In situ electrosynthesis of hydrogen peroxide with an improved gas diffusion cathode by rolling carbon black and PTFE

Abstract

A simply structured gas diffusion electrode (GDE) was constructed by rolling carbon black and PTFE as a conductive catalyst layer to enhance the producibility of hydrogen peroxide. A Box–Behnken design (BBD) coupled with response surface methodology was employed to assess the individual and interactive effects of the three main independent parameters (pH, current density and air flow rate) on the H₂O₂ concentration. Analysis of variance (ANOVA) showed a high coefficient of determination value. Optimal operating conditions were a pH value of 4.0, current density of 52 mA cm⁻², and an air flow rate of 55 mL min⁻¹. The predicted H₂O₂ concentration under the optimal conditions determined by the proposed model was 309.85 mM, demonstrating the improved GDE without using noble metals and other chemical promoters is a potential method for in situ electrosynthesis of H₂O₂. Results also revealed that the current density, air flow rate, and their interaction effect had a significant effect on the H₂O₂ concentration, whereas changes to the initial pH had no apparent effect. Experiments showed that current density has a direct effect on the decomposition reaction in the electrolytic process.

---

1. Introduction

Hydrogen peroxide (H₂O₂), as a versatile oxidizing agent, has been widely used in many industrial areas, particularly in the chemical industry and environmental domain. The only degradation product of its use is water, thus it has played an important role in environmentally friendly methods in the chemical industry.^1,2 Currently, H₂O₂ is produced on an industrial scale using the anthraquinone oxidation (AO) process. However, it can hardly be considered a green method, since the process involves the sequential hydrogenation and oxidation of an alkylanthraquinone precursor dissolved in a mixture of organic solvents followed by liquid–liquid extraction to recover H₂O₂.¹ The transport, storage, and handling of bulk H₂O₂ involve hazards and escalating expenses.³ Thus, novel, cleaner methods for the in situ production of H₂O₂ are being explored.

Various procedures are available for in situ synthesis of H₂O₂, including direct synthesis of H₂O₂ from O₂ and H₂ catalyzed by a variety of catalysts⁴ or activated by dielectric barrier discharge,⁵ and the application of microbial fuel cells.⁶ However, these methods are hindered by several limitations, such as the necessity to remove the employed catalyst and the high cost of catalytic materials. On the contrary, electrosynthesis methods present various advantages, e.g., the use of catalysts immobilized in the electrode structure, and the inexpensive carbonaceous materials with high catalytic performances for H₂O₂ production.⁷ Carbon materials make excellent cathodes for the two-electron reduction of oxygen to H₂O₂ and are the prime choice for an electrocatalyst support because of their large specific surface area, good thermal and chemical stability, and low price.⁸

Of all the electrode structures, the gas diffusion electrode (GDE) has attracted great attention owing to its relatively high H₂O₂ production.^9–12 The GDE is composed of an active layer and a gas diffusion layer, which allows an unlimited supply of gaseous reagents to pass through the porous structure to the electrode/electrolyte interface, thus preventing mass transport limitation of the reaction of interest.^7,13 In the electrolytic process, the catalytic layer faces the electrolyte while the gas diffusion layer faces the reactant gas which diffuses through the micro-pores of the GDE to the catalytic layer and reacts with the electrolyte at the interface between the electrolyte and the reactant gas.^9,14 Depending on the different cathode materials and electrode components, the H₂O₂ yield and current efficiency differ greatly. But for most of these cathodes the H₂O₂ yield and current efficiency are insufficient for their application in in situ electrosynthesis. Therefore, novel cathodes, with high efficiencies and low costs, must be developed. So far, GDEs used as cathodes for in situ electrosynthesis of H₂O₂ have mainly been focused on various carbon-based materials (e.g., graphite, carbon nanotubes and carbon black)^13,15–17 and their modification (e.g., modification by 2-ethylanthraquinone and azobenzene),^13,18 and their application to the degradation of different organic pollutants.^16,19–22 However, the effect of the construction of the cathode is usually neglected, which is absolutely vital for process effectiveness.²³ In the present work, an improved GDE prepared by rolling carbon black and polytetrafluoroethylene was introduced to the electrolytic process. Compared with the conventionally used GDE system, the improved GDE was proven to greatly enhance H₂O₂ productivity and accumulation.

