Introduction

During recent decades, advanced oxidation processes (AOPs) have proven a promising effect at organic pollutant degradation from wastewaters. The basis of AOPs is the in situ production of free hydroxyl radical (•OH) in aqueous solution via chemical, photochemical, electrochemical, or photocatalytic methods [1, 2]. Hydroxyl is a strong oxidant that is able to mineralize most of the organic pollutants unselectively through hydrogen abstraction, electron transfer reaction, and electrophilic addition on double bond, finally converting them into CO2, H2O, and inorganic ions [3, 4]. The AOP method consists of different methods including ozonation [5], UV irradiation [6], ultrasonication [7], Fenton [8], and electro-Fenton (EF) [9].

Fenton reaction, which is the aqueous combination of Fe2+ and H2O2, is considered a promising environmental remediation technology. This method is simple, does not need any special equipment, highly efficient in organic pollutant removal, and is an environmentally safe method. Although this method has high efficiency at the oxidation of organic pollutants, the application of this method at a large scale has been restricted due to the shipment and storage problems of H2O2 and the generation of iron sludge. Newly, to overcome the mentioned problems, the EF process has been developed. This process includes the ferrous ion addition and in situ electro-generation of H2O2 [10,11,12]. H2O2 can be produced by a two-electron reduction of oxygen at the appropriate cathodic potential on specific electrodes such as graphite [13], activated carbon fiber [14], carbon felt [15], graphene [16], C60-carbon nanotube composite [17], and gas diffusion electrode [18,19,20]. The reactions occurred can be expressed as follows (Eq. 1 and 2):

$${O}_{2}+ {2H}^{+}+ {2e}^{-} \to {H}_{2}{O}_{2}$$
(1)
$${Fe}^{2+} + {H}_{2}{O}_{2} \to {Fe}^{3+} + {^{\bullet}}OH+ {OH}^{-}$$
(2)

The advantages of the EF process include the following: (1) continuous in situ generation of H2O2 (Eq. 1), which eliminates shipment and storage costs of H2O2; (2) the reaction occurs easily at ambient temperature and pressure; (3) Fe2+ can be electrochemically regenerated at the cathode; therefore, the amount of iron sludge will be reduced (Eq. 3); and (4) air or oxygen flow enhances the mixing rate of EF reaction [21,22,23,24,25].

$${Fe}^{3+}+ {e}^{-} \to {Fe}^{2+}$$
(3)

The concentration of H2O2 is an essential factor in the efficiency of pollutant degradation in the Fenton reaction. Hence, an efficient cathode system for significant production of H2O2 and pollutant degradation seems to be necessary [25,26,27].

The elimination of dyes from wastewater is one of primary environmental concerns. More than 10,000 dyes are probably consumed in different industries such as textiles, leather, rubber, plastics, paper, food, cosmetics, and mineral processing industries. The concentration of these pollutants varies at 10–10.000 mg/L range [28, 29]. Synthetic dyes represent a massive problem due to their resistivity in biological treatment. Dye is one of the most obnoxious pollutants due to the following: (1) it is observable and even small quantities (≥ 0.005 mg/L) are not allowed [30], (2) the color of the dye can interfere sunlight transmission into natural streams, and (3) many of dyes and their intermediate products are toxic to aquatic life, carcinogenic and mutagenic to humans [31].

Neutral red (NR) is a cationic dye. It is widely applied as a pH indicator in laboratories and for nuclear counterstaining in biological research. This dye is considered toxic and hazardous because when it is decomposed, it gives out hazardous products, such as carbon monoxide, nitrogen oxides, and hydrogen chloride. These products may cause serious health problems for humans and animals. Therefore, NR draws significant attention to the innovation of effective techniques for its removal [32].

In this study, the removal of NR from the aqueous environment by the EF process using a couple of graphite electrodes as cathode and anode and its effective parameters effect were investigated. To find optimum amounts of variables, response surface methodology (RSM) was applied via Minitab 16 software. Finally, a verification study of the analysis including ultraviolet–visible spectroscopy (UV–Vis), Fourier-transform infrared spectroscopy (FT-IR), X-ray diffraction (XRD), Brunauer–Emmett–Teller (BET), temperature programmed desorption (TPD), total organic carbon (TOC), and gas chromatography–mass spectrometry (GC-MS) was performed using the optimum operational conditions.

