Degradation of Agro-Industrial Wastewater Model Compound by UV-A-Fenton Process: Batch vs. Continuous Mode

The degradation of a model agro-industrial wastewater phenolic compound (caffeic acid, CA) by a UV-A-Fenton system was investigated in this work. Experiments were carried out in order to compare batch and continuous mode. Initially, batch experiments showed that UV-A-Fenton at pH 3.0 (pH of CA solution) achieved a higher generation of HO•, leading to high CA degradation (>99.5%). The influence of different operational conditions, such as H2O2 and Fe2+ concentrations, were evaluated. The results fit a pseudo first-order (PFO) kinetic model, and a high kinetic rate of CA removal was observed, with a [CA] = 5.5 × 10−4 mol/L, [H2O2] = 2.2 × 10−3 mol/L and [Fe2+] = 1.1 × 10−4 mol/L (kCA = 0.694 min−1), with an electric energy per order (EEO) of 7.23 kWh m−3 order−1. Under the same operational conditions, experiments in continuous mode were performed under different flow rates. The results showed that CA achieved a steady state with higher space-times (θ = 0.04) in comparison to dissolved organic carbon (DOC) removal (θ = 0–0.020). The results showed that by increasing the flow rate (F) from 1 to 4 mL min−1, the CA and DOC removal rate increased significantly (kCA = 0.468 min−1; kDOC = 0.00896 min−1). It is concluded that continuous modes are advantageous systems that can be adapted to wastewater treatment plants for the treatment of real agro-industrial wastewaters.


Introduction
The rapid expansion of the agro-industry in both developed and developing countries is a major contributor of environmental pollution worldwide [1,2]. Agro-industrial wastewater characteristics are much more diverse than domestic wastewater, which is usually qualitatively and quantitatively similar in its composition. This industry produces large quantities of highly polluted wastewater containing toxic substances and organic and inorganic compounds such as: heavy metals, pesticides, phenols, and derivatives [3,4]. In the case of industries such as wine or olive oil production, large volumes of wastewaters are produced from different activities, namely tank washing, transfer, bottling, and filtration [5][6][7]. These wastewaters are characterized by an elevated content of suspended solids, low pH (3)(4)(5), and high organic load composed of phenolic compounds, sugars, organic acids, and esters [8,9]. Among these compounds, polyphenols are considered to be hazardous compounds because they are not mineralized by conventional biologic treatments [10,11], among which polyphenolic compounds are found to be the most abundant, including gallic, syringic, protocatechuic, vanillic, and caffeic acids [12][13][14][15]. Caffeic acid Table 1. Molecular formula, molecular structure, maximum absorption wavelength and molecular weight of caffeic acid (CA) [31].

Analytical Determinations
Different parameters were measured in order to determine the effect of the treatments. The dissolved organic carbon (DOC) in mg C/L and total nitrogen (TN) in mg N/L were determined by direct injection of filtered samples into a Shimadzu TOC-LCSH analyzer (Shimadzu, Kyoto, Japan), equipped with an ASI-L autosampler, provided with 324 180. 16

Analytical Determinations
Different parameters were measured in order to determine the effect of the treatments. The dissolved organic carbon (DOC) in mg C/L and total nitrogen (TN) in mg N/L were determined by direct injection of filtered samples into a Shimadzu TOC-LCSH analyzer (Shimadzu, Kyoto, Japan), equipped with an ASI-L autosampler, provided with an NDIR detector and calibrated with standard solutions of potassium phthalate. The hydrogen peroxide concentration was determined using titanium (IV) oxysulfate (DIN 38 402H15 method) at 410 nm using a portable spectrophotometer from Hach (Loveland, CO, USA), the pH and oxidation reduction potential (ORP) were measured by a 3510 pH meter (Jenway, Cole-Parmer, UK), and the iron concentrations were analyzed by atomic absorption spectroscopy (AAS) using a Thermo Scientific™ iCE™ 3000 Series (Thermo Fisher Scientific, Waltham, MA, USA).