In addition to the cathode properties, the operating conditions have a significant impact on H₂O₂ productivity and current efficiency. It is well documented in the literature that the efficiency of the electrolytic process depends on the pH, current density, air flow rate, supporting electrolyte and electrolytic time.^2,7,24,25 With typical multifactor experiments, different operating conditions should therefore be employed to achieve a higher H₂O₂ concentration and current efficiency. While in typical multifactor experiments, optimal conditions of these variables are usually found by varying a single factor while keeping the other variables constant, the methodology does not include possible interaction effects between variables and could lead to restricted conclusions.^26,27 Response surface methodology (RSM) is a widely accepted statistical-based method for designing experiments, evaluating the individual and interaction effects of independent variables, and optimizing the process parameters within a limited number of experiments.²⁸ For example, the oxygen reduction reaction (O₂ + 2H⁺ + 2e⁻ → H₂O₂), is influenced by the applied current, pH value and gas flow rate simultaneously. The operating process optimization using RSM is faster for collecting experimental results than the rather conventional, time consuming one-factor-at-a-time approach.²⁹

In this work, an improved GDE was proposed to enhance the electrochemical performance during the electrolytic process. RSM based on a Box–Behnken design (BBD) was employed to design and optimize the individual and interactive effects of the three main independent parameters (initial pH, current density and air flow rate) on H₂O₂ accumulation. The significance of each variable on the H₂O₂ concentration was determined and the optimal operating conditions were obtained and validated.

2. Experimental section

2.1. Materials

The carbon black powder (CB), Vulcan XC 72R, was purchased from Cabot Corporation and used without any treatment. The particle size distribution and nitrogen adsorption isotherm of the samples are shown in Fig. S1.† Polytetrafluoroethylene (PTFE, 60 wt%, Hesen, Shanghai, China) was used as a binder. Nafion-117 (Dupont, New York, NY, USA) was used as the cation-exchange membrane.

2.2. Preparation procedure of the gas diffusion cathode

The improved gas diffusion cathode (IGDE) consisted of a conductive catalytic layer (CCL) and a titanium mesh. Different from traditional GDEs comprising a gas diffusion layer and catalyst layer,^22,30 the CCL simultaneously acts as both the gas diffusion layer and catalyst layer. After a hydrophobic treatment with PTFE, the titanium mesh (40 meshes) was used as the matrix.

The IGDE fabrication procedure is presented in Fig. 1. The CCL was prepared firstly by distributing the CB powder of 2.0 g into an appropriate amount of dispersant (ethanol) in a beaker and ultrasonically agitating for 20 min at room temperature, followed by dripping 60 wt% PTFE suspensions of 0.83 g (CB : PTFE = 4 : 1) into the blend slowly. After being stirred uniformly, the mix was still ultrasonically agitated to disperse the carbon black and PTFE particles to form fine networks of gas channels.²³ The blend was stirred and the redundant alcohol was removed to give a paste. The paste was rolled on either side of the hydrophobic titanium mesh to form a flat sheet of 0.8 mm thickness. The flat sheet was thermolaminated by means of the thermal compression bonding method to obtain the final IGDE of 0.5 mm thickness. The pressure and temperature of the hot-pressing process was 10 MPa and 100 °C, respectively. Then the sheet was sintered for 10 min at 300 °C to sinter the PTFE in order to form a fibrous three-dimensional structure for gas transport.³¹

Fig. 1 Fabrication procedure for the novel GDE.

2.3. Electrolytic procedures

The electrolytic process was performed in a divided three-electrode cell under a constant current using a potentiostat (CH Instruments, Chenhua, Shanghai, China). A cation exchange membrane (Nafion-117) was used to separate the two chambers. The three chambers were for the cathode gas, catholyte and anolyte. The cathodic and anodic chambers had a volume of 15 mL and 30 mL, respectively. An aqueous solution of 0.2 mol L⁻¹ Na₂SO₄ was used as a supporting electrolyte and the initial pH was adjusted using H₂SO₄ or NaOH. Air was used as an oxygen source. When the prepared IGDE (5 cm × 2 cm) was used as the working electrode, a platinum plate (1 cm × 1 cm) was used as the anode because of its overpotential and high chemical stability. The distance between the electrodes was 1.5 cm. A schematic diagram of the experimental setup is shown in Fig. 2.

Fig. 2 Schematic diagram of the experimental setup.

The reaction solutions were collected to determine the H₂O₂ concentration after electrolysis. The yield of H₂O₂ was determined by a chemical titration using an aqueous solution of KMnO₄/H₂SO₄. The current efficiency (CE) of the H₂O₂ formation was calculated from the two-electron reaction against the quantity of charge passed and measured by a coulomb meter (eqn (1)).

where F is the Faraday constant of 96 485 C mol⁻¹.

