Combating paraben pollution in surface waters with a variety of photocatalyzed systems: Looking for the most efficient technology

2.1 Chemicals

Methyl-paraben (CAS Number 99-76-3), ethyl-paraben (120-47-8), propyl-paraben (94-13-3), and butyl-paraben (94-26-8) were of the highest purity (>98%) and were purchased from Sigma (Spain). A stock solution of the four parabens (5 mg dm^-3) was created with high purity water (from a Millipore Milli-Q^TM system). This initial concentration of parabens was selected in order to conform with a previous literature survey, which studied several experimental works that had the aim of removing parabens from WWTP influents or effluents [14]. The exact concentration of hydrogen peroxide (nominally 33% w/v, analytical grade, Sigma-Aldrich) was determined by a photometric method using Titanium (IV) oxysulfate-sulfuric acid solution (Fluka 89532) as reactant. FeSO₄·7 H₂O (99.9% analytical grade Sigma-Aldrich) was purchased from Merck (Spain), and the Fe²⁺ concentration was determined by permanganometry. TiO₂ (Degussa P25) was purchased from Degussa Portugal (AEROXIDE®-P25).

Surface water aqueous matrices were obtained from the Guadiana River (Badajoz, SW Spain) (type I), the reservoir of Peña del Águila (Villar del Rey, Badajoz, SW Spain) (type II), and secondary WWTP-Badajoz effluent (type III). Table 1 shows the physicochemical characterization of these surface -aqueous matrices.

2.2 Experimental

The reactor was filled with an aqueous solution containing the four parabens (5 mg dm^-3). A total volume of 0.350 dm³ of the solution was used for all experiments, excepting Fenton (250 ml). Finally, in all cases, the reaction was stopped by addition of bisulfite at time intervals. In every experience involving UV radiation, the experiments were performed in a cylindrical glass and the reactor was equipped with an axial radiation source (Heraeus TNN 15/32 lamp, 15W, λ = 254 nm). The constant operating conditions for all the systems under study (based in our previous studies) were: [H₂O₂]=2.7·10^-4 mol dm^-3, [Fe²⁺]=2.7·10^-5 mol dm^-3 and [TiO₂]=3.83·10^-4 mol dm^-3.

2.3 Analytical

MP, EP, PP and BP were determined by HPLC using a Waters HPLC chromatograph, equipped with a photodiode Array 996 detector and a Phenomenex-Pak C-18 column (5 μm, 150 mm x 3.9 mm). Well-defined peaks were obtained at 2.2, 2.8, 4.2 and 6.8 min for MP, EP, PP and BP, respectively. The mobile phase was a 60:40 methanol:water mixture (10⁻² mol dm^-3 orthophosphoric acid) in isocratic operation mode.

2.4 Design of experiments (DoE)

The removal efficiency of parabens after 30 minutes was considered as the response variable (Y). The operating levels of the statistical design of experiments (DoE) for the photo-Fenton process were: [H₂O₂]₀ codified values (-1, +1) corresponds to (4.31·10^-5 mol dm^-3, 2.51·10^-4 mol dm^-3), respectively; in the case of [Fe²⁺]₀ these codified values were (4.31·10^-6 mol dm^-3, 2.5·10^-5 mol dm^-3) respectively.

Ethical approval: The conducted research is not related to either human or animal use.

3.1 Preliminary experiments

The generation of hydroxyl radicals in Fenton processes is strongly influenced by several factors, including pH and initial concentrations of the reagents (namely, H₂O₂ and Fe²⁺). Due to the scarce solubility of iron hydroxides, the pH range to be used in Fenton processes is restricted to values of 4.5 and below. Furthermore, according to the experience of the authors, optimum results are attained in terms of removal efficiency of this kind of compounds by using a [H₂O₂ ]₀ : [F e²⁺]₀ ratio equal to 10 [40, 41]. However, the selection of an adequate initial concentration of hydrogen peroxide is far from easy. Hence, in order to optimize the initial H₂O₂ concentration, a set of preliminary experiments were performed with [H₂O₂]₀ ranging from 1.25·10⁻⁵ mol dm^-3 to 2.50 ·10⁻⁴ mol dm^-3.