Materials and Methods

Chemicals

Sodium aluminate (NaAlO2, Al2O3, 55 wt%), silicic acid (SiO2·xH2O, > 99 wt%), iron nitrate (Fe(NO3)2·9H2O), tetrapropyl ammonium bromide (TPABr, C12H28BrN), ammonium nitrate (NH4NO3), sodium sulfate (Na2SO4), sulfuric acid (H2SO4), and sodium hydroxide (NaOH) purchased from Merck (Germany).

NR, cationic dye, was selected as the model pollutant and purchased from Boyakhsaz Co. (Iran). The chemical properties of the NR are presented in Table 1.

Table 1 Characteristics of NR
Full size table

Instruments and Characterizations

For measuring the pH of the solution, pH-meter (AZ 86502, Iran) was used. The stirring of samples was performed via a magnetic stirrer (Philler scientific, Iran). To generate the electricity current of the process, ADAK-PS405 power supply was applied. The decolorization of the NR was monitored by SPECORD 250 UV–Vis spectrophotometer (Analytik Jena, Germany). The absorbance of NR was studied at 519 nm.

For FT-IR measurements, Nexus Model Infrared Spectrophotometer (Nicolet Co, USA) was applied at 4 cm−1 resolution. XRD experimentations were performed by a D8 Advance Bruker AXS X-ray diffractometer with Ni-filtered Cu Kα radiation (λ = 0.15418 nm) at 2θ range of 4–50° at 40 kV.

To investigate the surface area of nanocatalysts, BET measurements were done. For this purpose, the degassing of powders was performed for 3 h at 300 °C. Then, the total surface area (SBET), total volume (Vtotal), micropore volume (Vmicro), and mesopore volume (Vmeso) of nanocatalysts were calculated. For calculating the mentioned parameters, the BET isothermal equation, nitrogen adsorbed volume at P/P0 = 0.99, t-plot method, and the difference of the total calculated data and the corresponding micropore data were applied.

Temperature programmed desorption of ammonia (NH3-TPD, Micromeritics, USA) with an on-line TCD detector was applied for studying the acidity of the samples. The textural properties of the samples were identified via N2 adsorption–desorption technique at − 196.2 °C (Quantachrome, USA).

For TOC measurements, a TOC analyzer (Shimadzu, Japan) was applied. An Agilent 6890 gas chromatography coupled with an Agilent 5973 mass spectrometer (Agilent Technologies, Palo Alto, CA) was used for GC-MS analysis. The temperature programming of this analysis was done according to our previous work [13].

Experimental Design

In this research, the central composite design (CCD) with five levels was applied to optimize the EF method. For investigating the influence of the operational parameters on the NR degradation efficiency, five factors, including pH (X1), applied current (X2), reaction time (X3), amount of catalyst (X4), and initial NR concentration (X5), were chosen. Thirty-one experiments were recommended by the software. The range of parameters and applied levels are presented in Table 2.

Table 2 Coded levels and the natural values of the independent test variables
Full size table

Electrochemical System

EF treatments were performed in a 100-mL glass beaker considered as undivided cell at room temperature under a magnetic stirrer using a couple of graphite sheets with dimensions of 1 cm × 3 cm × 0.3 cm as electrodes. For NR solution preparation, 50 mL of NR dye solution, iron(II) sulfate solution (0.1 M) and 5 mL of sodium sulfate solution (0.05 M) were added, respectively. The pH was adjusted via adding H2SO4 and NaOH to the solution. Then, the needed oxygen was provided via air injector to produce the required H2O2. The electrical current of the process was conducted with a power supply at the range of 0.1–0.5 A. For the heterogeneous EF method, the provided catalysts were used instead of iron(II) sulfate solution and the other chemicals were the same. The decolorization efficiency of NR was measured by Eq. 4, which A0 and At are the absorbance of initial NR solution and the absorbance NR after t minutes of reaction, respectively.

$$\mathrm{Decolorization efficiency} (\mathrm{\%})=\frac{{A}_{0}-{A}_{t}}{{A}_{0}}\times 100$$
(4)