2.4. Experimental design and statistical model

The optimization of the experimental conditions for the H₂O₂ electrosynthesis involving the IGDE was conducted using the Box–Behnken design (BBD) technique coupled with RSM. The software Design Expert 8.0 was used for the experimental design, data analysis, quadratic model building, and graph plotting. The independent variables of the initial pH, current density and air flow rate were coded with low and high levels in the BBD as shown in Table 1, while the response was expressed as the H₂O₂ concentration after a 1 h reaction. The results along with the experimental conditions are presented in Table 2.

Table 1 Factor levels for the experiments

Process variables	Code	Real values of coded levels
		−1	0	+1
Initial pH	A	2	4	6
Current density (mA cm^−2)	B	20	50	80
Gas flow rate (mL min^−1)	C	20	50	80

---

Table 2 Experimental design matrix using the BBD design and the response in the H₂O₂ concentration

Run	pH	Current density (mA cm^−2)	Gas flow rate (mL min^−1)	H2O2 conc. (mM)
1	6.0	20	50	75.51
2	2.0	50	20	191.26
3	4.0	50	50	308.41
4	6.0	50	20	186.51
5	6.0	50	80	233.25
6	4.0	80	80	167.92
7	6.0	80	50	123.25
8	4.0	50	50	310.62
9	4.0	20	20	85.17
10	2.0	80	50	168.81
11	4.0	80	20	98.27
12	4.0	20	80	94.35
13	2.0	20	50	88.42
14	4.0	50	50	315.17
15	2.0	50	80	215.21

---

The experimental results of the BBD were fitted with a quadratic model as below:²⁶

Y = k₀ + k_aA + k_bB + k_cC + k_abAB + k_acAC + k_bcBC + k_aaA² + k_bbB² + k_ccC² (2)

where Y is the predicted response; k₀ is a constant; k_a, k_b, k_c are the linear coefficients; k_ab, k_ac, k_bc are the cross-coefficients; k_aa, k_bb, k_cc are the quadratic coefficients.

Pareto analysis (P_i)³² gives more significant information to help interpret the results. In fact, this analysis calculates the percentage effect of each factor on the response, according to the following relation:

Analysis of variance (ANOVA) was conducted to analyze the results and to verify the statistical significance of the fitted quadratic models. The interaction between the process variables was illustrated by the three-dimensional (3D) response surface and two-dimensional (2D) contour plots. The optimum process parameters for the electrolytic process were calculated using the fitted models and validated by the experiments.

3. Results and discussion

3.1. In situ electrosynthesis of H₂O₂ by the IGDE

The performance of the IGDE for H₂O₂ electrosynthesis was tested under the conditions of various current densities at pH 3.0, an air flow rate of 40 mL min⁻¹, and 0.2 mol L⁻¹ Na₂SO₄. An apparent increase in the H₂O₂ concentration and a gradual decrease in the current efficiency with electrolytic time are present in Fig. 3a and b. The IGDE at the current density of 60 mA cm⁻² exhibits a higher catalytic activity toward oxygen reduction generating H₂O₂ than that at 40 and 80 mA cm⁻². In this case, a maximum H₂O₂ accumulation of 315.67 mM was achieved by electrolysis for 120 min, and then there is no obvious improvement in the H₂O₂ concentration. This behavior can be explained assuming that, in the steady state, H₂O₂ is electrogenerated and simultaneously destroyed at the same rate in the electrolytic process. These results confirm that the novel preparation procedure of the IGDE was an effective way to improve the H₂O₂ concentration. Table 3 describes an extensive summary of H₂O₂ accumulation using different electrode structures found in the literature. Comparably, a higher H₂O₂ concentration was obtained compared to that of other GDEs reported in publications, indicating that the IGDE is an efficient cathode for H₂O₂ accumulation.

Fig. 3 Effects of current density on (a) H₂O₂ concentration and (b) current efficiency. Reaction conditions: pH 3.0, air flow rate 40 mL min⁻¹, 0.2 mol L⁻¹ Na₂SO₄.

