With the development of industrialization, the traditional physical, biological and chemical methods cannot meet the requirements of efficient wastewater treatment of refractory organics [1-3]. Advanced oxidation processes (AOPs) are expected to reach high requirements and strict regulations in organic effluent disposal [4]. AOPs include photocatalysis [5-7], catalytic ozonation [8], catalytic wet air oxidation [9, 10], electrochemical oxidation [11, 12], and Fenton reaction [13, 14]. One of the most widely used is Fenton oxidation, that is, ferrous ion (Fe²⁺) activates hydrogen peroxide (H₂O₂) to produce hydroxyl radical (^•OH) [15]. The ^•OH has high oxidation ability, which can quickly and non-selectively degrade various refractory organics in water [16].

Fenton system has some of the advantages, such as mild reaction conditions, simple equipment, convenient operation, and high oxidation rate [17]. However, the pH range of Fenton reaction is narrow (pH 3-5) [18]. H₂O₂ is easy to be decomposed, leading to the decrease of utilization [19]. Besides, the liquid H₂O₂ is inconvenient for storage and transportation. Moreover, Fe²⁺ is hard to recover and liable to produce the precipitation of ferric ion (Fe³⁺) [20]. These factors limit the Fenton reaction efficiency and its further application.

To overcome the bottlenecks of conventional Fenton reaction, the improved Fenton and Fenton-like processes have been well developed recently [21]. Introducing complexing agent into Fenton reaction is the most common way, which can effectively prolong the existence time of dissolved Fe²⁺ and Fe³⁺, broaden the application range of pH value, promote electron transfer, and prevent iron precipitation [22]. The usual chelation agents are ethylenediaminetetraacetic acid and hydroxylamine, but they would generate toxic products during reactions and secondary pollution [23-25]. So diverse small molecule organic acids have attracted great attention, such as ascorbic acid (AA) [26], citric acid [27] and oxalic acid [28]. These carboxylates could not only have the favorable complexing capability, but also be nontoxic and degradable.

Calcium peroxide (CaO₂) is a stable solid peroxide, which is easy to transport and store. CaO₂ could slowly and stably produce H₂O₂ over a broad range of pH, reducing the disproportionation of H₂O₂. Hence, CaO₂ has been extensively researched as the substitution of H₂O₂ [29]. On the other side, Fe³⁺ is more stable than Fe²⁺, and Fe³⁺ could also catalyze H₂O₂ by initiative superoxide and per hydroxyl actuated reactions to generate a series of reactive oxygen species [30], such as ^•OH and superoxide anion free radical (O₂^•−). But the direct application of Fe³⁺ would easily form iron sludge, exhausting the Fenton reactivity. Also, the reduction of Fe³⁺ to Fe²⁺ is slowly.

To sum up, using the AA complexed Fe³⁺ to react with CaO₂ is a prospective improved method to surmount the application barriers of traditional Fenton reaction, which is seldom reported. In addition, response surface method (RSM) combines math and statistics means to model and optimize the influence of a few independent variables, which is widely and efficiently applied for the modelling and optimization in the various engineering issues. Therefore, we established a Fe³⁺-AA complex catalyzed CaO₂ (Fe³⁺/AA/CaO₂) Fenton-like process to promote the removal of multiple organic dyes in water, and the operation parameters (Fe³⁺ concentration, AA dosage, and CaO₂ amount) were optimized by the RSM for the methyl orange (MO) decolorization. Furthermore, the H₂O₂ decomposition, Fe ions change, and active species identification were studied to reveal the synergistic reaction mechanisms.

The sources properties of chemicals applied in this study were introduced in Text S1 (Supporting Information). The tests were implemented in flask (100 mL) at room temperature on magnetic stir. The desired concertation of simulated wastewater was added in the beaker, and its initial pH was regulated by sulfuric acid and sodium hydroxide. Then, the reaction was initiated through adding Fe³⁺, AA, and CaO₂ in order. Sample aliquots (1 mL) was extracted at the given time intervals, and added 1 mL sodium thiosulfate to quench oxidation process, then filtered by 0.22 µm filter membrane for determination of concentration. The dye concentration was measured by the UV-spectrophotometer at different specific wavenumbers. The total organic carbon (TOC) was monitored through TOC analysis meter. The H₂O₂ amount was determined by titanium potassium oxalate method [31]. The concentrations of Fe³⁺ and total Fe were detected by 1, 10-phenanthroline monohydrate color method [32]. The determination of Fe⁴⁺ was used by dimethyl sulfoxide (DMSO) oxidation method [33]. Tert-butyl alcohol (TBA), p-benzoquinone (BQ) and L-histidine were used to quench ^•OH, O₂^•−, and singlet oxygen (¹O₂), respectively. The qualitative test for active species was determined through electron spin resonance (ESR) spectrometer. The detailed analyses parameters were described in Text S2 (Supporting information).