Fenton-Based Experimental Procedure
The Fenton-based batch experiments were performed in a self-designed lab-scale reactor with 500 mL capacity and a solution depth of 1.4 cm (Figure 1). The lab reactor had a rectangular shape with a bottom and mirror walls. The UV-A LEDs system was composed of 12 Indium Gallium Nitride (LnGaN) LEDs lamps (Roithner AP2C1-365E LEDs) with a λ max = 365 nm. 250 mL of CA solution with a concentration of 5.5 × 10 −4 mol L −1 (pH = 3.90 ± 0.19, DOC = 68.7 ± 2.1 mg C L −1 ), which was constantly agitated

Electrical Energy Determination
Assuming the degradation of CA as a pseudo-first order (PFO) kinetic model (ln[CA]t = −kt + ln[CA]0), the electrical energy per order (E EO ) can be determined by conversion of the units of the first order kinetic constants to min −1 , which results in Equation (2) [33]: For the Fenton-based continuous experiments, a reservoir filled with CA with a concentration of 5.5 × 10 −4 mol L −1 and H 2 O 2 was pumped by a peristaltic pump (Shenchen, Hvidovre, Denmark) into the UV-A reactor, in which different flow rates (1-4 mL min −1 ) were applied to the aqueous mean, with a hydraulic retention time (HRT) of 15 min (Figure 1). The H 2 O 2 was added in continuous mode to prevent radical scavenging and to enhance CA degradation.
To determine the CA and DOC removal, Equation (1) was applied [32].
where C 0 and C t are the initial and final concentrations of CA and DOC.

Electrical Energy Determination
Assuming the degradation of CA as a pseudo-first order (PFO) kinetic model (ln[CA] t = −kt + ln[CA] 0 ), the electrical energy per order (E EO ) can be determined by conversion of the units of the first order kinetic constants to min −1 , which results in Equation (2) [33]: where P is the nominal power of the reactor (kW), V is the volume (m 3 ), and k obs is the pseudo-first order kinetic observed (min −1 ). All the CA removal experiments were performed in triplicate, and the observed standard deviation was always less than 5% of the reported values. Differences among means were determined by analysis of variance (ANOVA) using OriginLab 2019 software (Northampton, MA, USA), and the Tukey's test was used for the comparison of means, which were considerate different when p < 0.05, and the data are presented as mean and standard deviation (mean ± SD).

Chemical Degradability of Caffeic Acid
Considering the high content of impurities present in agro-industrial wastewaters, it is difficult to understand the degradation of the single compounds present in their constitution. Therefore, in this work caffeic acid (CA) was selected as a model compound because it exists in many types of agro-industrial wastewater. To evaluate the capacity of the UV-A reactor for the degradation of CA phenolic, several experiments were carried out: (1) CA+H 2 O 2 , (2) CA+UV-A, (3) CA+Fe 2+ +UV-A, (4) CA+H 2 O 2 +UV-A, (5) CA+H 2 O 2 +Fe 2+ , (6) CA+H 2 O 2 +Fe 2+ +UV-A. In Figure 2a are represented the results of CA removal after each treatment. From the results, it was possible to observe that CA is resistant to oxidation with the application of H 2 O 2 , UV-A, Fe 2+ +UV-A, and H 2 O 2 +UV-A, with 1.5, 4.7, 7.9, and 12.7%, respectively. With the application of H 2 O 2 , UV-A, and Fe 2+ +UV-A, there is no generation of hydroxyl radicals (HO • ) and CA shows to be resistant to degradation by radiation, H 2 O 2 , and iron. In accordance with these results, CA is harder to degrade regarding other phenolics, such as gallic acid [34]. The combination of H 2 O 2 + UV-A increased the degradation of CA, due to the conversion of H 2 O 2 into HO • radicals (Equation (3)). The highest CA removals were observed with application of the Fenton and UV-A-Fenton processes, with 92.3 and 99.9% after 15 min of reaction. An analysis of the ORP values showed higher values with the application of Fenton and photo-Fenton, which could be linked to the production of HO • radicals. To understand these results, the conversion of the H 2 O 2 and the concentration of Fe 2+ available in solution were studied (Figure 2b) [35,36], thus a higher HO • radicals generation occurred. Fe2 Fe(HO) 2+ +UV → Fe2 + + HO • (5) (Figure 2b). With the application of H2O2 and H2O2 + UV-A, there was a low consumption of H2O2 (0.24 and 0.45 mM), thus a low generation of HO • radicals occurred. With application of Fenton and UV-A-Fenton, the H2O2 consumption increased (0.80 and 1.20 mM). This increase could be due to the reaction of Fe 2+ with H2O2 (Equation (4)). This difference in H2O2 consumption could be due to the Fe 2+ available. Due to the reaction of Fe 2+ with H2O2, the Fe 2+ was oxidized to Fe 3+ , precipitating and ferric hydroxide. The UV-A was able to regenerate the Fe 3+ to Fe 2+ (Equation (5) These results were in agreement with the work of Cruz et al. [37], who observed an efficient removal of Sulfamethoxazole by the combination of catalyst + H2O2 + UV radiation.