Table 3 Performance comparison with the literature

Electrode structure	Configuration	Time	H2O2 conc.	Ref.
Sheet	Stainless steel mesh, acetylene black–PTFE film	450 min	1130 mg L^−1	14
Sheet	Oxygen-fed graphite/PTFE, 2-ethylanthraquinone	2 h	414 mg L^−1	13
GDE	Silver-plated nickel web, XC-72 carbon layers, acetylene black layers	6000 s	0.12 M	11
Dual GDE	Carbon fiber, diffusion layer, catalyst layer	180 min	1928 mg L^−1	9
GDE	Carbon black layer, tert-butyl-anthraquinone	90 min	301 mg L^−1	10
GDE	Nickel mesh, carbon–PTFE layer	60 min	12 mM	12
GDE	Titanium mesh, conductive catalytic layer	2 h	315.67 mM	Present work

---

It is known that the electrode structure greatly influences the performance of H₂O₂ generation, especially its accumulation.⁹ Indeed, all of the carbon material catalysts so far identified for H₂O₂ electrosynthesis are equally effective for its sequential hydrogenation or decomposition to water.³³ As a result, a higher current efficiency would be achieved at an early stage and gradually decreases with the increasing H₂O₂ concentration. We have now addressed this problem and show that the IGDE can reduce the sequential hydrogenation and decomposition of H₂O₂, thereby producing a higher accumulation of H₂O₂. Additionally, the H₂O₂ production rate in earlier periods is higher than in later ones, inferring that H₂O₂ is also chemically decomposed to H₂O at the IGDE surface.

The surface morphologies of the CCL exposed to the electrolyte were scanned at a magnification of 10 000. The SEM images are presented in Fig. 4. The rope networks should be formed by rolling PTFE which binds the CB particles together and forms air transport channels. From these images, it is clearly shown that the cross-linked networks gradually increased with the decrease of the CB/PTFE ratio. It has been proven that the oxygen reduction reaction (ORR) takes place at the three-phase interface catalyst–air–electrolyte.³⁴ The solid phase provides electron transport and catalyzes the ORR, and the gas phase is responsible for gas diffusion, whereas the liquid phase is responsible for the proton supply and product diffusion.³¹ PTFE is a usual binding material used in the preparation of gas diffusion oxygen reducing electrodes for its hydrophobic properties facilitating oxygen permeability and diffusion. Moreover, PTFE also provides hydrophobicity and enhances the air permeability.² Insufficient PTFE content brings about the uneven distribution of pore channels, with a large cross-section connection (Fig. 4a). The airflow could easily pass through the large channel, while the other compacted surface hardly has any contact with air, which causes insufficient or absent oxygen in the two-electron reaction. There were still a number of macro pores and cross-linked ropes that uniformly existed throughout the CCL even though the CB/PTFE ratio reached 1. However, excessive amounts of PTFE evolved into the formation of a PTFE film, which covers the CCL surface Fig. 4c. For this reason, the number of active sites on the GDE for catalyzing the reduction of oxygen gas to H₂O₂ were greatly reduced. Consequently, the maximum H₂O₂ concentration was obtained at the CB/PTFE ratio of 4 because of its uniform channel size and the adequate number of active site (Fig. 4b).

Fig. 4 SEM images of the conductive catalyst layer surface of the IGDEs where (a) CB/PTFE = 6, (b) CB/PTFE = 4, (c) CB/PTFE = 1.

3.2. Fitting model and analysis of variance

According to the experimental results, an empirical relationship between the response and independent variables was obtained and expressed by the following second-order polynomial equation:

Y = 313.40 − 5.65A + 26.85B + 18.69C − 8.16AB + 5.70AC + 15.12BC − 51.14A² − 146.27B² − 53.71C² (4)

The data of the H₂O₂ concentration was fitted to the quadratic models, and the significance and the adequacy were tested by the ANOVA. The ANOVA results are presented in Table 4. A P-value less than 0.050 indicates that the model terms are significant at a 95% confidence level or higher, while the values greater than 0.100 are usually considered as insignificant.³⁵ For the model predicted by eqn (4), a P value less than 0.0001 shows that these terms are significant for describing the H₂O₂ concentration.

