media, the surface of the TiO₂ can be protonated (positively charged) or deprotonated (negatively charged) (Eqs. [10] and [11], respectively). Besides, the RB5 in aqueous solution is negatively charged because it is a sulphonated dye ( 28 ).

In this study, photocatalytic degradation is favored at pH 3. Under this condition RB5 degradation is faster because TiO₂ particles are positively charged, so RB5 is adsorbed easier on TiO₂ as a result of the electrostatic attraction of the positively charged photocatalyst with the dye ( 29 ). This behavior agrees with what has been reported by other authors who claim that the degradation of azo dyes increases with decreasing pH ( 13 , 16 ). At low pH, reduction by electrons in CB may play an important role in the degradation of dyes due to the reductive cleavage of azo bonds ( 16 ).

Promising removal of the dye was also found at pH 7 (15^th test), which is good considering that actual polluted effluents (e.g., from textile industries) have a high pH. Although, as reported Threrujirapapong et al. ( 29 ), the diverse composition of synthetic dyes in the wastewater is an important reason for the different results.

A second-order response surface model for AUC on TiO₂, pH and RB5 was fitted, using data shown in Table 4.

Table 4. Area under the curve (AUC) of Concentration vs Time, for the RB5 decolorization after 14 h, by photocatalysis with TiO2/UV.

Test	AUC (D%*h)
1	647.13
2	966.57
3	812.78
4	805.11
5	858.03
6	581.68
7	730.31
8	279.56
9	754.79
10	1031.42
11	795.64
12	678.65
13	968.17
14	918.45
15	1079.77

All main effects and the interaction TiO₂*pH were statistically significant (p-value < 0.05), as shown in Table 5.

Table 5. Estimated parameters of the second-order response surface model for AUC on TiO2, pH and RB5, and their corresponding p-values.

Effect	Estimate	p-value
Intercept	825.307
TiO2	179.491	0.0032
pH	-115.168	0.0076
RB5	-329.434	0.0003
(TiO2)2	-82.1054	0.0951
TiO2*pH	-122.45	0.0200
TiO2*RB5	-0.9275	0.9849
(pH)2	-44.1379	0.3070
pH*RB5	-74.555	0.0846
(RB5)2	81.1796	0.0978

Using a numerical method implemented in Statgraphics, a maximum expected AUC of 1168.50 D%*h was found, which is attainable at TiO₂: 0.3 g L^-1, pH: 3.0, and RB5: 50 mg L^-1. This is consistent with the results of the 10th and the 15th test (AUC: 1031.42 D%*h and 1079.77 D%*h, respectively), whose conditions were close to those for which the maximum AUC is predicted. The Figure 4 shows the evaluated region for pH and TiO₂, with RB5 fixed at 50 mg L^-1.

For the two test controls (with TiO₂ and without UV light, and with UV light but TiO₂ free), there were no changes, as the dye oxidation is mainly mediated by the action of the TiO₂ semiconductor in the presence of UV light, which activates the catalyst. In addition, a low decolorization percentage (about 7%), was found in the absence of UV light. This result could be associated with the adsorption of the dye on the solid TiO₂ surface, as reported by Chong et al. ( 4 ).

According to the optimization tests, under the conditions of the 21^st test (0.5 g L^-1 of TiO₂, pH 3, and 50 mg L^-1 of dye), the highest decolorization percentage (99.51%) was reached in 10 h. The six trials (assays 16 to 21) resulted in decolorization percentages ˃ 94% at 10 h. From the above, it is possible to conclude that the catalyst load used in the optimization tests is not a determining factor in the improvement of the process productivity (Figures 5 and 6).

With the optimization tests, the time required to achieve a decolorization percentage over 94% was reduced to 10 h (Figure 6). This behavior was also observed with the 10^th and 15^th tests (Figure 2). Besides, it was confirmed that the RB5 decolorization by photocatalysis with TiO₂/UV was favored at an initial pH of 3. The decolorization percentages reached in the optimization tests did not differ significantly from each other. However, to reduce costs and to facilitate the catalyst separation, low concentrations of TiO₂ (0.25 g L^-1 – 0.35 g L^-1) should be used. Further, it was found that at low dye concentrations (50 mg L^-1), the degradation process is faster, similar to the findings reported by Puentes-Cárdenas et al. ( 27 ), who obtained decolorization percentages higher than 80% in 10 h, when using dye concentrations of 10 mg L^-1, 50 mg L^-1, and 75 mg L^-1.

Conversely, Chong et al. ( 4 ) achieved 97% decolorization with a RB5 concentration of 1 mg L^-1 after 150 min of treatment with a TiO₂ concentration of 0.1 g L^-1. In contrast, in the present work, under the conditions of the 16th test (0.25 g L^-1 of TiO₂, 50 mg L^-1 of dye), a decolorization percentage of 94.6% was reached at 600 min (Figure 6). This result is due to the higher proportions of dye and catalyst that were used in this study, compared to those used by Chong et al. ( 4 ). Besides, there have been previously reported lower times compared to those obtained for the RB5 degradation. Though, as mentioned above it is important to note that the concentrations of dye used in the present study (50 mg L^-1) are higher than those reported by other authors; therefore, it would take more time to remove dyes (14 h). However, longer treatment times than those used in this work have also been reported in the literature, for instance, Threrujirapapong et al. ( 29 ) reported a retention time of 2 days in a photocatalytic reactor for sufficient wastewater treatment.

The cytotoxicity test was done in the 10^th and 21^st trials, as those had the best decolorization percentages. The evaluated samples did not show toxicity in the HepG2 cell line. All samples showed a lethal concentration of 50 (LC₅₀) > 25% w/v, which indicates that concentrations of the reaction mixture higher than 25% w/v are required to cause the death of 50% of the HepG2 cells (Table 6).

Table 6 shows the results obtained for the samples and the positive control (doxorubicin). It is important to note that LC50 was the same for all of the samples, i.e. LC₅₀ had the same value for those samples corresponding to t = 0, as for those corresponding to t = 14 h, and t = 10 h, respectively. It is concluded that the photocatalytic process presented in this work do not generate cytotoxic by-products.

Table 6. LC50 values in HepG2 cells, obtained for RB5 photocatalytic decolorization mixtures (tests 10 and 20), compared to doxorubicin (positive control).

Samples	LC50 (% w/v)
1	> 25
2	> 25
3	> 25
4	> 25
Doxorubicin	0.936

The biological evaluation is presented in Figure 7. The growth of the cells submitted to the different treatments was similar to that of the untreated control cells, i.e., the fact that the cells were in the presence of the resulting mixture of the photocatalytic process, did not affect their growth. Hence, the applied photocatalytic treatment is a viable alternative for the removal of azo dyes, since it makes it possible to achieve high decolorization percentages of RB5, with no formation of toxic products.

Similarly, Ferraz et al. ( 30 ) evaluated the cytotoxicity of the final products of the degradation of different azoic dyes after the application of a photo-electrochemical treatment with TiO₂ as a catalyst. In their study, they verified that after the process, the reaction mixture did not induce DNA damage or chromosomal mutation in the HepG2 cell line, with cell viability higher than 90%.