Effect of pH
The pH of the solution has an important effect in the degradation efficiency of CA. Previous authors observed that the downward trend of HO • radical production increased with higher pH, because the Fenton reaction's optimal pH is around pH 2-4 [38]. The pH of the CA solution was varied (3.0-7.0), with results showing a CA removal of 99.9, 87.7, 43.0, and 41.3%, respectively, for pH 3.0, 4.0, 6.0, and 7.0 ( Figure 3a). As the pH increased above pH > 3.0, the availability of the iron decreased (0.090, 0.079, 0.068, and 0.063 mM, respectively) ( Figure 3b). At pH above 3.0, the Fe 3+ produced by the oxidation These results were in agreement with the work of Cruz et al. [37], who observed an efficient removal of Sulfamethoxazole by the combination of catalyst + H 2 O 2 + UV radiation.

Effect of pH
The pH of the solution has an important effect in the degradation efficiency of CA. Previous authors observed that the downward trend of HO • radical production increased with higher pH, because the Fenton reaction's optimal pH is around pH 2-4 [38]. The pH of the CA solution was varied (3.0-7.0), with results showing a CA removal of 99.9, 87.7, 43.0, and 41.3%, respectively, for pH 3.0, 4.0, 6.0, and 7.0 ( Figure 3a). As the pH increased above pH > 3.0, the availability of the iron decreased (0.090, 0.079, 0.068, and 0.063 mM, respectively) ( Figure 3b). At pH above 3.0, the Fe 3+ produced by the oxidation of Fe 2+ in the Fenton process began to precipitate in the form of amorphous Fe(HO) 3 . This formation not only decreased the iron concentration, but also inhibited the regeneration of Fe 2+ [39]. In accordance with Equations (6) and (7), the reaction between Fe 3+ and H 2 O 2 generates ferric-hydroperoxyl complexes that decomposes to produce HO • 2 radicals and Fe 2+ , as shown in Equation (8); however, these reactions have a very slow reaction rate (2.7 × 10 −3 M s −1 ) [40]. Thus, at pH > 3.0, a lower concentration of H 2 O 2 was converted to HO • radicals, which decreased the efficiency of the UV-A-Fenton process. Fe(HO)( FeHO 2+ + H2O2 → Fe(HO)(HO 2 ) + + H + Fe(HO)(HO 2 ) + → Fe 2+ + HO -+ HO 2 A similar result was observed in the work of Zha et al. [41], in the degradation of landfill leachate by the photo-Fenton process. Considering that the pH of winery wastewaters is usually between 3 and 4, the application of photo-Fenton at pH 3.0 decreases the requirement of reagent consumption for pH adjustments.

Effect of H2O2 Concentration
In this section, the effect of the oxidant (H2O2) concentration in the efficiency of the UV-A-Fenton to degrade the CA phenolic was evaluated. To keep the UV-A-Fenton competitive with other processes, it is essential that this process represents a low-cost operation, thus the control of the H2O2 concentration is implied. In observation of Figure 4a, different H2O2:CA molar ratios (R) were tested, with results showing a tendency to increase the CA removal by increasing the molar rate. With application of an R = 1 and 2, a low CA removal was observed; however, when the R = 4 was applied, a near complete removal of CA was observed, reaching a plateau after 10 min of reaction. Above this ratio, no considerable CA removal was observed. To understand these results, the H2O2 consumption and Fe 2+ concentration, along with the reaction, were studied ( Figure 4b). With application of R = 1 and 2, the H2O2 consumption was low (0.10 and 0.30 mM, respectively), despite the high concentration of Fe 2+ available (0.088 and 0.095 mM, respectively), which could mean that the H2O2 consumed was insufficient to produce HO • radicals in a necessary amount to degrade the CA phenolic. With application of R = 4, 8, and 16, the iron concentrations remained similar (0.090, 0.087, and 0.086 mM, respectively); however, a higher H2O2 consumption was observed (1.2, 1.5, and 1.9 mM), thus a higher amount of A similar result was observed in the work of Zha et al. [41], in the degradation of landfill leachate by the photo-Fenton process. Considering that the pH of winery wastewaters is usually between 3 and 4, the application of photo-Fenton at pH 3.0 decreases the requirement of reagent consumption for pH adjustments.