Table 4 ANOVA test for the quadratic models

Source of variations	Sum of squares	DF	F-value	P-value
Model	90 546.72	9	450.02	<0.0001
Residual	89.42	4
Lack of fit	65.66	2	2.76	0.2657
Pure error	23.76	2
Total	90 636.15	13

---

These results (Table 5) show that the regression model has a high coefficient of determination value (R² = 0.999). The R²-value provides a measure of how much variability in the observed response values can be explained by the experimental factors and their interactions. This implies that 99.9% of the variations for the H₂O₂ concentration are explained by the independent variables and this also means that the model does not explain 0.1% of the variation. The value of the adjusted determination coefficient (adjusted R² = 0.9968) also proved the high significance of the model. Additionally, the low value of the coefficient of variation (C. V. = 2.56%) suggested the high precision and reliability of the experiment. In addition, the F-test of the regression models produced very low P-values (<0.0001), indicating that the models were of high significance. According to the above explained ANOVA test results, the application of the model explained the reaction quite well and can be employed to navigate the design space at least in terms of H₂O₂ concentration.

Table 5 Statistical parameters obtained from the analysis of variance for the regression models^a

Response	R^2	Adj. R^2	CV	S. D.	A. P.	PRESS
a A. P.: adequate precision; S. D.: standard deviation; CV: coefficient of variance; PRESS: predicted residual error sum of squares.
H2O2 conc.	0.9990	0.9968	2.56	4.73	56.747	N/A

---

Fig. 5 represents the Pareto graphic analysis. It shows that the current density, air flow rate and initial pH are the most determining factors on H₂O₂ concentration, their effect is over 90% of the investigated response.

Fig. 5 Pareto graphic analysis. (A) initial pH, (B) current density (mA cm⁻²), (C) gas flow rate (mL min⁻¹).

As shown in Fig. 6, the comparison of the actual and predicted H₂O₂ concentration of the process efficiency shows that the predicted data are in good agreement with the experimental ones. Therefore, the regression models can be used to predict the H₂O₂ concentration from the initial experimental conditions.

Fig. 6 Regression plots of actual data against predicted values from the response surface models describing the H₂O₂ concentration.

3.3. Response surface and contour plots

Factors giving significant interaction effects in the new simplified fitted models were chosen for the axes of the response surface plots to account for the curvature of the surfaces.³⁶ The object of this work aims at enhancing the H₂O₂ concentration rather than CE, thus the following research just focuses on the H₂O₂ concentration. The three-dimensional (3D) response surface and two-dimensional (2D) contour plots of the model-predicted responses, while some variables were kept constant and others were varied within the experimental range, were utilized to assess the interactive relationships between the process variables and the H₂O₂ concentration.

3.3.1 Effect of the initial pH and current density. The effect of the initial pH and current density on the H₂O₂ concentration is illustrated in Fig. 7. The peak shown in the contour plot indicates that the highest H₂O₂ concentration is achieved in the range located in that circular contour. It illustrates that there is an obvious interaction between the initial pH and current density on the H₂O₂ concentration. The H₂O₂ concentration increased sharply with increasing current density at a variety of initial pH values until the current density was above 55 mA cm⁻², and then it decreased with the increasing current density. The results also show that the highest H₂O₂ concentration was obtained at pH 4.0, and it was affected by a too high or too low pH value.

Fig. 7 Response surface plot and contour plot of the H₂O₂ concentration as a function of the initial pH and current density. Reaction conditions: air flow rate 50 mL min⁻¹, electrolytic time 60 min.

In this study, a cation exchange membrane (Nafion-117) was used to separate the two electrolytic cells. It obstructs the penetration of anions and H₂O₂ molecules, but allows cations (H⁺ and Na⁺), to freely penetrate through it.² Therefore, H₂O₂ at the cathode will be confined in the cathode chamber, avoiding its decomposition at the anode. In the electrolytic process, protons electrolyzed in the anode chamber will be electrically driven to the cathode chamber, partially supplementing the proton consumption.

Fig. S2† shows the pH values as a function of electrolytic time. A diaphragm electrolytic device could keep the pH < 1 in the anode chamber after 10 min of electrolysis because the oxidation of H₂O releases oxygen gas and protons at the anode (eqn (5)). In the case of the cathode chamber, the initial pH is below 2 keeping the electrolyte acidic in the whole electrolytic process, while the pH rapidly becomes alkaline when the initial pH is above 4. It has been reported that H₂O₂ can be electrochemically generated by the oxygen reduction reaction (ORR) in both acidic (eqn (6))² and alkaline solutions (eqn (7)).³⁷ Therefore, high H₂O₂ concentrations were obtained whether the original solution was acidic, neutral or alkaline.

