Optimization and scalability of two-electron water oxidation using C-fibre paper as electrode assisted by sodium stannate

Abstract

Electrochemical H₂O₂ generation offers a safer, cost-effective, and sustainable alternative for H₂O₂ production compared to classical methods. The two-electron water oxidation reaction (2e⁻ WOR) has a significant advantage over the cathodic pathway, as it can be coupled with other reduction processes, such as CO₂ valorization. Several anodic materials have been explored, with boron-doped diamond (BDD) standing out due to its high performance, but its high cost limits scalability. In this regard, recent studies have demonstrated the possibility of using cost-effective carbon materials by controlling the media conditions. In this work, we present a comparison between C-fibre paper (CFP) and BDD as anodes for 2e⁻ WOR. By thoroughly adjusting the operating parameters, CFP achieves more stable production and maintains similar or even higher generation rates and faradaic efficiencies (FE). Moreover, electrode passivation issues related to the use of Na₂SiO₃ additive were identified, and Na₂SnO₃ was proposed as a novel alternative stabilizer for H₂O₂ electrogeneration with excellent results in terms of production and anode durability. Up to 23.3 µmol cm⁻² min⁻¹ production rates and ∼38% FE were achieved using a bare CFP electrode in both H and flow-cell configurations, and CFP electrode behaviour was stable after three 6-hour cycles in a flow cell, showing the scalability potential of the developed system.

---

Introduction

Hydrogen peroxide (H₂O₂) is an essential chemical with a wide range of applications across industries. One of its primary uses is as a powerful oxidizing agent in environmental and industrial processes, including water treatment¹ and paper and textile bleaching.² H₂O₂ serves as an environmentally friendly alternative to harsher chemicals, since it decomposes into only water and oxygen. Moreover, the healthcare and pharmaceutical sectors heavily rely on H₂O₂ for its antiseptic and disinfectant properties. In recent years, its role in sanitation and infection control has expanded, particularly during public health crises such as pandemics.³ Additionally, advancements in propulsion and energy storage systems are exploring H₂O₂ as a potential clean fuel component,⁴ adding another layer to its industrial significance. As demand continues to grow across both traditional and emerging sectors, the need for improved and sustainable production methods for hydrogen peroxide is becoming increasingly urgent.

The most common industrial method for producing hydrogen peroxide (H₂O₂) is the anthraquinone process, also known as the Riedl-Pfleiderer process.⁵ This method involves the cyclic hydrogenation and oxidation of an organic compound, anthraquinone. In the process, anthraquinone is first hydrogenated using hydrogen gas to form anthrahydroquinone. This reduced form is then oxidized with air or oxygen, during which hydrogen peroxide is formed. The anthraquinone is regenerated in its original form and reused in the cycle, making the process highly efficient and cost-effective. The H₂O₂ is then extracted and purified, typically in an aqueous solution of varying concentrations.

This method is preferred in industry due to its scalability, high yield, and the relatively mild conditions under which it operates. It also allows continuous production, which is ideal for meeting global demand. Additionally, the use of organic solvents in the process allows for efficient separation of hydrogen peroxide from the reaction mixture. Despite its effectiveness, the anthraquinone process does involve complex equipment and careful handling of flammable materials, prompting ongoing research into alternative, greener production methods such as direct synthesis from hydrogen and oxygen. Nonetheless, the anthraquinone process remains the backbone of global H₂O₂ production today.

Electrochemical production of H₂O₂ is an alternative to the classical industrial production through the anthraquinone process, which requires large plants and infrastructures, and where the produced H₂O₂ must be transported to the point of use. Additionally, H₂O₂ is unstable and potentially explosive at high concentrations, imposing safety challenges for transportation and storage. The electrochemical method provides the possibility of on-site production of H₂O₂ with lower concentrations, increasing safety and lowering costs.⁶

Electrochemical production of hydrogen peroxide (H₂O₂) primarily occurs via two routes: cathodic and anodic pathways.⁷ In the cathodic approach, oxygen is reduced at the cathode—typically through a two-electron oxygen reduction reaction (2e⁻ ORR)—to produce H₂O₂ in aqueous media. Conversely, the anodic process involves the direct, thermodically unfavoured, oxidation of water or hydroxide ions at the anode to form H₂O₂ or related peroxo compounds. While cathodic production is more widely studied and closer to commercialization, the anodic route may be more attractive due to its potential to operate without the need for external oxygen gas, reducing system complexity and cost.⁸ Additionally, it can be paired with several large-scale cathode reactions that produce valuable chemical substances in an electrochemical cell.⁹

For this reason, the research must be focused on the optimization of H₂O₂ production process through 2e⁻ WOR to produce as much highly stable H₂O₂ as possible, avoiding its decomposition (eqn (4)) and minimizing secondary reactions toward other possible compounds whose formation is thermodynamically favored. The catalytic activity toward the 2e⁻ WOR (eqn (1)) for evolving H₂O₂ is governed by the thermodynamics of reaction (eqn (1)) at the surface of catalysts, which competes with 4e⁻ WOR (eqn (2)), and the one-electron H₂O₂ oxidation to form O₂ (eqn (3)).¹⁰

2H₂O ↔ H₂O₂ + 2H⁺ + 2e⁻ (E₀ = 1.76 V vs. SHE) (1)

2H₂O ↔ O₂ + 4H⁺ + 4e⁻ (E₀ = 1.23 V vs. SHE) (2)

H₂O₂ ↔ O₂ + 2H⁺ + 2e⁻ (E₀ = 0.67 V vs. SHE) (3)

2H₂O₂ ↔ O₂ + 2H₂O (4)

The activity and selectivity of WOR are also correlated with the adsorption free energies of several oxygenated intermediates at the catalyst surface.¹¹ These energies are associated with the surface properties of the catalyst. Moreover, the stability of the produced H₂O₂ depends on its spontaneous decomposition, which can be accelerated by the medium conditions.¹² It is therefore needed broader research focus on catalytic materials and their potential for H₂O₂ production, as they can promote 2e⁻ WOR over competing pathways. In this context, several anode materials, such as stannate¹³ or bismuth oxides¹⁴ have been studied, with boron-doped diamond (BDD) standing out for its high yield and faradaic efficiency.¹⁵ However, BDD electrodes are expensive to manufacture due to the complex chemical vapor deposition (CVD) fabrication process, the high-purity diamond substrate, and the precise boron doping. More recently, Pangotra et al. demonstrated that controlling the reaction medium conditions (mainly composition and pH) as well as using Na₂SiO₃ during the electrochemical process as a stabilizing additive enable the use of cheaper commercial carbon materials.¹² The present work shows a thorough comparison between BDD and carbon fibre paper (CFP) as electrodes for anodic H₂O₂ production. The comparison reveals the feasibility of CFP as a cost-effective substitute for the BDD electrode owing to its similar production rates but more stable H₂O₂ concentrations, when suitable reaction conditions and stabilizing agents are selected. In this context, long-term experiments have highlighted the drawbacks associated with the use of Na₂SiO₃ as a stabilizing agent, and Na₂SnO₃ as alternative stabilizer is here reported with excellent results. The combination of optimized reaction parameters and the selected novel additive has provided higher and stable H₂O₂ production rates via water oxidation. Finally, the results obtained in an H-cell have been transferred to a flow cell, demonstrating the scalability of the process.

