Optimizing the removal of methylene blue from aqueous solutions using persulfate activated with nanoscale zero valent iron (nZVI) supported by reduced expanded graphene oxide (rEGO)

Background: To remove methylene blue (MB) from aqueous solutions, nanoscale zero-valent iron (nZVI) predicated on reduced expanded graphene oxide (rEGO) was used as the activator of persulfate. Methods: Scanning electron microscope (SEM) and energy dispersive spectroscopy (EDS) analyses were used to investigate the surface morphology and to examine the surface elemental composition. X-ray diffraction (XRD) was used to determine the chemical compositions of the synthesized compound. In this study, the effects of pH (3-9), activator dose (0.4-1.6 g L-1), persulfate concentration (0.192-0.768 g L-1), and reaction time (0-60 minutes) on the removal of 10 mg L-1 MB were studied by nZVI -reduced expanded graphene oxide/persulfate (nZVI@rEGO/PS)


Introduction
The colored wastewater produced by industrial activities, such as textile industry and color production, has toxic effects on aquatic ecosystems (1). The presence of aromatic rings in the structure of azo dyes has increased the toxicity of these compounds and decreased their biodegradability (2). Dyes are among the most dangerous chemical compounds that can interfere with the process of photosynthesis in water resources. Methylene blue (MB) with molecular formula of C 16 H 18 ClN 3 S and with molar mass of 319.85 g/mol is one of the azo-cationic dyes (3,4). Its chemical structure is shown in Figure 1. It is also used in various industries, such as textile, paper paints, hair color, etc. Intense exposure to MB leads to severe damages to the human body (2,5). MB belongs to azo dyes family which contain the functional group R-N=N-R′ in their structure and constitute a very large group of all coloring agents used. Azo dyes are a triggering factor for histamine, which may exacerbate the symptoms of asthma, and cause uterine contractions in pregnant women, resulting in miscarriage. Additionally, in combination with benzoates, they can cause hyperactivity in children. Even for nonallergic persons, they may be the cause of, for example, urticaria (6). Therefore, due to environmental problems and human health, treatment of wastewater containing these compounds has become a vital issue (5).
Graphene has a two-dimensional structure and a suitable surface area that make it a pre-eminent nanoparticle carrier (28,29). The reduction of graphene oxide (GO) improves the performance of graphene when used as a nanoparticle carrier (30). For example, reduced graphene oxide-supported nanoparticles contributed to preventing the aggregation (31). Furthermore, expanded graphene oxide (EGO) with high quality and large surface area is suitable for environmental applications (32). Composites of nZVI with reduced graphene oxide (rGO) achieved the desired results in the removal of contaminants, such as Pb(II) (33), As(III), As(V) (34), and Cd(II) (35). Ahmad et al reported nZVI-rGO as an excellent and efficient persulfate activator than unsupported nZVI for degradation of trichloroethylene (36). EGO is a graphite derivative by chemical expansion of natural flake graphite. EGO has lower mobility in the porous media compared to GO. This feature makes it an environmentally friendly material. Recently, Dong et al have suggested a chemical method for the expansion of graphite as a pretreatment step for the synthesis of EGO. The expansion of graphite layers results in the fast diffusion of oxidant between layers even at low concentrations; as a result, the oxidation process is performed at room temperature. Furthermore, EGO with high quality and high surface area is suitable for environmental applications (32).
In this study, EGO was prepared by the chemically expanded graphite. Reduced EGO (rEGO) was used as a bed for the immobilization of nZVI. The nZVI@ rEGO was used as the activator of PS to remove MB from aqueous solutions. Moreover, the effects of essential factors affecting the performance of PS/nZVI@rEGO process, such as pH, nZVI@rEGO dose, PS dosage, contact time, and temperature were evaluated.