2H₂O → 4H⁺ + O₂ + 4e⁻ (5)

Alkaline medium: H₂O + O₂ + 2e⁻ → HO₂⁻ + OH⁻ (6)

Acidic medium: O₂ + 2H⁺ + 2e⁻ → H₂O₂ (7)

3.3.2 Effect of air flow rate and current density. Fig. 8 shows the response surface assuming current density and air flow rate as independent factors. The peak shown in the contour plot indicates that the highest H₂O₂ concentration is achieved in the range located in that circular contour. There is an optimum value of the current density for each air flow rate level. It is evident that there is an obvious interaction between the current density and air flow rate on H₂O₂ concentration. As shown in Fig. 3, current density plays an important role in H₂O₂ concentration. The H₂O₂ concentration dropped dramatically once the current density was above 60 mA cm⁻², leading to a low CE.

Fig. 8 Response surface plot and contour plot of H₂O₂ concentration as a function of current density and gas flow rate. Constant conditions: initial pH 4.0, electrolytic time 60 min.

These results can be explained by using the cell potential. During the electrolytic process, H₂O₂ is produced at the cathode surface through the ORR in the acidic or alkaline medium. However, side reactions simultaneously occur in the electrolytic process:² (a) four-electron reaction (eqn (8)), (b) decomposition reaction (eqn (9) and (10)) due to the H₂O₂ accumulation at the GDE interface, and (c) hydrogen evolution reaction (eqn (11)).

O₂ + 4H⁺ + 4e⁻ → 2H₂O (8)

Acidic medium: H₂O₂ + 2H⁺ + 2e⁻ → 2H₂O (9)

Acidic medium: H₂O₂ + HO₂⁻ → H₂O + O₂ + OH⁻ (10)

2H⁺ + 2e → H₂ (11)

To confirm the effects of the side reactions on the H₂O₂ concentration, an electrolyte containing 1.0 wt% H₂O₂ was added into the cathode chamber. The reaction parameters represent the optimum conditions established for the synthesis of H₂O₂. The only difference was that pure nitrogen, instead of air, was the gas source. The results are shown in Fig. 9. The amount of H₂O₂ decomposing increases with the increasing electrolytic time. Moreover, this phenomenon is more remarkable when the current density is higher than 60 mA cm⁻². The inset panel of Fig. 9 shows the variation of the cell potential with the current density. It indicates that a voltage higher than 5.0 V (40 mA cm⁻²) accelerates the H₂O₂ decomposition reaction, demonstrating that the decomposition reaction in the electrolytic process was the major reason for the inhibition of H₂O₂ accumulation. Besides, the ORR through four-electron reaction (eqn (8)) leads to the formation of H₂O instead of H₂O₂ as through two-electron reaction (eqn (6) or (7)) at potential values higher than 4.3 V.^24,38 Indeed, a high potential value should be supplied to the system to obtain a high current density, which accelerates the decomposition of H₂O₂. Moreover, the competitive electrode reactions such as the discharge of hydrogen evolution reaction inhibit the generation of H₂O₂.⁷

Fig. 9 H₂O₂ decomposition as a function of electrolytic time.

3.3.3 Effect of the initial pH and air flow rate. Fig. 10 illustrates the response surface assuming the air flow rate and pH are independent factors. As aforementioned, a high H₂O₂ concentration was obtained under all working conditions. For a constant initial pH, the H₂O₂ concentration in the cathode chamber was roughly proportional to the applied current in an air flow rate, and a steady-state condition was rapidly reached. But there was a considerable decrease in the amount of H₂O₂ when the air flow rate was fixed at a high level (over 50 mL min⁻¹).

Fig. 10 Response surface plot and contour plot of the H₂O₂ concentration as a function of the initial pH and air flow rate. Constant conditions: current density 50 mA cm⁻², electrolytic time 60 min.

The increase of the H₂O₂ concentration with increasing gas flow rate can be explained by two major factors. First, the origin of the effect could be due to oxygen being consumed in the H₂O₂ electrosynthesis by the two-electron reaction (eqn (6) or (7)) and in the nonselective production of H₂O by the four-electron reaction (eqn (8)). Second, the hydrodynamics determine the rate of mass transfer between the liquid phase and the GDE surface.³⁹ A H₂O₂ concentration gradient over the GDE surface is created, increasing the H₂O₂ decomposition reaction rate. It is physically analogous to increasing the stirring rate in a batch reactor. Thus, enhancing the air flow rate promotes the mass transfer rate, which is beneficial to the H₂O₂ accumulation.