Effect of H 2 O 2 Concentration
In this section, the effect of the oxidant (H 2 O 2 ) concentration in the efficiency of the UV-A-Fenton to degrade the CA phenolic was evaluated. To keep the UV-A-Fenton competitive with other processes, it is essential that this process represents a low-cost operation, thus the control of the H 2 O 2 concentration is implied. In observation of Figure 4a, different H 2 O 2 :CA molar ratios (R) were tested, with results showing a tendency to increase the CA removal by increasing the molar rate. With application of an R = 1 and 2, a low CA removal was observed; however, when the R = 4 was applied, a near complete removal of CA was observed, reaching a plateau after 10 min of reaction. Above this ratio, no considerable CA removal was observed. To understand these results, the H 2 O 2 consumption and Fe 2+ concentration, along with the reaction, were studied ( Figure 4b). With application of R = 1 and 2, the H 2 O 2 consumption was low (0.10 and 0.30 mM, respectively), despite the high concentration of Fe 2+ available (0.088 and 0.095 mM, respectively), which could mean that the H 2 O 2 consumed was insufficient to produce HO • radicals in a necessary amount to degrade the CA phenolic. With application of R = 4, 8, and 16, the iron concentrations remained similar (0.090, 0.087, and 0.086 mM, respectively); however, a higher H 2 O 2 consumption was observed (1.2, 1.5, and 1.9 mM), thus a higher amount of HO • radicals were generated, leading to the degradation of the CA phenolic. The consumption of H 2 O 2 within R = 8 and 16 could be linked to the scavenging reactions between the excess of H 2 O 2 with HO • (Equations (9) and (10)), leading to the production of hydroperoxyl radicals (HO • 2 ) and superoxide anion radicals (O •− 2 ) with a lower oxidation potential (E • = 1.70 and 0.40 V) than HO • radicals (E • = 2.80 V) [42].
sumption of H2O2 within R = 8 and 16 could be linked to the scavenging reactions between the excess of H2O2 with HO • (Equations (9) and (10)), leading to the production of hydroperoxyl radicals (HO 2 • ) and superoxide anion radicals (O 2 •-) with a lower oxidation potential (E° = 1.70 and 0.40 V) than HO • radicals (E° = 2.80 V) [42]. To have a better understanding of the effect of the H2O2 in phenolic degradation, the results showed a good fitting to the pseudo-first order (PFO) kinetic model. The k values vs.
[H2O2] were plotted and two different kinetic rates were separated (Figure 5a). In a first one, up to 22 × 10 −4 mol L −1 H2O2, the PFO constant is directly proportional to the concentration of H2O2 applied, with a slope of 0.012 mol L −1 min −1 . In the second kinetic rate (up to 88 × 10 −4 mol L −1 ), the PFO constant increased linearly with the concentration of H2O2, with a slope of 0.489 mol L −1 min −1 . Based in these results, it can be assumed that the LED light fully penetrates the solution, independent of the concentration of H2O2 used, so the geometry of the UV-A reactor and the depth of the solution appeared to be adequate. These results were shown to be in agreement with the work of Li and Cheng [43], who observed a direct relation between the increase in the kinetic rate of malachite green degradation with higher H2O2 concentrations.
It is also necessary to study the how the concentration of H2O2 affects the energy consumption of the reactor, because the energy consumption determines part of the treatment costs. In Figure 5b

Effect of Fe 2+ Concentration
As previously observed in Figure 2, the Fe 2+ catalyst had a major effect regarding UV-A radiation in the conversion of H2O2 into HO • radicals. Therefore, the H2O2:Fe 2+ molar ratio (R1) was varied to study the effect of the Fe 2+ concentration in the UV-A-Fenton process (Figure 6a). The results showed that by increasing the catalyst concentration, the rate of the CA phenolic degradation was increased. Clearly, the application of an R1 from 120 It is also necessary to study the how the concentration of H 2 O 2 affects the energy consumption of the reactor, because the energy consumption determines part of the treatment costs. In Figure 5b,  × 10 −3 mol L −1 . For concentrations below 2.2 × 10 −3 mol L −1 , the degradation of CA was too slow, which implied a higher energy consumption. For concentrations higher than 2.2 × 10 −3 mol L −1 , only small gains were observed, which are not profitable, considering the high cost of H 2 O 2 [44].