Experimental section

Reagents and solutions

Potassium carbonate (K₂CO₃, ≥99.5%), potassium bicarbonate (KHCO₃, ≥99%), sulfuric acid (H₂SO₄, 95%), and potassium hydroxide (KOH) were acquired from VWR. Sodium oxalate (Na₂C₂O₄) was purchased from Merck, sodium stannate trihydrate (Na₂SnO₃·3H₂O, 99%) was obtained from Cymit, while sodium metasilicate (Na₂SiO₃) and potassium permanganate (KMnO₄, ACS reagent, ≥99.0%) were supplied by Sigma-Aldrich.

All reagents were used as received, without further purification or modification, except for potassium permanganate, whose stock solution was hot filtered to remove impurities.

Aqueous solutions were freshly prepared using deionized ultrapure water from a Milli-Q purification system (resistivity ≥18.2 M Ω cm). The following solutions were prepared: 2 M K₂CO₃, 2 M KHCO₃, 1 M H₂SO₄, 10 M KOH, and 0.02 N KMnO₄. A stock solution of 0.2 N KMnO₄ was prepared in Milli-Q water and filtered at 80 °C. The solution used for titration (0.02 N KMnO₄) was obtained by dilution of the stock with Milli-Q water and was previously standardized to accurately determine the concentration of the titrant employed in the quantification of H₂O₂.

Materials and equipment

Toray carbon paper 060 5 wt% Wet Proofed (Fuel Cell Store) and BDD/Nb NeoCoat-Electrodes with 2500 ppm boron concentration (NeoCoat SA) were selected as materials to be used as working electrodes in the anodic compartment with 3 M KCl Ag/AgCl (Metrohm) as a reference electrode for H₂O₂ production in H-cell. In the cathodic compartment, a Pt foil counter electrode 10 × 10 mm (MTI) was used. A Nafion 115 proton-exchange membrane (Fuel Cell Store) was used for compartment separation. A Peltier cell 40 × 40 mm (Cetronic) connected to a 5 A, 0–24 V DC laboratory power supply (Cetronic) and a syringe pump NE-1000 (New Era Pump Systems Inc.) allowed the control of the temperature and pH of the process.

For scaling-up experiments, a dedicated test rig was assembled with a filter-press flow cell (Apria Systems, Spain), a titanium plate as cathode, and two centrifuge pumps.

Electrochemical procedures were conducted using an Autolab/PGSTAT 128N potentiostat/galvanostat and its accompanying NOVA 2.1 software for data acquisition. Electrogenerated H₂O₂ was quantified by KMnO₄ titration (848 titrino plus, Metrohm).

Electrochemical generation of H₂O₂

Electrochemical generation of H₂O₂ was carried out, except for scaling-up experiments, in a two-compartment, three-electrode, and low-volume (75 mL each compartment) electrochemical H-cell. Here, a Nafion 115 proton-exchange membrane separated the anodic and cathodic compartments (Fig. S1) to prevent H₂O₂ degradation due to exposure to the hydrogen evolution reaction (HER) occurring in the cathodic compartment.¹⁵ A boron-doped diamond (BDD) material and CFP were used as working electrodes (3.75 cm²) in the anodic compartment, with silver/silver chloride (3 M KCl) as reference electrode, and platinum foil acting as counter electrode in the cathode. Both anodic and cathodic compartments containing 2 M K₂CO₃ and 2 M KHCO₃, respectively, were continuously stirred at 450 rpm using PTFE-encapsulated stirring bars. H₂O₂ production was performed by subjecting the working electrode to different potentials or currents through chronoamperometric/potentiometric procedures.

CFP and BDD electrodes were not subjected to any pretreatment prior to use. After use, the electrodes were rinsed in water to remove any weakly adsorbed residues before the characterization. It must be highlighted that, for CFP system optimization, new electrodes were used for each test, in order to avoid introducing additional variables related to electrode surface after used. For CFP recycling experiments in the flow cell, the electrodes were washed with water for 10 minutes at the maximum flow rate before each new experiment and immediately after the previous one, in order to maintain cleanliness and preserve electrode durability.

An H-cell system was used to optimize medium conditions and electrochemical parameters to maximize H₂O₂ production and the stability of the compound. The optimal working potential of 2.5 V vs. Ag/AgCl for H₂O₂ production with BDD electrodes was fixed based on previous studies.¹⁵ Optimal working potential for CFP as anode was determined at 2.2 V by studying its performance in the range between 2.0 and 2.4 V vs. Ag/AgCl, avoiding higher values that can lead to potential degradation of the CFP material (Fig. S2 and S3). In first instance, 3-hour experiments were carried out under potentiostatic conditions with the optimal concentration of Na₂SiO₃ as stabilizer or in its absence for production process optimization and material comparison. However, for subsequent studies, different concentrations and types of stabilizing compounds, as well as longer experiments and galvanostatic production, were evaluated.

Automatic addition of KOH with a syringe pump (Fig. S4) was implemented to keep the pH constant at the desired value. A Peltier device was used to cool the anolyte when it was required (Fig. S4). Subsequently, the scaling up of the best results achieved in the H-cell was performed in the flow filter-press cell designed by Apria systems with an electrode of 50 cm². This cell consisted of a cathodic chamber with a titanium plate as cathode (counter electrode) and an anodic chamber with CFP as working electrode, separated by the same aforementioned Nafion 115 proton-exchange membrane, as well as two centrifugal pumps, and two 1 L plastic tanks (Fig. S5). Each tank contained 700 mL of electrolyte. The pH was kept at 11.5 by regularly adding KOH. The temperature was maintained between 20 and 25 °C with the help of a refrigerated glass coil.