The results of the preliminary trials with MP are shown in Figure 1. According to these results, it can be concluded that, as a rule, the removal efficiencies after 60 min of treatment are relatively low (at most, 54.9, 58.0 and 67.0%) for MP, EP and PP, respectively. For BP, on the contrary, up to 76.9% of removal is attained even using low initial concentrations of H₂O₂. For higher initial concentrations of this reagent, namely 1.25 and 2.50·10⁻⁵ mol dm^-3, complete abatement of the parabens is achieved after 45-60 minutes of treatment. It is worth noting that above a given concentration of hydrogen peroxide no remarkable effect is observed in terms of removal efficiency of the pollutants with an increase in hydrogen peroxide concentration. This saturating effect is in good agreement with the fact that H₂O₂ may serve as a radical scavenger as suggested by Eq. (4) and the formation of hydroperoxyl radicals (HO₂·), a chemical species with a lower oxidation potential than HO·.

Figure 1

MP removal efficiency (%) vs. time as at different initial concentrations of hydrogen peroxide. [MP]₀=5 mgdm^-3, [H₂O₂ ]₀ : [F e²⁺]_{0 =} 10. T=25 ^⍛C.

Taking into consideration these results, an experimental design was planned using an initial H₂O₂ concentration of 1.25·10⁻⁵ mol dm^-3 as the central point (0) and 2.50·10⁻⁵ mol dm^-3 as the factorial point (+1).

3.2 Parabens abatement by several AOPs

The oxidation of the four pollutants, simultaneously in ultrapure water, was first performed with Fenton’s reagent (H₂O₂/Fe²⁺) and with the following AOPs that employ UV radiation: UV/ H₂O₂, UV/Fe²⁺/ H₂O₂ (photo-Fenton system), UV/TiO₂ and UV/TiO₂/H₂O₂. In order to perform a comparison of all of the AOPs here studied, the experimental conditions indicated in Section 3.1 were considered as optimal and consistently used in all cases.

The removal efficiencies for MP by all of the examined AOPs are shown in Figure 2. Similar results were obtained for EP, PP and BP (data not shown). It can be readily concluded that the photo-Fenton (UV/Fe²⁺/H₂O₂) ternary system yields the most effective process. The other ternary system (i.e., UV/TiO₂/ H₂O₂) also exhibits removal efficiencies that are larger than those of the binary systems, i.e., Fenton (Fe²⁺/ H₂O₂), UV/ H₂O₂ and UV/TiO₂, but in this case, the efficiencies are more similar to those of the binary systems. The sequence shown in Figure 2 suggests that, in combination with H₂O₂, Fe²⁺ is somewhat more efficient than TiO₂. The synergistic effect observed for the photo-Fenton system is consistent with results obtained in the 1990s [42, 43, 44] and more recently by Serna-Galvis et al. [45]. Figure 2 confirms that UV radiation accelerates the Fenton process, thus improving the degradation of the parabens. A reasonable explanation for this result is the fact that during the reaction, Fe²⁺ oxidizes to Fe³⁺ (Eq. (2)) and that the regeneration of the ferrous ions occurs in a process that is mediated by UV radiation (Eq. (3)), thus facilitating the production of hydroxyl radicals (Eq. (4) and Eq. (5)).

Figure 2

Methylparaben removal efficiency (%) for the different AOPs. [MP]₀=5 mg dm^-3, [H₂O₂]₀=2,7·10^-4 mol dm^-3, [Fe²⁺]=2,7·10^-5 mol dm^-3, [TiO₂]₀=0,38 ·10^-3 mol dm^-3, T=25 ^⍛C. UV radiation source = Heraeus TNN 15/32 (15W).

Moreover, the Fenton system is more effective than UV/H₂O₂ because the hydroxyl radicals generated in this latter system by the photocatalytic degradation of H₂O₂ (Eq. (1)) are consumed during the reaction [46]. In the Fenton system, the rate of degradation of the contaminant increases, since the reaction of hydroxyl radical generation may still occur in the medium by the reduction of ferric ions to ferrous ions (Eq. (5)) in the presence of H₂O₂. Furthermore, titania catalyzes the reaction of the organic contaminant by photodegradation due to the intrinsic nature of its electrical structure [47] so that it facilitates the ·OH radical generation. However, the presence of H₂O₂ exerts an important influence on the photodegradation process of the organic compounds, because H₂O₂ acts as a powerful oxidant, generating large amounts of ·OH radicals [48]. In the UV/TiO₂ process, the contribution of radical oxidation is very low, like that of the photooxidation process. However, in the UV/H₂O₂ process, oxidation by free radicals (especially ·OH) is much more remarkable due to the presence of H₂O₂ [49].