Catalyst Synthesis

The zeolite nanocatalysts were prepared by the hydrothermal method and impregnation with the iron promoter (1 wt%) at one of the samples according to Rostamizadeh et al. [33]. In the current study, the crystallization of the nanocatalyst was performed for 48 h at a temperature of 180°C under autogenous pressure in an autoclave. The obtained powder was washed several times. Then, it was dried for 48 h at 105°C. Then, the calcination of the nanocatalyst was performed for 12 h at a temperature of 530°C. The heating rate of the furnace was 3°C/min with air flow. The powder was ion-exchanged four times by 1 M solution of NH4NO3 at 90°C for 10 h, following with drying at 105°C overnight and calcination at 540°C in air for 12 h (3°C/min). The parent nanocatalyst was denoted as HZC zeolite. The iron promoter was incorporated on the parent HZC through a multi-step impregnation in rotary evaporator including the following temperatures and pressures: (1) 65°C and 300 mmHg for 60 min, (2) 70°C and 250 mmHg for 30 min, (3) 70°C and 200 mmHg for 30 min, (4) 75°C and 200 mmHg for 30 min. Finally, the solution was dried at 110°C for 12 h to obtain Fe impregnated ZSM-5 nanocatalyst (FeZC).

Results and Discussions

EF Process Optimizing with RSM

For this part of the research, Minitab 16 software was applied and 31 tests were designed by the software. Experiments were performed according to Table 3. According to the experimental results, the coefficients of correlation and relationship was obtained between the response variable and independent variables according to Eq. 5:

Table 3 The 5-factor central composite design matrix and the value of response function (decolorization efficiency %)
Full size table
$$\begin{aligned}Y&=-53.74+{18.09X}_{1}+{0.0055X}_{2}+{0.9978X}_{3}+{35.87X}_{4}+{1.918X}_{5} \\ & \quad -{0.01616X}_{1}{X}_{2}+{0.0440X}_{1}{X}_{3}+{0.986X}_{1}{X}_{4}-{0.3259X}_{1}{X}_{5} \\ & \quad -{0.000693X}_{2}{X}_{3}+{0.01659X}_{2}{X}_{4}+{0.003122X}_{2}{X}_{5}-{0.0951X}_{3}{X}_{4} \\ & \quad -{0.00537X}_{3}{X}_{5}-{0.1827X}_{4}{X}_{5}-{1.767X}_{1}{^{2}}-{0.000150X}_{2}{^{2}} \\ & \quad -{0.004371X}_{3}{^{2}}-{5.738X}_{4}{^{2}}-{0.00627X}_{5}{^{2}} \end{aligned}$$
(5)

In this equation, Y is the response variable (decolorization efficiency) and Xi is the experimental levels of the variables. Table 3 shows the experimental and predicted results by the software based on the Eq. 5 in uncoded form. Based on the results, it can be concluded that the resulted model can accurately modeling the experimental results and can be used to predict the output variable under other conditions.

Analysis of variance (ANOVA) is commonly used to test the significance and suitability of the model. Table 4 shows the ANOVA results. The obtained correlation coefficient (R2 = 0.997) is acceptable (Fig. 1), indicating how much variability in the response variable can be explained by the experimental factors and their interactions. The obtained value indicates that approximately 99.7% of the variability in decolorization efficiency can be justified by the independent variables and the model is unable to explain only 0.3% of the variation.

Table 4 Analysis of variance (ANOVA) for fit of decolorization efficiency from central composite design
Full size table
Fig. 1
figure 1

Comparison of experimental and predicted decolorization efficiency results

Full size image

The variables and their interactions impact on the response value can be calculated via Pareto analysis. Each parameter’s influence is determined via Eq. 6 and the obtained coefficients of each parameter via Eq. 5.

$${P}_{i}=\left(\frac{{a}_{i}^{2}}{\sum {a}_{i}^{2}}\right)\times 100 i\ne 0$$
(6)

In this equation, Pi represents the effect of each parameter and the ai is related to each parameter’s coefficient. Figure 2 shows each parameter’s impact and their interactions according to Pareto analysis. According to the results, Fe (%) amount and pH have the most influence among parameters, respectively.