Synthesis of rEGO-supported nZVI
At first, expanded graphite (EG) was papered by mixing 1g of NFG, 8.5 g of CrO 3 , and 70 mL of HCl for 2 hours. In order to remove the remaining CrO 3 , the prepared EG was soaked in distilled water 4 consecutive times. To complete the chemical expansion reactions, the resulted EG was submerged in 40 mL of H 2 O 2 for 20 hours. Then, the EG was washed with distilled water and ethanol 4 consecutive times (32). The previous stage can be explained through the following equations: The presence of CrO 3 and HCl together leads to the production of CrO 2 Cl 2 between graphite layers (Eq. 7), the penetration stage. Adding H 2 O 2 to the penetrated EG leads to the hydrolysis of CrO 2 Cl 2 to CrO 5 at acidic condition (Eqs. 8 and 9). The generated CrO 5 interacts with H 2 O 2 and oxygen gas is released between graphite layers which leads to its expansion (Eq. 10), the expansion stage (37,38). As shown in Figure 2a, it almost doubled in size to 25 times.
Before starting the oxidation process, the EG was soaked in H 2 SO 4 to eliminate the remained internal H 2 O between graphite layers for 10 minutes. Then, 2 g of KMnO 4 was mixed with 20 mL of H 2 SO 4 in an ice bath for 30 minutes. After that, the mixture was removed from 25 ical structure of methylene blue. ice bath and it was leaved without mechanical mixing for 4 hours. Afterwards, the EG was added to the mixture and it was leaved for 2 hours to generate EGO. To complete the oxidation process, the produced EGO was soaked in diluted H 2 O 2 solution and cooled in an ice bath. Then, the EGO was washed with distilled water and ethanol 4 consecutive times (32). Figure 2b shows the mechanism of oxidation process.
To predicate nZVI on EGO, Fan et al. method was used (33). For this purpose, 1 g of EGO was scattered in distilled water for 2 hours and 5 g of FeSO 4 . 7H 2 O was mixed in 50 mL of distilled water, then, it was added to the suspended EGO where mixing was continued for 12 hours. The reducing solution was prepared by dissolving 1.4 g of NaBH 4 as a reducing agent in 50 mL of distilled water. Then, it was slowly added to the previous mixture where Fe +2 was converted to nZVI according to Eq. (11) (39)(40)(41). Figure 2c shows the mechanism of nZVI@rEGO formation.
Characteristics and analytical methods VEGA3/TESCAN-Libusina trida device was employed to provide the scanning electron microscope (SEM) and energy dispersive spectroscopy (EDS) analyses, which show the surface morphology and surface elemental composition of compounds. X-ray diffraction (XRD) patterns of samples were recorded on an X-ray diffractometer (X'Pert PRO MPD, PANalytical Company) with a CuKa radiation source at 40 kV and 40 mA.
The MB concentration was measured by measuring the absorption of light at a maximum wavelength of 664 nm. For this purpose, first, different concentrations of dye were prepared. Then, their absorption was measured at a maximum wavelength of 664 nm using a DR 5000 spectrophotometer manufactured by HACH. Then, the calibration curve was obtained by plotting the amount of adsorption in terms of concentration. Finally, by placing the adsorption of the samples in the resulting line equation, the dye concentration was calculated (42). Heidolph MR Hei-Standard magnetic stirrer was also used for mixing the solution. Solution pH values were adjusted by 0.1 M HCl and NaOH solutions. A Metrohm 691 pH meter was used to measure the pH of solutions. To quench the reaction, Soubh et al method was applied (43).
Batch oxidation experiments The batch method was employed to perform tests. Briefly, MB solution was placed in a glass flask, then, a certain amount of PS and nZVI@rEGO was added to the solution. Afterwards, the mixture was stirred at 80 rpm. Finally, the concentration of MB was determined at specified time intervals.
Where C i and C f are the initial and final MB concentrations, respectively. 3a and 3b show the XRD patterns of NFG and nZVI@rEGO, respectively. Figures 4a to 4b show the SEM images of NFG and nZVI@rEGO, respectively. In order to investigate the effect of pH on the MB removal efficiency, different values of pH (3, 5, 7, and 9) were investigated, the results are presented in Figure 5. The effect of different nZVI@rEGO doses (0.4-1.6 g L -1 ) with equal intervals of 0.4 g L -1 on the removal efficacy of MB was studied at MB concentration = 10 mg L -1 , PS concentration = 0.288 g L -1 , pH = 3, and reaction time = 25 min, the results are illustrated in Figure 6. The effect of different PS concentrations (0.192-0.768 g L -1 ) on the removal efficiency of MB was studied at MB concentration = 10 mg L -1 , nZVI@rEGO dose = 1.2 g L -1 , pH = 3, and reaction time = 25 minutes, and the results are illustrated in Figure  7. In order to determine the optimal time of the nZVI@ rEGO/PS process, the experiments were performed again under optimal conditions (MB concentration = 10 mg L -1 , nZVI@rEGO dose = 1.2 g L -1 , PS concentration = 0.576 g L -1 , pH = 3, and reaction time = 0-180 minutes), and the results are shown in Figure 8. The effect of temperature on the performance of nZVI@rEGO/PS and PS processes under optimal conditions (MB concentration = 10 mg L -1 , nZVI@rEGO dose = 1.2 g L -1 , PS concentration = 0.576 g L -1 , pH = 3, and reaction time = 20 minutes) was studied. For better understanding of the impact of temperature on the removal efficacy of MB by the nZVI@rEGO/PS process, the degradation rate constants were evaluated at temperature of 5, 25, and 50°C in the presence and absence of nZVI@rEGO.