These results allow us to conclude that the photo-Fenton system is clearly the most effective process for the destruction of these pollutants. Therefore, this system was studied in more detail.

3.3 Photo-Fenton Experimental Design

Two operational variables (namely, A: [H₂O₂]₀ and B: [Fe²⁺]₀), their mutual interaction (AB), and their respective quadratic effects (A² and B²) are considered in the ANOVA test. A nonlinear equation regression was performed (Eq. (6)) for each paraben compound.

Table 2 shows the fitting coefficients obtained for the removal of the different parabens. The positive influences of [H₂O₂]₀ and [Fe²⁺] are due to the fact that these variables positively impact the generation of ·OH radicals. The value of the correlation factor (R²) reaches 0.959, which is consistent with adequate modelling of the process.

Table 3 lists the values of [H₂O₂]₀ and [Fe²⁺]₀ that optimize the removal of each contaminant separately. It can be clearly appreciated that, for all parabens, with respect to [Fe²⁺]₀ an optimum for the separate removal has been found in the range of coded values comprised between 0.683 and 0.776 (excepting for propyl-paraben, 1.410). This corresponds to initial concentrations of 2.17·10⁻⁵ - 2.27·10⁻⁵ mol dm^-3. However, for [H₂O₂]₀, no optimum is predicted in this range. Nevertheless, the initial concentration of hydrogen peroxide should not be excessively increased due to the aforesaid scavenging effect. Table 3 also shows the coded and natural values of [H₂O₂]₀ and [Fe²⁺]₀ that optimize the removal of all contaminants simultaneously [50]. Operating under these conditions (i.e., [H₂O₂]₀ = 2.92·10⁻⁴ mol dm^-3 and [Fe²⁺]₀ = 1.85·10⁻⁵ mol dm^-3), total removal of the four compounds should be obtained.

3.3.1 Graphical analysis

A model was developed according to five factors, which correspond to Eq. (6). The so-called Pareto plot (Figure 5) may be considered as a plot of the results of the ANOVA test. Grey bars represent negative factors, and white bars correspond to positive factors.

The main effect plots allow us to conclude that [H₂O₂]₀ represents a positive influence in this range of study. This fact is consistent with an increase in the formation of ·OH radicals. With respect to [Fe²⁺]₀, in all cases except for PP, the curve shows a maximum. This suggests that the presence of ferrous ions favours the oxidation process due to the catalytic role of Fe²⁺, but only up to a coded value around 0.5-0.7. Larger coded values exert a negative effect on the removal of all compounds except for PP.

Figure 3 displays these results regarding BP degradation. The response surface is remarkably convex along with the whole operational range. Both variables (namely, [H₂O₂]₀ and [Fe²⁺]₀) affect the compound removal in a similar manner but only for [Fe²⁺]₀ can an optimum be determined.

Figure 3

Response surface and contour plot for the removal of BP by photo-Fenton system. Influence of initial concentration of hydrogen peroxide and ferrous salt. T=25 ^⍛C. [BP]₀=5 mg dm^-3

3.4 Influence of the aqueous matrix on the photo-Fenton system

With regard to the influence of the aqueous matrix, three different aqueous surface waters type I (Guadiana River), type II (Peña del Águila reservoir) and type III (Badajoz-WWTP effluent) were tested. Experiments were carried out at the previously determined optimal conditions for the Photo-Fenton process. Figure 4 shows the evolution of the EP removal vs. time in the four aqueous matrices (including Milli-Q ultrapure water). Similar results were obtained for MP, PP and BP. As can be seen, the removal efficiency follows the trend ultrapure water ≈ river water > WWTP effluent ≈ reservoir water. It can be stated that the removal is faster when ultrapure water is used, followed by river water and WWTP effluent. The removal of the parabens from reservoir water is very slow in all cases.