Fig. 2
figure 2

Pareto graphic analysis of parameters

Full size image

The obtained F value of the model (the ratio of the mean squares to the regression or model and the residuals) is greater than the critical value, indicating that the factors adequately represent the data changes from their mean value and the effects of the factors estimated in the model are true and accurate [34, 35]. If the obtained F value is compared with the critical F values reported in the statistical tables at degrees 30 and 32, it can be seen that the obtained F in this experiment is greater than the critical F value and the null hypothesis is rejected, which means the origin of the regression or model changes and the residuals are not the same [36].

Effects of Operational Parameters on the Decolorization Efficiency of NR

Influence of Applied Current and pH on Decolorization Efficiency

Figure 3a shows two-dimensional and three-dimensional graphs of the decolorization efficiency as a function of applied current and pH for a solution with an initial NR concentration of 30 mg/L, reaction time of 60 min and Fe ion amount of 3% (V/V). According to this, at pH = 3, by increasing the current intensity about 300 mA, the decolorization efficiency is slightly increased and after 400 mA, it is rapidly decreased.

Fig. 3
figure 3

The response surface plots and contour plots of the decolorization efficiency (DE. %) as the function of a) applied current (mA) and pH, b) pH and Fe ion amount (%), and c) initial NR concentration (mol/L) and EF time (min)

Full size image

Increasing the current intensity to the optimum amount leads to the formation of more H2O2 in the cathode, and thus the production of •OH. It also increases the hydrolysis of the water molecules and accelerates the EF process by accelerating electrical coagulation. On the other hand, increasing the current intensity above the optimum value results in competitive electrode reactions such as oxygen depletion at the anode and hydrogen formation at the cathode, which results in a decrease in the •OH concentration in the solution and decreases the decolorization efficiency [37, 38].

pH is one of the most critical parameters in the EF process. It directly affects the oxidation of organic compounds and •OH production [39]. Studies in this field show that the effect of the initial pH of the environment varies greatly depending on the type of contaminant in the different processes.

At very low pH, available or electronically produced H2O2 cannot be decomposed by the Fe2+ ions present in the reaction medium to •OH. Therefore, it absorbs a positive charge (proton) and converts it to an oxonium ion (Eq. 7), which is more stable than H2O2 and reducing its reaction with Fe2+, resulting in a decrease in •OH production and process efficiency [40].

$${H}_{2}{O}_{2} + {4H}^{+} \to {2H}_{3}{O}^{2+}$$
(7)

At high pH, H2O2 is unstable and may decompose spontaneously according to Eq. 8.

$${H}_{2}{O}_{2} \to {O}_{2}+ {H}_{2}$$
(8)

In this range, Fe2+ and Fe3+ are converted to Fe(OH)2 and Fe(OH)3 through the reaction with −OH, respectively. Fe(OH)3 exits the EF process cycle after formation but can promote electrical coagulation, while Fe(OH)2 may lead to the development of the EF process (Eq. 9) [37].

$${2H}_{2}{O}_{2}+ {Fe(OH)}_{2} \to {Fe}_{2}{O}_{3}+ {H}_{2}O+ 2{HO}$$
(9)

Influence of pH and Fe Ion Amount on Decolorization Efficiency

In Fig. 3b, the simultaneous effect of initial solution pH and Fe ion amount on the decolorization efficiency was investigated at initial NR concentration of 30 mg/L, applied current of 300 mA and 60 min of reaction time. These two-dimensional and three-dimensional curves show that by increasing until pH = 3, the decolorization efficiency increases, and after this point, it decreases. Fe2+ amount also plays an important role in the NR removal during the EF process. It was observed by increasing the Fe2+ amount until 3% the decolorization efficiency increases. By adding more amount of Fe ion and passing the optimum amount, the decolorization efficiency decreases.

Influence Initial NR Concentration and Reaction Time on Decolorization Efficiency

Figure 3c shows two-dimensional and three-dimensional graphs of decolorization efficiency as a function of initial NR concentration and reaction time. It can be seen by increasing the initial NR concentration to the optimum level; the decolorization efficiency gradually increases, whereas, at the same conditions with lower initial concentration, the decolorization efficiency quickly reaches higher efficiencies. It must be noted after passing the optimum initial NR concentration, the amount of hydroxyl radical will not be sufficient to remove the high contaminant concentration and the decolorization efficiency will be decreased [41]. At higher initial NR concentrations, the optimal time range for NR decolorization is more narrow than it is at lower concentrations.