As the air flow rate increased, the H₂O₂ concentration increased to the maximum value at the gas flow rate of 50 mL min⁻¹, after that point no further increase in the H₂O₂ concentration was observed. It is seen that a rate of 50 mL min⁻¹ is adequate to maintain the steady-state of oxygen during the electrolytic process. At higher gas flow rates, the mass transfer between the GDE surface and electrolyte will increase, but probably more important is the increase in the mass transfer through the liquid layer surrounding the catalyst surface, resulting in the low catalytic efficiency.³⁹

3.4. Determination of the optimal conditions for the electrosynthesis of H₂O₂ and verification

In the case of multiple responses, RSM describes a range of specific operating conditions that at least keeps them in the desired ranges or in some sense maximizes all responses.³⁵ In this study, the desired goal in terms of the H₂O₂ concentration was defined as the maximization to achieve the highest electrosynthesis performance. The H₂O₂ concentration contour plots in Fig. 7, 8 and 10 show clear peak areas, demonstrating that the optimum conditions of initial pH, current density and air flow rate were within the design boundary. Accordingly, the optimum values of the process variables were demonstrated in Table 6. After verification through further experiments with the predicted values, it indicates that the maximum H₂O₂ accumulation was obtained when the values of each factor were set at the optimum values. The results imply that the strategy to optimize the operating conditions and to obtain the maximum H₂O₂ concentration using RSM for the electrolytic process was successful.

Table 6 Optimum conditions of the process variables for the maximum H₂O₂ concentration

Initial pH	Current density (mA cm^−2)	Gas flow rate (mL min^−1)	H2O2 concentration (mM)
			Actual	Predicted
4.0	52	55	309.85	313.72

---

4. Conclusions

This work has demonstrated that the improved GDE constructed by rolling carbon black and PTFE as a conductive catalyst layer was an efficient cathode for H₂O₂ accumulation. The main reason hindering H₂O₂ accumulation was the subsequent decomposition reaction on the IGDE surface. Response surface methodology (RSM) based on a Box–Behnken design (BBD) was employed to assess the individual and interactive effects of several critical process conditions on H₂O₂ concentration, and to optimize the electrolytic process. The results of ANOVA showed that the regression model was highly significant and can be used to predict H₂O₂ accumulation from the initial experimental conditions. The optimal conditions for the H₂O₂ concentration were found to be an initial pH of 4.0, a current density of 52 mA cm⁻², and an air flow rate of 55 mL min⁻¹. Under the optimal conditions, the H₂O₂ concentration was 309.85 mM after 60 min of electrolysis. The obtained results demonstrated the usefulness of response surface methodology in predicting the electrolytic process as well as the interactive effects of manipulating process variables.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (Grant No. 51408370). The authors are also greatly appreciative to Wenxiang Zhang (University of Technology of Compiegne, France) for reviewing the manuscript and his suggestive comments.