Effect of Fe 2+ Concentration
As previously observed in Figure 2, the Fe 2+ catalyst had a major effect regarding UV-A radiation in the conversion of H 2 O 2 into HO • radicals. Therefore, the H 2 O 2 :Fe 2+ molar ratio (R1) was varied to study the effect of the Fe 2+ concentration in the UV-A-Fenton process (Figure 6a). The results showed that by increasing the catalyst concentration, the rate of the CA phenolic degradation was increased. Clearly, the application of an R1 from 120 to 40 was revealed to be insufficient to achieve a complete degradation of CA. Qualitatively, the results indicated that increasing the Fe 2+ concentration in solution, for a constant H 2 O 2 concentration, increased the degradation rate of CA. To understand the effect of the Fe 2+ concentration, the H 2 O 2 consumption and the concentration of iron that remained in solution were analyzed (Figure 6b). With the application of an R1 from 120 to 40, lower concentrations of H 2 O 2 were consumed, regarding the values obtained within R1 from 20 to 2, thus more HO • radicals were generated. Figure 6b shows that within the first 5 min a higher concentration of H 2 O 2 was consumed with the application of R1 = 2, which is consistent with the higher concentration of iron present, thus more HO • radicals were generated. However, from this point onwards, the H 2 O 2 consumption decreased to values lower than those observed with the application of R1 = 4, which could be a strong indicator that Fe 2+ began consuming the HO • radicals. To confirm this theory, the Fe 2+ present in the solution was determined. The results showed higher Fe 2+ concentrations available within the application of R1 from 20 to 2 (Figure 6b). A similar behavior was observed in the work of Faggiano et al. [45], in which increasing the concentration of iron to a certain extent lead to a reduction of the H 2 O 2 consumption.   (Figure 7a). In a first dominion, up to 1.1 × 10 −4 mol L −1 , the PFO kinetic constant is directly proportional to the concen-tration of Fe 2+ , with a slope of 0.071 mol L −1 min −1 . A second dominion was observed from 1.1 × 10 −4 to 11 × 10 −4 mol L −1 , in which the PFO kinetic constant increased linearly with the concentration Fe 2+ , with a slope of 0.749 mol L −1 min −1 . In a closer look, from 5.5 × 10 −4 to 11 × 10 −4 mol L −1 , the kinetic rate decreased. This could be attributed to scavenging reactions between the excess of iron present in solution and the HO • radicals generated (Equation (11)) [46]. In Table 2, it is shown research that compares the E EO values obtained by other treatment processes applied in the degradation of phenolic compounds, dyes, and compounds of emerging concern. When compared to the work of Yáñez et al. [12], the methodology created in this work was shown to be more efficient, reducing the energy required to degrade the CA phenolic. In a different work, the application of a low-pressure Hg lamp (8 W power, 250 nm) presented an ideal process to convert the H2O2 into HO • radicals [49]; however, without a catalyst, this was observed to be a slow process with high energy consumption.
In the work of Kim et al. [50], three different reactors were applied: (1) UV-C lamp (6 W power, 254 nm) with H2O2, (2) ozone batch reactor (1.12 mg O3/min), and (3) E-beam (1 MeV and 40 kW). The results showed that the ozonation process achieved worse E EO results in comparison to the UV-A reactor used in this work. Although the UV-C and Ebeam reactors showed promising results, the concentration of the SMX degraded was much lower than the concentration of CA used in this work. The UV-A reactors with an emission wavelength of 365 nm were applied in photocatalytic degradation of dyes and phenolics [33,48], with results showing efficient removals; however, they are not revealed as being competitive in comparison to the results obtained in this work. In Table 2, several examples of the application of photo-Fenton process to treat real wastewaters are also shown, such as in winery wastewater [51] and poultry slaughterhouse wastewater [52]. It is shown that the demand for electric energy to treat real wastewaters is much higher in comparison to single compounds, as is the amount of reagents required to degrade the organic carbon. Table 2. Disclosed results of electric energy per order (E EO ) obtained after degradation of single compounds by different treatment processes.  These results were in agreement with the work of Xavier et al. [47], who observed that the application of an excess of Fe 2+ increased the scavenging reactions and decreased the efficiency of the photo-Fenton process in the degradation of Magenta MB. Parallel to the kinetic rate, the energy consumption was determined to evaluate the feasibility of the reactor. The results (Figure 7b) showed two different dominions, one for [Fe 2+ ] ≤ 1.1 × 10 −4 mol L −1 and one for [Fe 2+ ] ≥ 1.1 × 10 −4 mol L −1 . These results clearly stipulate that for concentrations below 1.1 × 10 −4 mol L −1 , the reactions were too prolonged, implying larger electric consumptions, thus it could result in higher costs for treatment plants. For application of Fe 2+ above 1.1 × 10 −4 mol L −1 , the gains were very low and do not justify the application of those amounts of iron. The selection of the appropriate radiation type with the correct geometry influences the kinetic rate of degradation and the costs of treatment. This fact was observed in the work of Tapia-Tlatelpa et al. [48], who observed from testing UV lamps with different geometries, that LED lamps with a radial position (similar to this work) achieved the lowest energy consumption.