H₂O₂ analytic quantification

The concentration of electrogenerated (EG) H₂O₂ was determined by titration using a potentiometric titrator (See “Materials and Equipment”). The method is based on the redoxreaction between potassium permanganate and hydrogen peroxide in acidic medium, described by the following ionic reaction:

2MnO₄⁻(aq) + 5H₂O₂(aq) + 6H⁺(aq) → 2Mn²⁺(aq) + 5O₂(g) + 8H₂O(l) (5)

During the electrooxidation process, 5 mL samples from the H-cell/flow-cell anodic compartment were collected to monitor the reaction progress. Each sample was added to 150 mL of 1 M H₂SO₄ under constant stirring. Titration was performed with 0.02 N KMnO₄ solution until the equivalent point was reached to determine H₂O₂ concentration. Faradaic efficiency at each point was as well determined according to previously defined method.¹²

Materials characterization after electrolysis

The morphology and surface composition of the electrodes after 2e⁻ WOR process under different reaction conditions were examined by scanning electron microscopy (SEM) using an eLINE Plus system (Raith GmbH) at the Interdepartmental Research Service of the Universidad Autónoma de Madrid (UAM). Particular attention was paid to the formation of passivating surface layers in the presence and absence of stabilizing additives. Elemental composition was determined by energy-dispersive X-ray spectroscopy (EDX) coupled to the SEM system. X-ray Photoelectron Spectroscopy (XPS) data were obtained at the Institute of Catalysis and Petrochemistry (ICP-CSIC) with a SPECS GmbH system equipped with a hemispherical energy analyzer PHOIBOS 150 9MCD. A non-monochromatic Al X-ray source was used with a power of 200 W and voltage of 12 kV. Samples were placed first in the pre-treatment chamber at room temperature and degassed for several hours before being transferred to the analysis chamber. Pass energies of 50 and 20 eV were used for acquiring both survey and high-resolution spectra, respectively. The photoelectron spectra were calibrated using the graphitic peak (sp²,–C=C−) at 284.4 eV of the CFP. Adventitious carbon at 284.8 eV was used as the reference for the reactive samples. The data fitting was carried out with CasaXPS software version 2.3.26. A handbook¹⁶ and an online database¹⁷ were used to identify the most common species. The solid precipitates formed during electrolysis (pH 11) were collected, dried, and analyzed by X-ray diffraction (XRD) at the Institute of Materials Science of Madrid (ICMM-CSIC) using a D8 Advance TT powder diffractometer.

Results and discussion

H₂O₂ anodic production optimization with CFP

In water oxidation reactions, medium conditions (composition, pH, temperature) and the use of a stabilizing agent are crucial for achieving high H₂O₂ production rates and a stable H₂O₂ concentration over time.^7,8,12,18,19 Related to medium composition, 2 M KHCO₃ was used in the cathode to ensure good conductivity, and 2 M K₂CO₃ was employed as electrolyte in the anodic compartment, based on previous studies that reported the key role of CO₃²⁻ ions for the H₂O₂ production by WOR^12,19 and the results of the developed confirmation tests (Fig. S6). Although KHCO₃ provides a more stable environment for the formed H₂O₂ given the lower alkalinity of the medium (pH 8.8 for a 2 M concentration), the activity of HCO₃⁻ ions is much lower than that of CO₃²⁻ ions, and this high activity of CO₃²⁻ greatly promotes the production of H₂O₂ accordingly to previously reported mechanisms^18,19(Fig. S7), despite the more alkaline environment (pH 12.6 at 2 M concentration).¹² This higher activity of carbonate as electrolyte is reflected not only on production rate, but also on the FE of the process, favouring the selective production of H₂O₂ (See Table S1).

Concentration of Na₂SiO₃ as stabilizing agent. The positive effect of the addition of sodium metasilicate (Na₂SiO₃) as a stabilizer of hydrogen peroxide during the electrogeneration process was previously reported.¹² Based on the mentioned study, the stabilization capability of different concentrations of Na₂SiO₃ over time was evaluated (Fig. 1a). For both concentrations of stabilizer, 90 mM and 180 mM Na₂SiO₃, H₂O₂ production increased by more than 100%. However, although for short duration electrogeneration experiments (1 h), the H₂O₂ production is maximum with 180 mM Na₂SiO₃, these higher concentrations provide a more unstable environment, causing H₂O₂ losses over time. In this context, 90 mM Na₂SiO₃ seems to provide a more stable concentration for longer experiments, even though the production in the first hour was slightly lower than for 180 mM Na₂SiO₃. The use of a lower silicate concentration allows better control over the evolution of the system; although the initial H₂O₂ generation rate is slightly slower, production is maintained more steadily over time. In contrast, at 180 mM Na₂SiO₃, H₂O₂ decomposition may proceed faster than its stabilization by silicate, since the higher amount of Na₂SiO₃ further increased the alkalinity of the solution.

Fig. 1 Optimization of reaction parameters for H₂O₂ anodic electrogeneration: (a) H₂O₂ concentration obtained during WOR for 2 M K₂CO₃ solutions containing different Na₂SiO₃ concentrations, (b) 2 M K₂CO₃ solutions and 90 mM Na₂SiO₃ at different pH, and (c) 2 M K₂CO₃, 90 mM Na₂SiO₃ at pH 11.5 and two temperature ranges.

Effect of the pH. Higher H₂O₂ production rates have been linked to the increased presence of CO₃²⁻, which occurs at high pH in the carbonate-bicarbonate system.^12,19 However, although high CO₃²⁻ concentrations increase the production, pH values above 10 accelerate the decomposition rate of H₂O₂ (eqn (4)), especially in the presence of carbonaceous materials.²⁰ Moreover, H₂O oxidation generates H⁺ that acidifies the anolyte and produces a loss of the CO₃²⁻ ions needed to increase the activity. Thus, it is necessary to maintain the desired pH with the addition of a base. In this regard, KOH was added periodically to keep the studied pH value. These optimization studies revealed the key role of pH on the obtained H₂O₂ concentration and selectivity of the process, where between 11.5–12.5 pH values, the production of H₂O₂ and FE increases very quickly as the pH decreases. However, at pH 11 and below, the anolyte became increasingly turbid over time, and a whitish solid precipitated on the electrode surface, causing the current to drop rapidly around 30 minutes after the reaction started (Fig. S8). This behaviour was attributed to electrode surface passivation caused by the formation of a solid layer. In order to complete the 3-hour experiment and compare it with the other pH curves, the electrode was removed, slightly scratched cleaned, and re-immersed several times; however, the current continued to decrease steadily, leading to similar H₂O₂ concentrations and faradaic efficiencies compared to pH 11.5 system (Fig. 1b and Table S1)

The precipitate formed at pH 11 was initially characterized by XRD (Fig. S9), revealing a predominantly amorphous structure, with a broad halo in the 2θ range of 20–30° and minor crystalline reflections. Based on these XRD results, together with complementary surface analyses by SEM, EDX, and XPS (discussed in detail in the following sections on passive layer characterization), the precipitated material is attributed to the formation of an amorphous Si–O–based compound, compatible with hydrated SiO₂ or a related silicon oxide phase. The formation of this layer is linked to electrode surface passivation and a loss of electrode durability at pH ≤ 11.5.

At pH > 12, the production of hydrogen peroxide was reduced by about 50%, as well as the selectivity of the process, due to the excessive alkalinity of the medium, despite the presence of stabilizer. Based on these results, pH 11.5 was identified as the optimum working pH, giving good production and selectivity while avoiding the precipitation observed at lower pHs, maximizing the durability of the working electrode.