References

1. J. M. Campos-Martin, G. Blanco-Brieva and J. L. Fierro, Angew. Chem., Int. Ed., 2006, 45, 6962–6984.
2. Z. Qiang, J.-H. Chang and C.-P. Huang, Water Res., 2002, 36, 85–94.
3. T. Murayama and I. Yamanaka, J. Phys. Chem. C, 2011, 115, 5792–5799.
4. M. Teranishi, S.-I. Naya and H. Tada, J. Am. Chem. Soc., 2010, 132, 7850–7851.
5. F. Thevenet, J. Couble, M. Brandhorst, J. Dubois, E. Puzenat, C. Guillard and D. Bianchi, Plasma Chem. Plasma Process., 2010, 30, 489–502.
6. I. Yamanaka, S. Tazawa, T. Murayama, R. Ichihashi and N. Hanaizumi, ChemSusChem, 2008, 1, 988–992.
7. R. M. Reis, A. A. Beati, R. S. Rocha, M. H. Assumpcao, M. C. Santos, R. Bertazzoli and M. R. Lanza, Ind. Eng. Chem. Res., 2011, 51, 649–654.
8. M. Assumpção, R. De Souza, D. Rascio, J. Silva, M. Calegaro, I. Gaubeur, T. Paixão, P. Hammer, M. Lanza and M. Santos, Carbon, 2011, 49, 2842–2851.
9. X. Yu, M. Zhou, G. Ren and L. Ma, Chem. Eng. J., 2015, 263, 92–100.
10. R. B. Valim, R. M. Reis, P. S. Castro, A. S. Lima, R. S. Rocha, M. Bertotti and M. R. Lanza, Carbon, 2013, 61, 236–244.
11. M. Giomo, A. Buso, P. Fier, G. Sandonà, B. Boye and G. Farnia, Electrochim. Acta, 2008, 54, 808–815.
12. M. Panizza and G. Cerisola, Electrochim. Acta, 2008, 54, 876–878.
13. J. Forti, J. Nunes, M. Lanza and R. Bertazzoli, J. Appl. Electrochem., 2007, 37, 527–532.
14. Y. Sheng, S. Song, X. Wang, L. Song, C. Wang, H. Sun and X. Niu, Electrochim. Acta, 2011, 56, 8651–8656.
15. M. Assumpção, R. De Souza, R. Reis, R. Rocha, J. Steter, P. Hammer, I. Gaubeur, M. Calegaro, M. Lanza and M. Santos, Appl. Catal., B, 2013, 142, 479–486.
16. X. Zhang, J. Fu, Y. Zhang and L. Lei, Sep. Purif. Technol., 2008, 64, 116–123.
17. C. McDonnell-Worth and D. R. MacFarlane, RSC Adv., 2014, 4, 30551–30557.
18. J. Forti, R. Rocha, M. Lanza and R. Bertazzoli, J. Electroanal. Chem., 2007, 601, 63–67.
19. G. Xia, Y. Lu and H. Xu, Electrochim. Acta, 2015, 158, 390–396.
20. F. Yu, M. Zhou and X. Yu, Electrochim. Acta, 2015, 163, 182–189.
21. W. Zhang, G. Huang, J. Wei and D. Yan, Desalination, 2013, 311, 31–36.
22. Q. Shi, H. Wang, S. Liu and Z. Bian, RSC Adv., 2014, 4, 56263–56272.
23. H. Dong, H. Yu, X. Wang, Q. Zhou and J. Feng, Water Res., 2012, 46, 5777–5787.
24. A. Khataee, M. Safarpour, M. Zarei and S. Aber, J. Electroanal. Chem., 2011, 659, 63–68.
25. A. Moraes, M. Assumpção, R. Papai, I. Gaubeur, R. Rocha, R. Reis, M. Calegaro, M. Lanza and M. Santos, J. Electroanal. Chem., 2014, 719, 127–132.
26. T. Xu, Y. Liu, F. Ge, L. Liu and Y. Ouyang, Appl. Surf. Sci., 2013, 280, 926–932.
27. Z. Zhang and H. Zheng, J. Hazard. Mater., 2009, 172, 1388–1393.
28. J. Virkutyte, E. Rokhina and V. Jegatheesan, Bioresour. Technol., 2010, 101, 1440–1446.
29. M. Alim, J.-H. Lee, C. Akoh, M.-S. Choi, M.-S. Jeon, J.-A. Shin and K.-T. Lee, LWT--Food Sci. Technol., 2008, 41, 764–770.
30. H. Wang and J. Wang, Electrochim. Acta, 2008, 53, 6402–6409.
31. C. Santoro, K. Artyushkova, S. Babanova, P. Atanassov, I. Ieropoulos, M. Grattieri, P. Cristiani, S. Trasatti, B. Li and A. J. Schuler, Bioresour. Technol., 2014, 163, 54–63.
32. A. R. Khataee, M. Zarei and L. Moradkhannejhad, Desalination, 2010, 258, 112–119.
33. J. K. Edwards, B. Solsona, E. Ntainjua, A. F. Carley, A. A. Herzing, C. J. Kiely and G. J. Hutchings, Science, 2009, 323, 1037–1041.
34. H. Dong, H. Yu, H. Yu, N. Gao and X. Wang, J. Power Sources, 2013, 232, 132–138.
35. I. Arslan-Alaton, G. Tureli and T. Olmez-Hanci, J. Photochem. Photobiol., A, 2009, 202, 142–153.
36. H. Li, S. Zhou, Y. Sun and J. Lv, Waste Manage., 2010, 30, 2122–2129.
37. V. Antonin, M. Assumpção, J. Silva, L. Parreira, M. Lanza and M. Santos, Electrochim. Acta, 2013, 109, 245–251.
38. A. Özcan, Y. Şahin, A. S. Koparal and M. A. Oturan, J. Electroanal. Chem., 2008, 616, 71–78.
39. S. J. Freakley, M. Piccinini, J. K. Edwards, E. N. Ntainjua, J. A. Moulijn and G. J. Hutchings, ACS Catal., 2013, 3, 487–501.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra09636g

---