The RSM from Box-Behnken was designed on the basis of Design-Expert 8.0.6 Trial software. The experimental design of three factors (Fe³⁺ dose, AA concentration, and CaO₂ amount) and three levels was adopted to investigate the MO removal. The detailed design and analysis of RSM described in Text S3 (Supporting information), and the numbered and actual values for the variables were presented in Table S1 (Supporting information).

Fig. 1a compares the MO removal by the Fe³⁺/AA/CaO₂ and other control processes. There were no MO decolorization by the Fe³⁺ alone, AA alone, CaO₂ alone, AA/CaO₂, and Fe³⁺/AA systems, and only 28.90% of MO was decolored in the Fe³⁺/CaO₂ after 15 min treatment. Nevertheless, 93.40% of MO was removed by the Fe³⁺/AA/CaO₂, indicating the significant improvement in the synergistic process. In contrast, the MO removals were 25.40% and 15.30% for the Fe³⁺/H₂O₂ and Fe³⁺/AA/H₂O₂, respectively, which was clearly inferior to the CaO₂ relevant processes. Above phenomena could be explained through the following reasons: firstly, the dissolved CaO₂ could stably supply H₂O₂ for the Fenton-like system and release O₂ simultaneously (R1 and R2) [34], avoiding the disproportionation of H₂O₂. Secondly, after adding the AA, the complexation of Fe³⁺ and AA could be formed (R3) [34, 35], which could accelerate the electron transfer, improving the Fe²⁺ formation from the Fe³⁺ reduction and decreasing the precipitation of Fe³⁺. Thirdly, except for the conventional Fenton reactions for the generations of ^•OH and O₂^•− (R4 and R5), the AA mediated Fe³⁺/CaO₂ process could produce more reactive species, such as O₂^•− and ¹O₂ (R6 and R7) [36, 37], which are the more selective species for oxidizing the electron-rich groups in the organic compounds.

(1)

(2)

(3)

(4)

(5)

(6)

(7)

Fig. 1b shows the change of MO absorption peak (465 nm) with treatment time. The characteristic peak intensity decreased with the increase of time, which indicated that the azo bond of MO was attacked and cleaved by the active substances [38, 39]. At the same time, the peak intensity at less than 300 nm increased gradually, proving that more small molecules were formed during the MO removal [40]. The mineralization of MO was verified by the TOC analysis. As show in Fig. 1c, the TOC removal enhanced with the treatment time, and reaches 19.40% after 120 min, proving that the dye molecular structure was decomposed into lesser organic matters and mineralized into H₂O and CO₂ under the synergistic process. The comparatively low TOC removal could be attributed to the addition of organic acid complexing agent, which would compete with the reactive species generated in the synergy and enhance the organic load of wastewater. Besides, the oxidation performance of Fe³⁺/AA/CaO₂ for different organic dyes were determined (Fig. 1d). Other four dyes (Congo red (CGR), acid fuchsin (AF), methylene blue (MB), and reactive brilliant blue (KN-R)) were treated under the Fe³⁺/AA/CaO₂ system with the identical conditions, and their decolorization efficiencies all were higher than 84.00% after 15 min reaction, certifying this coupling system has a good ability to treat various dye wastewaters.

Table S2 (Supporting information) lists the three levels and three variables full factor analysis of each experiment based on RSM results, including the experimental and predicted values of MO removal. According to the results and data analysis, the quadratic linear regression equation obtained by running Design-Expert 8.0.6 trial software was as follows [41, 42]:

(10)

Table S3 (Supporting information) displays the results of variance analysis. The P-value was less than 0.0001, indicating that the result was significant, and the 51.23 of F-value was a further proof of the significant model [43]. Meanwhile, due to the influence of interference error, only 0.01% chance would appear such a large F-value. The R² value indicates that the model is to what extent impact the variability of data. In this study, the R² value of quadratic linear regression model was 0.9850, demonstrating that 98.50% of the MO removal efficiency was affected by the independent variables, and only 1.50% of influence was unaffected by the model. R_adj² is optimized for R² by considering the covariates or predictors, which is more suitable to compare those models with various of arguments. The value of R_adj² (0.9658) was close to that of R², ensuring the reliability of quadratic model. In addition, the signal to noise ratio of 20.199 showed that the result was feasible for this model [44].