Effect of the temperature. Various cooling levels were investigated for an optimal electrogeneration of H₂O₂, seeking a compromise between minimizing its thermal decomposition and ensuring efficient reaction kinetics, as it is well known how an increase in temperature reduces the stability of peroxide solutions by accelerating the decomposition of H₂O₂,²¹ regardless of any other conditions employed.²² Since the electrochemical production of H₂O₂ is an exothermic reaction, an efficient cooling system is needed to prevent excessive heating and maintain higher production rates.

As shown in Fig. 1c, two temperature ranges were evaluated: near room temperature (20–25 °C) and in the range of 30–35 °C, to determine the actual cooling requirements of the system. To achieve temperature control, a Peltier device was assembled to the anolyte side of the cell. At the beginning of the experiments, H₂O₂ production and the selectivity of the process (Table S1) was nearly identical in both temperature ranges, as the system heating occurred gradually and was not significant during the first hour of operation. Therefore, the average temperature of the system, even without cooling, remained close to ambient temperature for a substantial portion of the initial period (1 h). However, during the second and third hours, the detrimental effect of overheating became evident, resulting in a noticeable decrease in H₂O₂ concentration and faradaic efficiency. Lower temperatures were not explored, as the objective was to find a balance between mitigating the negative thermal effects on H₂O₂ stability and maintaining favorable reaction kinetics. Excessive cooling would lead to a reduction in the reaction rate, thus lowering the overall efficiency of H₂O₂ production.^23,24

Comparison of BDD and CFP materials for H₂O₂ production

As mentioned above, BDD stands out for its remarkable H₂O₂ generation capacity via the two-electron water oxidation reaction (2e⁻ WOR). The wide potential window, chemical stability, and corrosion resistance enable operation under harsh electrochemical conditions, often achieving high selectivity and current efficiency for H₂O₂ production.^15,18,25,26 However, despite its excellent performance, the complexity and cost of the manufacturing process²⁷ have driven research toward exploring different types of materials for BDD substitution.

Although a variety of metal-based catalysts have been reported for selective 2e⁻ water oxidation to H₂O₂, these systems are often evaluated under conditions that emphasize intrinsic selectivity rather than high production rates, long-term operation, or scalability.^28–31 Furthermore, recent literature consistently highlights several limitations that restrict their practical applicability, as in their pristine form, these materials typically exhibit insufficient intrinsic activity, and limited operational stability under the potentials required for sustained H₂O₂ production (Table S4).^8,32,33 Significant performance improvements often rely on advanced strategies such as doping, defect engineering, and facet control, which involve complex nanostructuring and precise atomic-level modifications that can complicate scalability and reproducibility.^34,35 For this reason, the comparative analysis in this work will focus on carbon-based electrodes and BDD systems investigated under more comparable reactor configurations and operating strategies.

In recent years, there has been a growing interest in using carbon-based materials as a cost-effective alternative, not only for H₂O₂ production via oxygen reduction reaction (ORR),³⁶ but also for WOR, highlighting the potential of carbonaceous electrodes as promising candidates for replacing BDD in H₂O₂ electrogeneration, significantly reducing materials costs.^12,18,37,38 However, these materials have not been directly compared within the same study. In this section, a comparison between the two types of materials (CFP and BDD) under different reaction conditions is discussed. It should be noted that BDD was tested at 2.5 V vs. Ag/AgCl to achieve maximum selectivity,¹⁵ whereas CFP was evaluated at 2.2 V vs. Ag/AgCl, as determined through previous optimization studies (Fig. S2) to prevent oxidative degradation at harsh conditions (Fig. S3).

Comparison of the effect of Na₂SiO₃ addition. First, the influence that the addition of a stabilizer had on each of the materials was evaluated without controlling the reaction conditions. According to previous studies, Na₂SiO₃ acts by adsorbing onto anodic substrates, modifying the active sites, and avoiding the complexation of H₂O₂, preventing the subsequent reactions that lead to H₂O₂ decomposition.³⁹ Thus, different concentrations of Na₂SiO₃ additive¹² were tested in this work (0, 90, and 180 mM) to study the effect of the stabilizing agent on H₂O₂ production for each material (Fig. 2a). Experimental results reveal that, for both electrode materials, the need for a stabilizing additive to ensure long-term production of hydrogen peroxide is clear. BDD exhibit more unstable behaviour, showing a stronger dependence on higher amounts of stabilizer to maintain H₂O₂ production. This could be attributed to the higher oxidative power of BDD materials when compared with carbon ones, which leads to the formation of more hydroxyl intermediates capable of reacting with the electrogenerated H₂O₂, thereby promoting its destruction.⁴⁰ In fact, metasilicate concentration of 90 mM appears insufficient to counteract the combined effects of medium alkalinization and the formation of reactive intermediates, which accelerate H₂O₂ decomposition. As a result, H₂O₂ production at 2 and 3 hours is lower than expected, indicating that the rate of destruction exceeds the stabilizing effect provided at this concentration. Increasing metasilicate concentration to 180 mM enhances stabilization, partially compensating for these destructive processes and allowing higher H₂O₂ production during the second hour.

Fig. 2 Comparison between CFP and BDD materials for H₂O₂ production via 2e⁻ WOR: (a) effect of Na₂SiO₃ as stabilizing agent without controlling the reaction conditions, and effect of pH and temperature on H₂O₂ production without stabilizing agents using (b) CFP and (c) BDD.

In the absence of additives, BDD doubled the amount of H₂O₂ concentration produced on CFP surfaces. However, with 90 mM Na₂SiO₃, the H₂O₂ concentration was equalized at the first hour, and it was remarkable that, in the BDD system, H₂O₂ decomposed faster in the subsequent hours. For 180 mM Na₂SiO₃, CFP reached similar values to those observed at 90 mM, indicating no benefit of further adding Na₂SiO₃ when using the CFP electrode. There was some higher decomposition when increasing the stabilizer, which may be explained by the basification of pH produced by the additive. These results highlight the potential of CFP as a cost-effective alternative for BDD as electrode for H₂O₂ production via WOR.

Comparison of the effect of pH and temperature. Fig. 2b and c shows the influence of the reaction medium conditions for each material on H₂O₂ production in the absence of stabilizer. Although when controlling the reaction conditions an improvement was observed for both materials, CFP electrodes seem to be much more sensitive to these conditions, whereas in the case of BDD, and based on previous results (see Comparison of the effect of Na₂SiO₃ addition), the stabilization of the formed product using additives is the main factor. Under suitable pH and temperature conditions, and despite of stabilizing absence, CFP anodes even exceed the H₂O₂ concentration obtained with BDD materials, demonstrating their potential as a competitive alternative for hydrogen peroxide generation via the water oxidation reaction (WOR).