Characterization of NFG, EGO, and nZVI@rEGO
As shown in Figure 3a, the XRD patterns of NFG show an intense peak at 2θ angle of 26°. While it decreased significantly in the XRD patterns of EGO and nZVI@ rEGO (Figure 3b), this is due to the processes of expansion and oxidation (32). The XRD patterns of nZVI@rEGO confirm the presence of ZVI at 2θ angles of 44.7° and 65°, which is consistent with the results of previous studies (33,44).
The SEM images for NFG, EGO, and ZVINFs/rULGO are shown in Figures 4a to 3C, respectively. As shown in Figure 4b, the EGO surface became more suitable to carry nanoparticles where the nZVI is placed on the EGO surface in the chain structure with a diameter about 100 nm ( Figure 4c). As shown in Figure 4f, EDS analyses of

Effect of pH
The solution pH has a significant effect on the quantity and type of radicals formed in sulfate radical advanced oxidation processes (SR-AOPs) (45). In the nZVI@ rEGO/PS process, pH has a significant effect on the decomposition of PS through its effect on dissolving nZVI (46). As shown in Figure 5, the removal efficiency of MB increased by decreasing the solution pH, and the maximum removal efficiency of MB was 79% at pH 3. That is because, at acidic conditions, nZVI is decomposed to Fe +2 according to Eq. (1). This led to an increase in activting PS according to Eq. (6). By increasing pH, the removal efficacy of MB was decreased to a minimum value at pH 7. This is due to the formation of iron hydroxide and the reduced presence of iron ions in the solution, according to Eqs. (13)(14)(15)(16) (47). By reaching pH to 9 and 11, the removal efficacy of MB increased slightly because alkaline conditions contribute to the activation of PS (48,49 Maximum removal efficiency of anthraquinone dye reactive blue 19 (RB19) using PS/ZVI system has been reported at acidic conditions, i. e. pH 3 (50). In another study, the maximum degradation of nonylphenol by PS/ nZVI-biochar was also reported at pH 3 (24). According to the results obtained from the experiments in this section, pH 3 can be considered as an optimal pH for the nZVI@rEGO/PS system.