Verification of model feasibility is a vital process, which is the indispensable way to evaluate whether the result is suitable. The linear distribution points of normal probability plots can prove the residuals obey normal distribution. As shown in Fig. S1a (Supporting information) the points were distributed on both sides of the straight line, proving it obeyed normal distribution. Also, the result suggested that the quadratic formula given by the model could well predict the effect of Fe³⁺, AA and CaO₂ on the MO decolorization. Fig. S1b (Supporting information) presents the comparison of MO removal ratio for the predicted and experimental values. All the spots were dispersed around the regression curve, testifying that the simulated outcomes were consistent with the actual data. From Fig. S1c (Supporting information), the residuals were randomly distributed around 0 with the change of ± 3.0, indicating that the response of model was identical to the normal distribution. Fig. S1d (Supporting information) reveals the matching correlation for the residual and simulative values. The residual was a random distribution, manifesting that the change of original value was constant for the response value [44].

The response surface and contour could manifest the interaction of various factors on the decolorization of MO. Fig. 2a shows the effect of Fe³⁺ and AA concentrations on the MO removal. The dose of CaO₂ was fixed at 3 mmol/L. The decolorization of MO underwent an increase and decrease with the increscent AA amount (0.25 to 1.75 mmol/L). This is because the excessive AA would react with active substances. With the increase of Fe³⁺ concentration from 1 mmol/L to 3 mmol/L, the removal of MO also enhanced, proving that the Fe³⁺ was a crucial factor as a catalyst in this synergistic method. Fig. 2b confirmed that there was interaction between Fe³⁺ and AA, then the generated functions of complexation and reduction could improve the Fenton-like reaction and MO removal. Fig. 2c reveals the effect of Fe³⁺ and CaO₂ concentrations on the decolorization of MO. The concentration of AA kept constant. When AA was 1 mmol/L, the MO removal increased with the enhancing CaO₂ amount. This is due to the fact that the more oxidant added, the more active species would be produced in the system, benefiting the dye removal. Fig. 2d also verified the interrelationship between Fe³⁺ and CaO₂. Figs. 2e and f present the effect of AA and CaO₂ quantities on the MO elimination. The Fe³⁺ concentration remained at 2 mmol/L. With the decrease of AA dosage and the increase of CaO₂ amount, the MO decolorization was augmented. The result suggested that there was an optimum for the AA addition.

As the predesigned experiments were completed, Design-Expert 8.0.6 was applied to optimize parameters for the Fe³⁺/AA/CaO₂ system. The simulation result indicated that the best MO decolorization was 99.30% under the optimal conditions of 2.76 mmol/L Fe³⁺, 0.68 mmol/L AA and 4 mmol/L CaO₂. Furthermore, the proof test demonstrated that the elimination of MO could reach 98.90% under the same conditions. The experimental outcome was agreed well with the predicted value, attesting that the RSM model was reliable and its optimization was successful.

Owing to the significant influence of solution pH on the Fenton process, the MO removal in the Fe³⁺/AA/CaO₂ was investigated at the different initial pH. Fig. 3a displays that the MO eliminations varied little when the wastewater pH changed from 3 to 11, and the decolorizations were all higher than 90%. The result demonstrated that the presented collaborative Fenton-like system could perform well in a wide range of pH.

Fig. 3b presents change of H₂O₂ concentration during the diverse processes. The H₂O₂ amounts all rose at the beginning and then declined with the time for the three experiments, and their sequence was: CaO₂ > Fe³⁺/CaO₂ > Fe³⁺/AA/CaO₂. The phenomena suggested that the addition of AA could be beneficial to improve the H₂O₂ decomposition.