The different behaviour observed in each case can be explained by the structural differences between both electrodes: carbonaceous materials such as CFP usually provide porous surfaces with great adsorption ability due to the sp²-hybridized orbitals characteristic of these materials.^41,42 Furthermore, they are usually enriched with oxygenated groups whose reactivity strongly depends on the medium conditions.^42–44 In contrast, BDD materials are characterized by their relatively inert surface with low adsorption capability due to sp³-hybridized orbitals present in the diamond structure, which also provide good stability in a wide range of media.^41,45,46

Therefore, changes in the reaction conditions will have a more severe influence on the CFP surface, affecting the final product and its stability (Fig. 2b). However, for BDD electrodes, the employment of stabilizing additives seems to be more relevant, as the different environmental conditions have less influence on the material surface (Fig. 2c). Moreover, the slightly decrease in concentration for BDD after the first hour can be attributed to its strong oxidative character which causes the evolution of the electro-generated hydrogen peroxide into other compounds due to the reaction with hydroxyl ions.^47,48

Faradaic efficiency for CFP and BDD systems. Table S2 corroborate the potential of CFP as a highly competitive and cost-effective alternative to BDD for H₂O₂ production, as it is capable of providing not only higher and more stable concentrations of H₂O₂ over time under the optimal conditions as discussed above, but also greater process selectivity. Under baseline conditions (without stabilizer or pH and temperature control), both materials exhibit similar initial faradaic efficiency and decline over time, dropping from 17–20% to roughly 5% after three hours of operation (due to the evolution of secondary reactions). However, the implementation of medium conditions control (optimal pH and temperature), as well as the addition of a stabilizing agent (90 mM Na₂SiO₃) reveal significant advances for the CFP electrode in terms of selectivity: although both materials show an initial boost in FE, for CFP electrodes it maintains a markedly more stable selectivity profile over time. For instance, even without pH or temperature control, when 90 mM of Na₂SiO₃ is added in both systems, the FE for CFP remains at 23.78% after 2 h process, whereas BDD FE drop to 13.80%. This observation aligns with the strongest oxidative nature of BDD materials which facilitates the over-oxidation of hydrogen peroxide into secondary non-selective species.^47,48

Effect of using Na₂SiO₃ on CFP durability and alternative additives

Long-term experiments at different Na₂SiO₃ concentrations. In the presence of Na₂SiO₃ and at pH 11.5, a drop in current was observed before the end of the three-hour experiments. Under the possible hypothesis of electrode passivation formulated after observing a precipitate at pH 11 the effect of the stabilizer on electrode durability was analysed. Tests with different concentrations of Na₂SiO₃ were carried out with the aim of reaching 6 h without a current drop.

The durability of the electrode was defined as the time at which the current starts to drop because of surface passivation.

According to the proposed stabilization mechanism, the stabilizer is adsorbed onto the electrode surface, modifying its active sites and preventing hydrogen peroxide decomposition by means of previous complexation.³⁹ This process likely involves the formation of a thin film that becomes thicker over time and causes surface fouling. The conducted experiments revealed that the process appears to be non-reversible, even if periodic cleaning of the electrode surface is done and more stabilizer is added, leading in all cases to a current drop and thus inhibiting hydrogen peroxide electrogeneration. For this reason, the modulation of Na₂SiO₃ concentration plays a key role in terms of finding a compromise between the minimum quantity needed to avoid production losses due to the lack of stabilization, but, at the same time, does not reach the critical layer thickness on the electrode that significantly limits its durability. Fig. 3a shows the relationship between the presence of Na₂SiO₃ and the durability of the electrode. According to the aforementioned hypothesis, the durability of the material increases with a decrease in metasilicate concentration, reaching 6 hours of total durability with 40 mM Na₂SiO₃. However, Fig. 3b clearly indicates that this concentration is not enough to stabilize the electrogenerated H₂O₂, leading to significant production losses. By reducing the amount of stabilizer to 50 mM, the life of the electrode was significantly extended, reaching the maximum peroxide concentration, although more slowly than at 90 and 70 mM. Therefore, 50 mM Na₂SiO₃ was determined to be the optimum concentration, achieving a compromise between increased electrode lifetime and stable H₂O₂ production.

Fig. 3 Na₂SiO₃ effect on H₂O₂ production and CFP lifetime: (a) evaluated durability during six-hour experiments at different Na₂SiO₃ concentrations, and (b) H₂O₂ concentration evolution with CFP during long-term experiments at different Na₂SiO₃ concentrations. System conditions: pH 11.5; 25 °C, 2M K₂CO₃ as electrolyte solution.

Alternative additives. In addition to considering the Na₂SiO₃ concentration reduction, other possible alternative stabilizers were evaluated to totally overcome the problems of electrode passivation. Although not in electrochemical processes, compounds such as phosphates⁴⁹ and sodium stannate⁵⁰ have been reported to enhance the stability of H₂O₂ during storage. Consequently, these compounds were investigated as potential stabilizers for the in situ electrogeneration of hydrogen peroxide.

It can be observed from Fig. 4a that, in the case of the phosphates, the experiment was not extended further than 3 h, as the produced H₂O₂ was even lower than that of the solutions without stabilizer, thus revealing a non-stabilization effect on the electrogenerated H₂O_2. The only stabilizer that showed significant improvements compared to the production without additives was Na₂SnO₃, with no significant differences when its concentration was doubled. The result with Na₂SnO₃ as stabilizer can be compared with the previous one with Na₂SiO₃ (Fig. 4b). The use of 90 mM stannate as a stabilizer gives similar results to those obtained with 40 mM silicate in terms of H₂O₂ production and provides enhanced durability of the electrode, as it avoids surface passivation throughout the whole experiment. These experimental results evidence different behaviour for Na₂SiO₃ and Na₂SnO₃ under the working conditions during 2e⁻ WOR. Silicate ions (SiO₃²⁻) exhibit higher adsorption affinities to solid surfaces than other common ions⁵¹ and in mild alkaline or near-neutral aqueous environments can undergo condensation to form oligomeric or colloidal silica networks, which can adsorb onto or precipitate over surfaces.⁵² The local acidification produced near the electrode may promote these polymerization reactions, leading to the formation of surface-bound silicate species that contribute to electrode passivation. Conversely, there is no evidence of polymerization tendencies for SnO₃²⁻ ions in electrochemical systems. In fact, sodium stannate has been used as an additive to mitigate anodic corrosion^53,54 and the high stability of their soluble complexes formed in aqueous alkaline media has also been reported.⁵⁵

Fig. 4 Evaluation of H₂O₂ electrogeneration in the presence of novel stabilizers: (a) H₂O₂ concentration obtained with Na₂HPO₄, KH₂PO₄ and Na₂SnO₃ as potential stabilizing agents, and (b) comparison of Na₂SnO₃ and Na₂SiO₃ stabilization potential for long-term 2e⁻ WOR reactions. System conditions: pH 11.5; 25 °C, 2 M K₂CO₃ as electrolyte solution.