Figure 4

Ethyl-paraben removal in different aqueous matrices by the photo-Fenton process in optimal conditions [EP]₀=5 mg dm^-3, [H₂O₂]₀ = 2.92·10⁻⁴ M and [Fe²⁺]₀ = 1.85·10⁻⁵ M, T=25 ^⍛C, Heraeus TNN 15/32 lamp, 15W, λ = 254 nm.

The removal rate of pollutants in different aqueous matrices by Fenton process -- or its derivatives -- has been investigated in recent years, and some controversies have arisen. For instance, Derbalah et al. [51] reported that the degradation of fenitrothion in river water by Fenton and photo-Fenton processes was slightly faster than in pure water and attributed this fact to the presence of Fe²⁺ and Fe³⁺ in natural water, which caused a higher rate of .OH radicals generation. Similarly, Moncayo-Lasso et al. [52] attribute the faster mineralization of dihydroxy-benzene by a photo-Fenton process in river-water compared to ultrapure water to the concurrence of three facts, namely i) the formation of photoactive intermediaries that contribute to mineralization by photoassisted electron transfers; ii) the presence of Fe³⁺ ions in river waters, that results in the generation of singlet oxygen, H₂O₂ and ·OH radicals that attack the dihydroxy-benzene compounds, and iii) the presence of humic acids that, at adequate pH (3-5) can form complexes with the Fe³⁺, increasing the amount of photoactive Fe³⁺ and therefore the efficiency of the process. Also, Jacobs et al. [53] reported that the presence of fulvic acids in river waters accelerates the degradation rate of caffeine approximately threefold.

However, more recently, Veloutsou et al. [54] reported a delay in the mineralization rate of atenolol and metoprolol in river waters by the photo-Fenton process with respect to an ultrapure aqueous matrix. An increase in the total consumption of H₂O₂ was reported, too. These authors indicate that hydroxyl radicals and other oxidizing species are neutralized by attacking other organic molecules naturally present in river waters. Also, ·OH radicals may be involved in reactions with inorganic ions such as chloride, nitrate, sulphate or phosphate. Nevertheless, the total disappearance of the initial compounds was also achieved after 10 min of illumination in river water as it is the case in the present study.

Likewise, Benitez et al. [55] reported a higher removal rate in ultrapure water when compared to other surface-water matrices. These authors asserted that the dissolved organic matter that naturally occurs in real water matrices may deplete at least part of the oxidizing species and, hence, the total amount of oxidizing agents available for the degradation of the target pollutants decreases. Obviously, in the case of ultrapure water, the emitted radiation and the oxidizing agents generated in situ are almost totally devoted to the degradation of the pollutants, which results in higher removal rates. The detrimental effect of the presence of non-dissolved organic matter in the aqueous media has also been reported by Polo-Lopez et al. [56] when investigating the removal of phytopathogen fungi spores in natural effluents.

On the other hand, Neamtu et al. [57] indicated that the presence of inorganic anions dissolved in the real water matrices could exert a coordination effect over the Fe²⁺ and, mostly, Fe³⁺ ions that are involved in the photo-Fenton process. For instance, the presence of chloride ion in real waters favours the formation of thermodynamically stable Fe³⁺ complexes as, for instance, FeCl₂⁺ or FeCl²⁺. Furthermore, sulphate ions may also form complexes such as Fe[(SO₄)]₂⁺. Consequently, the regeneration of Fe²⁺ ions is handicapped and the process efficiency decreases markedly. In addition, Fe²⁺ and Fe³⁺ ions may also form stable complexes with certain organic compounds that naturally occur in real waters, which also contribute to lower the amount of oxidizing species available to react with the pollutant. The same conclusion was reached by De Luna et al. [58] when investigating the degradation of dimethyl sulfoxide by different Fenton processes.

Finally, but not less important, it must be taken into consideration that the presence of organic matter in real water matrices may hinder the penetration of light due to increased turbidity. Table 4 summarizes the operational conditions that lead to optimum removal of different pollutants using the photo-Fenton process. It is worth noting that the experimental conditions that resulted in an optimized removal of the target molecules are, as a rule, similar to those reported in this work (i.e., [H₂O₂]₀ to [Fe²⁺]₀ ratio equal to 15.8 and pH = 3.5).