Effect of nZVI@rEGO and PS dose
As illustrated in Figure 6, the increase of nZVI@rEGO dosage from 0.4 to 1.2 g L -1 increased the removal efficacy of MB from 70% to 87%. This can be due to the increased activation of PS, according to Eqs. (6), and as a result, increasing the dose of dissolved Fe 2+ ions, according to Eqs.      (1-4). Subsequently, the increase over that dosage did not cause a significant improvement in the removal efficacy of MB. This is because the increasing of concentrations of dissolved Fe 2+ ions can quench SR, according to Eq. (16) (51,52).

• 3 4 4
Fe SO SO Fe In other study, the increasing of ZVI dosage over 0.8 g L -1 did not significantly improve the decolorization rate of anthraquinone dye reactive blue 19 by the combination of PS and ZVI. Therefore, the nZVI@rEGO dosage of 1.2 g L -1 was chosen as the optimal dose for subsequent experiments.
In SR-AOPs, PS dose is one of the effective factors in removing the target contaminant (8). As illustrated in Figure 7, by increasing the PS dose from 0.192 to 0.576 g L -1 , the removal efficiency of MB increased from 74 to 96%. By increasing the PS dose higher than 0.576 g L -1 , there was no improvement in the removal efficacy of MB. That is because the increase of PS above a certain limit can quench SR, according to Eq. (17) (52). Also, sulfate radicals can react with each other, as a result, they have a quenching effect on the sulfate radical production at high concentration, according to Eq. (18) (15). Le et al have reported that an increase in the initial PS concentration above 20 mM had a negative effect on the removal rate of anthraquinone dye reactive blue 19 in the nZVI/PS system (50). Therefore, the PS dose of 0.576 g L -1 was selected to perform the following experiments.

Effect of reaction time and temperature
As can be seen in Figure 8, at reaction time of 20 minutes, the removal efficiency of MB was 96%, after that the removal efficiency of MB remained almost constant. This can be explained by the fact that the role of divalent iron ions is much greater than trivalent iron in the activation of PS. In the first minutes of the experiment, due to the high concentration of divalent iron ions, the reaction rate was high, which over time and decreasing the concentration of this ion and its conversion to trivalent iron ions (Eq. 16) (36), the process of PS activation also decreased sharply, and as a result, leads to its poor performance. In a study by Hung et al, iron oxide was used as a PS activator to remove MB dye, and similar results were reported regarding the removal efficiency after the optimal time (1). Thus, the reaction time of 20 minutes was selected as the optimum reaction time. At optimum conditions, the removal efficiency of MB was 37% in the absence of activator. Because PS is hydrolyzed alone in the absence of activator, according to Eq. (19) (53).
In order to investigate the removal mechanism, some experiments were conducted using PS, EGO, nZVI, and nZVI@rEGO/PS (Figure 9). For this purpose, the effect of PS and nZVI@rEGO on the removal of MB was tested separately, and the related removal efficiencies were significantly lower than that of the nZVI@rEGO/PS system. These results confirmed that the sulfate radical is the main component involving in the removal of MB. In addition, the low removal efficiency obtained using EGO is an indication of its low adsorption efficiency. As shown in Figure 9, only 18.7% removal of MB was observed in the ZVI process. The removal of MB occurred due to the ZVI surface adsorption and reduction.
As shown in Figure 10, by increasing temperature from 25 to 50°C, the removal efficiency of MB was increased from 35% to 51% by the PS process and from 96% to 98% by the nZVI@rEGO/PS process. As heat is considered as an activator of PS that increases removal efficiency, therefore, increasing temperature improved the removal efficiency (54). At temperature of 5°C, the removal efficiency of MB by the PS process decreased to 9%. While, there was no significant change in the removal efficiency of MB by the nZVI@rEGO/PS process. This feature makes nZVI@ rEGO/PS process suitable to work at low temperatures.