The absence of a passivation effect using this additive with CFP suggests that it interacts to a weaker degree with carbon surfaces and tends to be stable in solution without forming blocking layers, achieving a balance between stabilization and electrochemical activity preservation. Thus, in this work, it is hypothesized that, whereas the stabilization mechanism of Na₂SiO₃ implies the formation of silicate species that gradually passivate the electrode surface, Na₂SnO₃ remains only mildly interactive and stable in aqueous alkaline media, avoiding complete blockage of active sites while still enhancing H₂O₂ stability, demonstrating its high potential as an alternative stabilizer.

Morphological and chemical surface characterization

SEM/EDX analysis. Fig. 5 illustrates the distinct surface modifications of CFP electrodes induced by Na₂SiO₃ and Na₂SnO₃ during the 2e⁻ WOR for H₂O₂ production process. Pristine CFP fibres (Fig. 5a) exhibit a clean and smooth surface without evidence of secondary phase deposition. In contrast, the use of Na₂SiO₃ as a stabilizing agent results in the formation of a surface layer (Fig. 5b and c). Nanometric or sub-micrometric aggregates, typical of gels or amorphous silica deposited via sol–gel processes or chemical precipitation are observed at high magnification (Fig. S10). The layer continuously covers the fibre, with no regions showing visible crystals. This morphology is characteristic of amorphous or gel-like material.⁵⁶

Elemental analysis by EDX reveals Si:O atomic ratios of approximately 2–3 (Fig. S11b, c, and Table S5), which is consistent with the formation of a hydrated amorphous SiO₂ layer on the CFP surface.^57,58 This assignment is further supported by the amorphous morphology observed in SEM images (Fig. S10), characteristic of non-crystalline silica deposits.

The extent of surface passivation strongly depends on the solution pH. At pH 11.5 (Fig. 5b), the deposited layer is relatively thin and discontinuous, which correlates with a moderate electrode lifetime of approximately 2 h. In contrast, under slightly more acidic conditions (pH 11, Fig. 5c), a substantially thicker and crust-like layer develops, leading to rapid blockage of the electrode surface and premature current decay.

Fig. 5 SEM images (×500) of the surface of CFP electrodes after 2e⁻ WOR in 2 M K₂CO₃ with different stabilizers: (a) pristine CFP, (b) 2 h 2e⁻ WOR with 90 mM Na₂SiO₃ at pH 11.5, (c) 0.5 h 2e⁻ WOR with 90 Na₂SiO₃ 90 mM at pH 11, (d) 6 h 2e⁻ WOR with 90 mM Na₂SnO₃ at pH 11.5.

Conversely, when Na₂SnO₃ is employed as the stabilizing agent, no significant passivating layer is detected. SEM images (Fig. 5d) show only minor adsorbed residues, and Sn is scarcely detected by EDX (Fig. S11d and Table S5). These observations are consistent with the enhanced electrode durability and the negligible current decay observed during electrochemical measurements.

XPS analysis. In the XPS spectra of the pristine electrode, the surface is predominantly composed of carbon and fluorine species (Table S6), with oxygen present at less than 1 at%. This observation is consistent with the deconvolution of the C 1s peak (Fig. 6a), which reveals a dominant graphitic contribution alongside a minor fraction of oxygen-containing functional groups, specifically C–O and C=O. The presence of these oxygen functionalities is further supported by the deconvolution of the O 1s peak (Fig. 6b). Additionally, the C 1s spectrum exhibits pronounced contributions from CF₂ and CF₃ groups,⁵⁹ corresponding to the backbone and terminal chain groups of the PTFE treatment on the purchased electrode surface.

Fig. 6 Photoelectron spectra of (a) C 1s and (b) O 1s of a CFP pristine electrode and electrodes used with 90 mM Na₂SiO₃ and 90 mM Na₂SnO₃ additives at pH 11.5. (c) Si 2p photoelectron spectra of the Na₂SiO₃ powder used as additive and the electrode used with 90 mM Na₂SiO₃ at pH 11.5.

Electrodes used with Na₂SiO₃ and Na₂SnO₃ additives at pH 11.5 exhibit a marked increase in surface oxygen content, reaching 18.2 at% and 6.0 at%, respectively (Table S6). For the electrode used with Na₂SnO₃, C–O functionalities remain the predominant oxygen-containing species (Fig. 6a). In contrast, the electrode used with Na₂SiO₃ displays a greater relative contribution of C=O groups, suggesting that the surface experiences a more oxidizing environment in the presence of this additive. Moreover, in both electrodes, the fluorine–carbon contributions are largely obscured by K 2p; however, in the electrode used with Na₂SiO₃, a distinct K 2p plasmon is observed at 298.8 eV, which can be attributed to the more oxidizing surface environment, resulting in electron withdrawal from potassium and reduced screening of its photoemitted.⁶⁰

The formation of a passivating surface layer is consistent with the high silicon content (11.7 at%) detected when Na₂SiO₃ is used as an additive (Table S5), in contrast to the significantly lower tin content (0.6 at%) observed for electrodes used with Na₂SnO₃. Furthermore, the Si 2p core level is shifted toward higher binding energies compared to that of the Na₂SiO₃ reagent (Fig. 6c), indicating a modified chemical environment for silicon on the electrode surface. Deconvolution of the Si 2p spectrum reveals two main components. For the Si 2p_3/2 level, one contribution is observed at 104.4 eV and a second at 105.5 eV.

Previous spectroscopic studies of silicate polymers demonstrate that the Si 2p binding energy is sensitive to the local oxygen coordination environment, with Si–OH (hydrated silica) and other non-bridging oxygen species exhibiting higher binding energies relative to bulk SiO₂ (typically ∼103.4 eV) due to increased electron withdrawal by electronegative ligands.⁶¹ Therefore, the lower binding energy component can be assigned to highly hydrated SiO₂. On the other hand, the contribution at 105.8 eV represents an unusually high binding energy for SiO₂, even in hydrated form. We therefore hypothesize that SiO₂ species located near the highly oxidized carbon surface undergo a much stronger electron-withdrawing interaction, leading to this pronounced shift. A shift toward higher Si 2p binding energies has been observed in SiO₂ samples treated by plasma surface activation using oxygen reactive ion etching.⁶²

This interpretation is further supported by the O 1s spectrum (Fig. 6b), which, in addition to contributions attributable to expected species such as C–O, C=O, Si–OH, SiO₂, and adsorbed H₂O,^61,63 exhibit an additional high binding energy component at 535.3 eV with significant intensity (see O 1s spectrum of Na₂SiO₃ powder for comparison, Fig. S12). This latter contribution may be associated with peroxo (–OO/–OOH) species and molecular oxygen. These species are usually not stable under standard environmental conditions, as only weakly physisorbed oxygen has been described as such high binding energies.⁶⁴ We speculate that they are trapped inside the carbon porous matrix and beneath the SiO₂-like passivation layer, stabilized owing to the hydrophobic conditions. This hypothesis of high-binding-energy oxygen trapped in cavities has also been previously theorized to explain oxygen signals detected in a fast-growth gold oxide film activated by high-discharge oxygen on gold foil.⁶⁵

These oxygen species with high binding energy are also detected to a much lesser extent in the sample with the Na₂SnO₃ additive, surviving in the hydrophobic porosities of the carbon electrode, but not as favored under a passivating layer. The low amount of Sn in the surface of the sample shows, unlike with Na₂SiO₃, that if there is some formation layer, this should be very thin. The Sn peak for the electrode is shifted 1 eV with respect to that of the Na₂SnO₃ powder (Fig. S13), which can also be explained by the loss of electron density of residual SnO_x deposits in the oxidized substrate.