Determination of Optimum Conditions for NR Decolorization

The optimal way to remove NR is to reach maximum efficiency. Therefore, using the Minitab 16 software, optimal test conditions were determined as stated in Table 5. The results showed a maximum of 85.88% NR removal at the applied current of 200 mA, the amount of Fe2+ 2%, the initial NR concentration of 40 mg/L, and pH = 3 during 40 min of reaction time. After obtaining the optimum conditions, an experiment was performed at optimal conditions and the results showed that there is a fairly good agreement between the experimental value (85.88%) and the value predicted by the model (85.77%). This demonstrates the success of using the RSM in modeling and optimizing the homogeneous EF process.

Table 5 Optimum operating parameters for maximum removal efficiency
Full size table

Nanocatalyst Characterization

FT-IR spectra of the synthesized nanocatalysts were documented in the range of 400–4000 cm−1. The bands nearby 450 cm−1 assigned to the internal SiO4 and AlO4 tetrahedral. The band about the wavelength of 550 cm−1 attributed to the zeolite with five-membered rings at HZC and FeZC (Fig. 4a). Surface hydroxyl groups band observed in the 3500–3800 cm−1 range (Fig. 4b). Recognized FT-IR bands for the vibration of bridging Si–OH-Al were at 3610 cm−1 and 3680 cm−1 range [42]. The band at 3680 cm−1 belongs to extra-framework aluminum species (Al–OH) of the nanocatalysts [43, 44].

Fig. 4
figure 4

FTIR spectra of nanocatalysts in the range of a) 400–4000 cm−1 and b) 3500–3800 cm−1

Full size image

Figure 5 shows the XRD patterns of the synthesized HZC and FeZC nanocatalysts. The lower peak intensity of HZC comparing to FeZC at 2θ of 20–30° claims amorphous SiO2 was not entirely formed and crystallized in the HZC structure. According to Table 6, the relative crystallinity of HZC and FeZC nanocatalysts were high. Minor defects at the catalyst framework, which led to the amorphous peak at 2θ of 25.5–27° were observed. The reason for this phenomenon was dealumination over the impregnation process at the nanocatalysts, which results in a shift in the peak positions and slight framework damage [45]. The crystallinity reduction of the impregnated nanocatalysts was in accordance with the impurity formation. The lack of further peaks, which are related to the promoter, established the uniform promoter dispersion.

Fig. 5
figure 5

The XRD patterns of nanocatalysts

Full size image
Table 6 Relative crystallinity and textural data of the nanocatalysts
Full size table

It was observed N2 adsorption–desorption of HZC and FeZC nanocatalysts was a mixture of types I and IV isotherm according to Fig. 6a. The rectangular hysteresis loops at P/P0 = 0.5–0.95 indicated the HZC and FeZC nanocatalysts structure are mesoporous, resulting from the capillary condensation [46]. Also, the crystal agglomeration was created an interparticle space at the mesoporous structure of nanocatalysts. The microporous structures of the HZC and FeZC nanocatalysts were proved by their high adsorption amount at P/P0 = 0.1 [47]. Also, by analyzing the pore size distribution of HZC and FeZC, the mesoporous structure of nanocatalysts was confirmed (Fig. 6b). The primary pore diameter of the nanocatalysts was 1.70 nm. The calculated data in Table 6 also exhibited the high surface area and formation of mesopore nanocatalysts. The results obtained for BET were in accordance with the XRD results. The FeZC nanocatalyst represented higher total pore volume owing to the formation of mesopores during the impregnation process.

Fig. 6
figure 6

a) N2 adsorption–desorption isotherm and b) pore size distribution of nanocatalysts

Full size image

The NH3-TPD measurements exhibited almost the same patterns for the HZC and FeZC with different strength and quantity of the acid sites (Table 7, Fig. 7a and b). The strong and weak acid sites of the HZC and FeZC nanocatalysts were characterized via a couple of desorption peaks. The temperature ranges of these peaks were 130–280°C and 300–500°C, respectively. The acid sites concentration was estimated via catalysts peak area. The HZC almost involved the same quantity of strong and weak acid sites (0.53 mmol NH3/g). The impregnation decreased nanocatalyst acidity due to Fe species involvement through the zeolite structure. This can be described by blocking pores and neutralizing surface acid sites with promoters [48]. The smaller effective ionic radii of Fe ion (about 0.77 Å [49]) made it possible to enter inside the channels of ZSM-5 (diameter of 5.5 Å) and interacted with interior acid sites.