The variations of Fe³⁺ and Fe²⁺ concentrations during the Fe³⁺/CaO₂ and Fe³⁺/AA/CaO₂ systems are shown in Figs. 3c and d, respectively. The total Fe ions content of Fe³⁺/AA/CaO₂ remained constant, and which was decreased sharply with time in the Fe³⁺/CaO₂. Besides, the Fe²⁺ concentration of the complexation system was obviously higher than that of Fe³⁺/CaO₂. Above result verified the complexing function of AA, which could not only maintain the dissolved state for Fe³⁺ and Fe²⁺, but also improve the remarkably promote the reduction of Fe³⁺ and reduce the iron sludge generation.

The reactive oxygen species masking experiment was employed to investigated the functions of various active species in the Fe³⁺/AA/CaO₂ system (Fig. 4). TBA can effectively quench ^•OH, and the reaction rate constant is 6.0 × 10⁸ L mol^-1 s^-1 [45, 46]. BQ is commonly used as a scavenger of O₂^•−, (9.6 × 10⁸ L mol^-1 s^-1), and L-histidine can mask ¹O₂ and ^•OH [47-49]. In Fig. 4a, the addition of TBA significantly inhibited the removal of MO, and the decolorization rate decreased from 93% to 55% as the TBA increased from 0 to 60 mmol/L. However, when the concentration of TBA was enhanced to 60 mmol/L, the suppression effect was obviously weakened. It could be concluded that excessive TBA could not completely confine the MO decolorization, and other active substances also were responsible for the elimination of MO. From Fig. 4b, while the amount of BQ was augmented from 0.1 mmol/L to 0.5 mmol/L, the MO decolorization was declined from 87% to 59%, showing that O₂^•− was also played a non-ignorable role for the dye decontamination. Fig. 4c displays that the scavenging effect was augmented with the increase of L-histidine. When the concentration of L-histidine was enhanced to 4 mmol/L, the MO decolorization declined to 25%, demonstrating that ¹O₂ was also an essential reactive species in this Fenton-like system. In addition, the ESR analysis was employed to further confirm the existences of these active species. The typical signals for ^•OH, O₂^•− and ¹O₂ were detected in the spectrum of Figs. 4d-f, respectively. Obviously, these peak intensities for the Fe³⁺/AA/CaO₂ was higher than those of Fe³⁺/CaO₂ and Fe³⁺/AA, corroborating that the generations of these reactive species in the collaborative system. More importantly, the ESR results were in accordance with the scavenging tests.

Moreover, to explore the role of Fe⁴⁺ in this oxidative process, the Fe⁴⁺ amounts were measured by DMSO oxidation method. From Fig. S2 (Supporting information), the difference for the consumptions of DMSO was not significant in the Fe³⁺/CaO₂ and Fe³⁺/AA/CaO₂ systems. Furthermore, the formations of dimethyl sulfone (DMSO₂) were quite small and comparative, indicating that the Fe⁴⁺ productions were tiny during the two processes. Besides, the little decrease of the DMSO could be attributed to the attack of ^•OH radical [33]. These results proved that the Fe⁴⁺ was not the dominant active species in the Fe³⁺/AA/CaO₂.

In this work, a Fenton-like system using CaO₂ as the oxidant coupled with AA chelated Fe³⁺ was proposed to effectively remove various organic dyes in water. The Fe³⁺/AA/CaO₂ process could significantly improve the MO decolorization compared with other control systems. By the RSM optimization, the mutual effect of operation conditions (concentrations of Fe³⁺, AA and CaO₂) was explored, and the optimal parameters were obtained (2.76 mmol/L Fe³⁺, 0.68 mmol/L AA and 4 mmol/L CaO₂). Using the optimized parameters, the elimination of MO could reach 98.90% in the proof experiment, which was agreed well with the simulated result (99.30%). The complexation process could perform well in a wide range of pH (3-11), improving the H₂O₂ decomposition, and promoting the dissolution and cycle of Fe³⁺/Fe²⁺. The quenching test demonstrated that ^•OH, O₂^•− and ¹O₂ were all responsible for the dye removal, and the ESR analysis certified the existence of these reactive species in the collaborative system. This work was fundamental to provide an improved and optimized method for Fenton-like technique.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgment

The authors thank the financial support from the Natural Science Foundation of China (No. 51908485), the Natural Science Foundation of Hebei province (Nos. E2020203185, B2020203033, B2018203331), and the University Science and Technology Program Project of Hebei Provincial Department of Education (No. QN2020143).

Supplementary materials

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2021.04.050.