The electrode surface characterization shows a different behaviour for Na₂SiO₃ than for Na₂SnO₃. Silicate additives promote passivation due to the strong tendency of silicon species to form silica networks,^66,67 a more evident process at lower pH values.⁵² In contrast, stannate contributes to the stabilization of the formed compound without inducing the formation of an appreciable passivating layer, owing to the high stability of soluble complexes formed in aqueous alkaline media.⁵⁵

Maximizing H₂O₂ production: from potentiostatic to galvanostatic mode

H₂O₂ production study under galvanostatic conditions. Current density is reported to be a key variable for H₂O₂ production boosting in water electrolysis processes.^18,68,69 Potentiostatic mode is excellent for lab-scale studies, mechanism research, and selectivity optimization. However, galvanostatic conditions are more suitable for scale-up objectives because they maintain stable current, avoids potential distribution across large electrodes, and simplify process control in industrial stacks. The experimental setup remained similar to previous experiments, but now the reference electrode was used only to register the anodic potential. In Fig. 7a, different current densities (from 125 to 200 mA cm⁻²) were studied to select the one that maximized H₂O₂ production. Higher current densities were avoided as they raise the potential over 2.2 V vs. Ag/AgCl that could degrade the CFP electrode (Fig. S3). The increase in current density led to a higher production of H₂O₂, with a more pronounced increase at 200 mA cm⁻². The monitored potential ranged from 1.85 V at the lowest current density (125 mA cm⁻²) to 2.2 V at 200 mA cm⁻².

Fig. 7 Galvanostatic production of H₂O₂: (a) effect of applied current density, and (b) comparison of maximum H₂O₂ concentration obtained with different stabilizers and the influence of the operational mode. System conditions: pH 11.5; 25 °C, 2M K₂CO₃ and 90 mM Na₂SnO₃ as electrolyte solution/stabilizing agent.

As shown in Fig. 7b, galvanostatic tests demonstrated that, under suitable operating conditions, H₂O₂ production with Na₂SnO₃ as stabilizer can surpass the previous results obtained with sodium metasilicate, while also extending the electrode lifetime. The fact that the measured potential at 200 mA cm⁻² is 2.2 V, the same as the working potential in the potentiostatic tests, confirms the advantages of this operational mode for H₂O₂ production, as it allows better process control and a uniform current distribution. It should be noted that galvanostatic assays with metasilicate were not evaluated due to the passivation issues characteristic of this compound. The aim of this study was therefore to achieve similar production rates by overcoming the main limitations of the well-studied stabilizer through the use of an alternative compound (Na₂SnO₃), which was successfully achieved by obtaining a production up to 23.3 µmol cm⁻² min⁻¹, exceeding the maximum production of 20.6 µmol cm⁻² min⁻¹ initially achieved with metasilicate additive.

Faradaic efficiency tendencies under different current densities. Table S3 highlights a non-monotonic dependence of FE on current density, which can be explained by the evolution of competing reactions under the different conditions. At lower current densities (125 mA·cm⁻²), the anodic overpotential is lower, and the kinetics of the competing oxygen evolution reaction (OER, 4e⁻) are less favourable,^7,70 allowing a significant fraction of the current to be channelled through the selective 2e⁻ pathway. From 125 to 150–175 mA cm⁻², a systematic FE drop is appreciated. This could be attributed to a higher anodic overpotential that promotes OER over H₂O₂ production via 2e⁻ WOR.^7,70 Since OER is thermodynamically and kinetically more favourable at high potentials, a greater proportion of the current is directed towards O₂ production, reducing the faradaic efficiency for H₂O₂ electrogeneration, although the production is higher (Fig. 7a) as a consequence of a higher electron transfer rate when current is increased. However, at 200 mA cm² the FE increases relative to intermediate currents. This behaviour stems from mass transport limitations and intense O₂ bubble evolution observed experimentally. At high current densities, intermittent shielding of the electrode surface by gas bubbles may hinder the complex, multi-electron OER more severely than the 2e⁻ pathway.⁷¹ Furthermore, the enhanced convective flow induced by bubbling facilitates the rapid removal of H₂O₂ from the electrode interface, reducing surface reoxidation or decomposition. Thus, H₂O₂ selectivity is governed by a balance between kinetics and transport: while OER competition peaks at intermediate currents, high-current physical effects partially recover selectivity. For this reason, 200 mA cm⁻² is selected as the optimal current density not only in terms of H₂O₂ production but also in terms of process selectivity.

Upscaling to a flow-cell configuration

Flow-cell and H-cell systems comparison. Motivated by the good results obtained under galvanostatic conditions, and to evaluate the potential of this process to be applied to an industrial environment, it was scaled to a flow cell configuration. A continuous current of 200 mA cm⁻² was set to ensure the maximum H₂O₂ production and selectivity, based on H-cell system results. Moreover, two different flow rates, 1.4 and 2.34 L min⁻¹, were evaluated (see Fig. S14). A higher production rate was observed in the first hour at 2.34 L min⁻¹ and, thus, was chosen for the flow-cell experiments. This result is explained by an enhancement of the convective mass transport, reducing the residence time of the electrolyte near the electrode surface and limiting H₂O₂ accumulation. This mitigates side reactions such as oxygen evolution and H₂O₂ decomposition, resulting in higher faradaic efficiency and production rate.¹⁸ As illustrated in Fig. 8a, H₂O₂ production after one hour was similar to that obtained in the H-cell. However, in the second and third hours, the production decreased more compared to the H-Type cell. The slightly lower production in the 2^nd and 3^rd hour can be attributed to the self-decomposition of the H₂O₂. The presence in the circuit of materials other than glass, for example, plastics and some metallic parts, can accelerate this process.⁷² Nevertheless, it can be concluded that the process is potentially scalable. Similarly, Fig. 8b shows the evolution of H₂O₂ concentration in both systems, tending to stabilize at slightly lower values for the flow-cell system due to the abovementioned causes.

Fig. 8 Scalability potential evaluation of the optimized 2e⁻ WOR process. Comparison of 2 e⁻ WOR for three hours in H-cell and flow cell systems: (a) H₂O₂ production rate and (b) H₂O₂ concentration. (c) Evolution of H₂O₂ concentration for the 2 e⁻ WOR with a recycled electrode in flow conditions during three 6 h cycles. System conditions: Q = 2.34 L min⁻¹ pH 11.5, 25 °C, j = 200 mA cm⁻² 2 M K₂CO₃ and 90 mM Na₂SnO₃ as electrolyte solution/stabilizing agent.