Table 7 Acidity and strong/weak ratio
Full size table
Table 8 Identified by-products during electrolysis of NR
Full size table
Fig. 7
figure 7

Acidity of the nanocatalysts

Full size image

TOC and GC-Mass Analysis

TOC analysis was also performed for the optimal sample to measure the degradation efficiency of the method. The TOC amount of NR over time is given in Fig. 8. The TOC amount of NR decreased over 2 h of reaction time, which indicates that the results of decolorization and degradation efficiency of NR are in good agreement and the mineralization efficiency reached 85.98%. This shows that NR has been successfully mineralized. The difference between obtained NR decolorization efficiency (92.44%) and degradation efficiency (85.98%) for heterogeneous EF can be attributed to the decomposition of organic materials and the formation of stable short-chain organic intermediates [13, 17].

Fig. 8
figure 8

TOC changes of NR solution over time

Full size image

GC-MS analysis was performed after preparing the optimal sample. The mass spectra obtained from GC-MS analysis were matched with the spectral information of the MS library (Wiley 7n) and 18 compounds were successfully identified according to Table 8 including the following: (1) acetamide, (2) propane-1,2-diol (3) ethanimidic acid, (4) (S)-2-amino-3-methylbutanoic acid, (5) ethane-1,2-diol, (6) 2,3-butanediol monoacetate, (7) (S)-2-aminopentanoic acid, (8) 2-methyl-1,3-propanediol, (9) 2-methylmalonic acid, (10) propanoic acid, (11) acetic acid, (12) butanoic acid, (13) 1,3-butanediol, (14) methyl 3-hydroxypentanoate, (15) 3-hydroxy-2-methylbutanoic acid, (16) octanoic acid, (17) 4-oxopentanoic acid, and (18) 2-butenedioic acid. It was possible due to the breaking of C–N and C–C bonds at NR in earlier stages, compounds with OH and COOH functional groups were formed. These intermediates can be altered to the short-chain carboxylic acids by further oxidation and finally to CO2 and H2O. It must be noted that, alongside the identified compounds, many other chromatographic peaks were also observed. But since their match factor of mass spectrums was very low, it was not possible to recognize them favorably.

Stability of the Electrodes

For examining the stability of the electrodes, repeated-batch operations were performed. According to Fig. 9, during ten repeated runs, the electrodes showed slight decrease of decolorization efficiency from the first run to the last one. The decolorization decrease amount was 4.46% (92.44–87.98%) after ten runs. Therefore, the results indicated that the electrodes had a good stability and possessed reasonable reusability in repetitive degradation operations.

Fig. 9
figure 9

DE (%) of graphite electrode during ten batch operations

Full size image

Conclusions

In the present work, the EF process for the decolorization of NR from contaminated water was investigated via a couple of graphite electrodes. Thirty-one experiments were done via RSM. The results showed that in the applied current of 200 mA, the amount of Fe2+ 2%, the initial NR concentration of 40 mg/L and pH = 3 during 40 min of reaction time, the maximum decolorization efficiency value was 85.88% at homogenous EF, 90.32% and 92.44% at heterogeneous EF using HZC and FeZC nanocatalyst, respectively. The results of the experimental part were compared by predictions using Minitab 16 software. The correlation coefficient found between the experimental and the model results was 99.7%. FT-IR, XRD, BET, and TPD analyses were performed to confirm the successful synthesis of HZC and FeZC catalysts. UV–Vis, TOC, and GC-MS analysis were performed to analyze the successful degradation of the NR. The resulting compound was analyzed by GC-MS, and 18 compounds were successfully identified. The TOC results showed about 85.98% of NR mineralization efficiency. Due to the low cost and high decolorization efficiency of this method, it can be used for the removal of organic pollutants such as dyes.