Removal of MB from real dye wastewater
The impact of nZVI@rEGO/PS process on dyecontaining wastewater was examined. For this purpose, a series of experiments were performed on the diluted real dye wastewater samples with MB (10 mg L -1 ) under optimum conditions. Dye wastewater was obtained from pump station of a tannery wastewater treatment plant. As concentration of MB in the supplied wastewater was high, for providing comparable concentration of MB (10 mg/L), supplied wastewater was diluted by distilled water. The characteristics of diluted wastewater were: COD (98 mg L -1 ) and Cl -(440 mg L -1 ). As shown in Figure 8, the removal efficiency of MB from real dye wastewater was 88%, while the removal efficiency of MB from synthetic wastewater was 96%. This decrease in removal efficiency can be explained by the fact that the real wastewater mainly contains a variety of organic and inorganic contaminants which could decrease contacts with MB (55,56), as well as chloride ion which can scavenge oxidant radicals, and as a result, leads to a decrease in the removal efficiency (57,58).

Identification of predominant radicals in nZVI@rEGO/ PS process
To identify the dominant radical species in nZVI@rEGO/ PS process, alcohols were used to quench the hydroxyl and sulfate radicals. Therefore, ethanol (alcohol with alpha hydrogen) as hydroxyl and sulfate radical scavenger and tert-butyl alcohol (TBA) with no alpha hydrogen were used as effective quenching agents for hydroxyl radicals (59). In this study, alcohol (EtOH, TBA) concentration used was 1 M. The results clearly showed that aniline degradation was significantly decreased by adding alcohols. As shown in Figure 11, aniline degradation in nZVI@rEGO/PS process without any radical scavenger was 96%. However, in the presence of ethanol (EtOH), only 18% MB removal was observed. While in the presence of TBA, the MB removal was 63%. The results revealed that the reaction was completely quenched by adding EtOH as • OH radical and SO 4 •− radical scavenger, while with the addition of TBA, the removal rate of MB was moderately influenced.

Kinetic modeling
Kinetics study indicated that nZVI@rEGO/PS decolorization is a pseudo-first-order reaction, which can be expressed according to Eq. (20) (50). As illustrated in Figures 10a and 10b, in the absence of nZVI@rEGO, the degradation rate constants of MB at temperatures of 5, 25, and 50°C were 2.5 × 10 -3 , 19.5 × 10 -3 , and 29.1 × 10 -3 min -1 , respectively (Figure 12a). However, in the presence of nZVI@rEGO, the degradation rate constants of MB at temperatures of 5, 25, and 50°C were 96.4 × 10 -3 , 135 × 10 -3 , and 150.3 × 10 -3 min -1 , respectively (Figure 12b). In other words, the presence of nZVI@rEGO contributed to strengthening of the degradation rate constants, at above-mentioned temperatures. It has been reported that increasing ferrous concentration from 1 to 4 mM resulted in an increase in the degradation rate constant of Orange G from 0.04 to 0.12 min -1 in ferrous/persulfate system (60).
The Arrhenius equation (Eq. 21) was used to evaluate the activation energy of reaction at above-mentioned temperatures (61).
Where k is the pseudo first-order rate constant (min -1 ), A is the pre-exponential factor (min -1 ), E a is the activation energy (kJ mol -1 ), R is the universal gas constant (0.0083 kJ mol -1 ), and T is the solution temperature (K). It could be observed that there is a strong linear relationship between lnk and 1/T (Figure 13), the presence of nZVI@rEGO decreased the activation energies of MB removal from 45.9 to 8.32 kJ mol -l (Table 1). In other words, it caused 5.5 times decrease in E a of MB removal. This decline in E a is attributed to the activation provided by the nZVI@ rEGO of PS. In other words, generation of SR at this level need less E a , and as a result, the existence of nZVI@rEGO with PS together in the solution (62). In other studies, the activation energy was decreased from 120.4 to 36.1 kJ mol -1 using Fe 0 as an activator of PS for degradation of acetaminophen (15), and the activation energy was 98 kJ mol -1 using ZVI as an activator of PS for decolorization of anthraquinone dye reactive blue 19 (50).

Conclusion
This study investigated the removal efficiency of MB in aqueous solutions by the nZVI@rEGO/PS process. The