AOP processes Conditions
In Table 2, it is shown research that compares the E EO values obtained by other treatment processes applied in the degradation of phenolic compounds, dyes, and compounds of emerging concern. When compared to the work of Yáñez et al. [12], the methodology created in this work was shown to be more efficient, reducing the energy required to degrade the CA phenolic. In a different work, the application of a low-pressure Hg lamp (8 W power, 250 nm) presented an ideal process to convert the H 2 O 2 into HO • radicals [49]; however, without a catalyst, this was observed to be a slow process with high energy consumption. Table 2. Disclosed results of electric energy per order (E EO ) obtained after degradation of single compounds by different treatment processes.  [54] In the work of Kim et al. [50], three different reactors were applied: (1) UV-C lamp (6 W power, 254 nm) with H 2 O 2 , (2) ozone batch reactor (1.12 mg O 3 /min), and (3) E-beam (1 MeV and 40 kW). The results showed that the ozonation process achieved worse E EO results in comparison to the UV-A reactor used in this work. Although the UV-C and E-beam reactors showed promising results, the concentration of the SMX degraded was much lower than the concentration of CA used in this work. The UV-A reactors with an emission wavelength of 365 nm were applied in photocatalytic degradation of dyes and phenolics [33,48], with results showing efficient removals; however, they are not revealed as being competitive in comparison to the results obtained in this work. In Table 2, several examples of the application of photo-Fenton process to treat real wastewaters are also shown, such as in winery wastewater [51] and poultry slaughterhouse wastewater [52]. It is shown that the demand for electric energy to treat real wastewaters is much higher in comparison to single compounds, as is the amount of reagents required to degrade the organic carbon.

Experiments in Continuous Mode
Previous sections have shown that batch reactors fitted with UV-A LED lights achieved a near complete removal of CA from the aqueous solution. It was also observed in the H 2 O 2 and Fe 2+ concentration variations sections that the degradation of CA occurred in two phases, a faster one within the first 5 min, followed by a slower rate, reaching a plateau. This occurs because, in the first period, the Fe 2+ reacted very fast with the H 2 O 2 (k = 78 mol −1 dm 3 s −1 ), generating a high content of HO • radicals. Following this period, the ferric ions produced earlier reacted with the H 2 O 2 producing HO • 2 , with a much lower kinetic rate (k = 0.02 mol −1 dm 3 s −1 ) [55]. Therefore, these results indicate that the residence time (τ, min) within the continuous mode is an important parameter to be considered. In an ideal continuous reactor, the space-time is equal to the mean residence time, and is determined in accordance with Equation (12) [55].
where V is the volume of the reactor (mL) and F is the flow rate (mL min −1 ). In Figure 8a are shown the removal results of CA and DOC as a function of dimensionless reaction time values (θ = t/τ), in which each θ is comparable to one τ [56]. The results showed that a CA steady state was achieved after 0.04 space-time values, while the mineralization of CA showed a steady state from 0 to 0.020 space-time values. By the analysis of the results obtained, the flow rate had a significant influence on the removal rate of CA and mineralization, in which, with F = 4 mL min −1 , higher CA and DOC removal was achieved (99.2 and 35.9%, respectively). The Fe 2+ concentration analysis revealed that the availability of iron remaining in the solution was similar to that observed in batch experiments; therefore, the concentration of H 2 O 2 was revealed to be the decisive factor. In observation of Figure 8, it can be seen that the application of F = 1 and 2 mL min −1 led to considerable lower additions of H 2 O 2 regarding the application of F = 4 mL min −1 , thus less HO • radicals were generated. These results were in agreement with the work of Rosales et al. [30], who observed that increasing the residence time enhanced the generation of HO • radicals, increasing the discoloration of textile wastewaters. To understand the results obtained by the continuous mode, the kinetic constants were determined by application of a continuous mode simplified model, assuming a first order reaction Equation (13) [33]. kτ= 1-C ss C 0 C SS C 0 (13) where C0 and CSS are the initial and steady state concentrations of CA (mol L −1 ) and DOC (mg C L −1 ). In Table 3 are shown the kinetic rate constants obtained under different flow rates. The results are plotted in Figure 9, showing that as the flow rate increases, the kinetic rate of CA and DOC also increased, meaning that it achieved a good penetration of the UV-A radiation into the water. Considering this behavior, the continuous mode in association with the UV-A LEDs with horizontal geometry could be considered as a good system for phenolic degradation. To understand the results obtained by the continuous mode, the kinetic constants were determined by application of a continuous mode simplified model, assuming a first order reaction Equation (13) [33].
where C 0 and C SS are the initial and steady state concentrations of CA (mol L −1 ) and DOC (mg C L −1 ). In Table 3 are shown the kinetic rate constants obtained under different flow rates. The results are plotted in Figure 9, showing that as the flow rate increases, the kinetic rate of CA and DOC also increased, meaning that it achieved a good penetration of the UV-A radiation into the water. Considering this behavior, the continuous mode in association with the UV-A LEDs with horizontal geometry could be considered as a good system for phenolic degradation. Table 3. Kinetic constants from CA and DOC removal observed for the assays with different flow rates (mL min −1 ). Means in the same column with different letters represent significant differences (p < 0.05) within each parameter (k CA and k DOC ) by comparing flow rates.   To understand the results obtained by the continuous mode, the kinetic constants were determined by application of a continuous mode simplified model, assuming a first order reaction Equation (13) [33]. kτ= 1-C ss C 0 C SS C 0 (13) where C0 and CSS are the initial and steady state concentrations of CA (mol L −1 ) and DOC (mg C L −1 ). In Table 3 are shown the kinetic rate constants obtained under different flow rates. The results are plotted in Figure 9, showing that as the flow rate increases, the kinetic rate of CA and DOC also increased, meaning that it achieved a good penetration of the UV-A radiation into the water. Considering this behavior, the continuous mode in association with the UV-A LEDs with horizontal geometry could be considered as a good system for phenolic degradation.

Conclusions
This work attempted to study the effect of several parameters in the kinetic rate and energy consumption of the hydroxyl radical-based oxidation processes. The experiments were carried out employing a UV-A LED system emitting at 365 nm, H 2 O 2 as the oxidant, and Fe 2+ as the catalyst. Based in the results, the UV-A-Fenton process at pH 3.0 achieved the highest CA degradation rate. It is concluded that increasing the H 2 O 2 concentration increases the kinetic rate of CA degradation; however, above 2.2 × 10 −3 mol L −1 , the reactor is not economically viable, because no significant E EO reduction is observed. It is concluded that the Fe 2+ achieves the best performance, with 1.1 × 10 −4 mol L −1 , with an E EO = 7.23 kWh m −3 order. Above this concentration, the kinetic rate increases; however, no significant gains in energy consumption were observed. When batch conditions are applied to continuous mode, it is concluded that the degradation of CA is dependent on the flow rate. It is also concluded that higher flow rates lead to higher CA and DOC removals (k CA = 0.468 min −1 ; k DOC = 0.00896 min −1 ).
Based in the results obtained from this work, a UV-A LED reactor adapted to continuous mode could provide an excellent process to reduce the soaring energy costs associated with these systems and make them viable for wastewater treatment plants.