Long-term experiments and recycling tests

The experiments (Fig. 8c) were conducted in a flow cell over three 6 h cycles using the same recycled CFP electrode. In all cycles, H₂O₂ accumulated over time, reaching high concentrations without appreciable abrupt performance losses. Although a slight decrease in concentration is observed between the second and third cycles, the overall trend indicates that the electrode retains most of its activity after prolonged operation.

The decrease in H₂O₂ concentration observed between consecutive cycles is mainly attributed to a progressive loss of selectivity of the CFP electrode caused by the well-known carbon material surface oxidation under prolonged anodic polarization,^73–76 and supported by the surface characterization section. This surface modification is expected to occur predominantly during the first cycle, when the carbon surface undergoes the most significant chemical transformation. These findings align with the faradaic efficiency results (Table 1), where lower FE values measured at equivalent times in later cycles suggest the presence of a modified CFP surface chemistry, likely associated with oxidation-induced changes in surface functional groups.

Table 1 Faradaic efficiency data for CFP recycling tests in flow-cell configuration. Conditions: 3 cycles of use. 6 h 2e⁻ WOR per cycle with 2 M K₂CO₃ + 90 mM Na₂SnO₃ at Q = 2.34 L min, pH 11.5, 25 °C and j = 200 mA cm⁻²

Faradaic efficiency (%)
Time (h)	Cycle of use
	1st	2nd	3rd
1	38.23	26.58	24.07
2	24.57	18.02	16.80
3	16.88	13.58	11.72
4	50.10	41.54	36.38
5	24.59	20.32	18.25
6	15.74	13.24	12.17

---

Such functionalization of carbon materials under anodic conditions is known to modify their electronic structure and adsorption properties,^73–78 often promoting OER activity relative to partial oxidation pathways.^79–82 Accordingly, the initial electrochemical oxidation of CFP can be regarded as a conditioning process that alters the surface chemistry and shifts the reaction selectivity. Importantly, the relatively small difference in H₂O₂ production between the second and third cycles indicates that, once this initial surface modification has occurred, the electrode exhibits stable behaviour under prolonged operation. The limited performance loss (≤10% between cycles) suggests only minor additional oxidation or gradual degradation, demonstrating that the system can operate reproducibly after an initial conditioning stage. This stability, together with the sustained production observed over extended operation times, supports the practical potential and scalability of the 2e⁻ WOR process in flow configurations using reusable CFP electrodes.

Surface passivation can be reasonably excluded as the main cause of the performance changes. A passivating layer would be expected to progressively hinder the electrogeneration of H₂O₂, leading to a continuous decline in H₂O₂ concentration. Instead, the experimental results show that H₂O₂ concentration tends to stabilize over time in each cycle, indicating the establishment of a dynamic equilibrium between production and destruction pathways rather than transport limitations imposed by a blocking layer. This steady-state behaviour and the surface characterization of the electrodes using Na₂SnO₃, with no evidence of layer formation, support the interpretation that the dominant effect is a change in catalytic selectivity rather than physical passivation. Moreover, the cell potential was the same between cycles and during them, excluding any interfacial resistance increase due to passivation.

While electrochemical 2e⁻ WOR for H₂O₂ production in flow reactors has been previously demonstrated,^12,18 there were no systematic reports of this extended operation over multiple long cycles with large electrolyte volumes and high flow rates. The present work provides new insights into the operational stability and scalability of this process. The sustained performance and electrode reusability demonstrate the robustness of the system and represent an important step toward practical, scalable implementation beyond short-term laboratory experiments.

Comparison with previous works

Production of H-cell systems. In the present work, under optimal conditions, a production of 23.3 µmol cm⁻² min⁻¹ and 37.46% FE with CFP as anode were reached. Compared with the results reported for BDD electrodes¹⁵ (Table 2), the 19.7 µmol cm⁻² min⁻¹ and 28% FE achieved in potentiostatic mode were overcome. When compared with the galvanostatic production, the results are below the reported 35 µmol cm⁻² min⁻¹ for BDD, although the achieved selectivity of the developed system is quite similar.¹⁸ However, these reported results were achieved after 10 minutes, with the H₂O₂ produced afterward decreasing drastically, so the developed system of this work offers clear advantages in terms of stability and longer-time productions over BDD electrodes for H₂O₂ production via 2e⁻ WOR. In the case of the results previously obtained for the same material,¹² CFP production is much higher and similar to that of a carbon electrode modified interfacially with PTFE,¹¹ and a super-aerophilic B-doped graphite catalyst.⁸³ For significantly higher production rates and faradaic efficiencies, more complex modification strategies beyond the scope of this work, such as a self-assembled membranes modification strategy⁸⁴ are required. Nevertheless, obtaining the competitive value of 23.3 µmol cm⁻² min⁻¹ and 37.4% FE with a pristine CFP electrode through simple adjustments of the reaction medium and operational parameters results very appealing scenario for large-scale application of the water oxidation method for H₂O₂ production. Moreover, the results obtained after one hour of operation are particularly relevant for scale-up considerations, as they reveal system stability, steady-state selectivity, and an accumulative yield. They also indicate good control of secondary effects such as H₂O₂ chemical decomposition, electrode fouling, and ohmic heating.

Table 2 Comparative table of results for anodic batch production of H₂O₂ in basic media

We material	Operational mode	Electrolyte solution	µmax (µmol cm^−2 min^−1)	FEmax (%)	Sampling time (min)	Ref.
BDD	Potentiostatic, 3.15 V vs. RHE	KHCO3, 2M (pH = 8)	19.7	28	5	15
BDD	Galvanostatic, 300 mA cm^−2	K2CO3 2M (pH = 12.6)	35	40	10	18
CFP	Galvanostatic, 50 mA cm^−2	KHCO3, 2M Na2SiO3 90 mM (pH = 11)	1.44	—	10	12
PTFE/CFP	Potentiostatic, 2.4 V vs. RHE	Na2CO3 1M Na2SiO3 30 mM (pH = 12)	23.4	66	420	11
B-doped graphite	Potentiostatic, 2.87 V vs. RHE	K2CO3 2M	26.7	60.6	40	83
SAMs-modified CFP	2.1 V vs. RHE	2 M K2CO3	79.8	82.5	10 (or reaching 10C)	84
CFP	Galvanostatic conditions, J = 200 mA cm^−2	K2CO3, 2M Na2SnO3 90 mM (pH = 11.5)	23.3	37.46	60	This work

---

Production in flow cells. The use of a flow cell allows continuous, efficient, and scalable H₂O₂ production, as well as more realistic operating conditions than batch H-cells. Here, the results in the developed flow configuration are compared with the most relevant articles for BDD and CFP using a similar system (Table 3). Compared with the same electrode material, CFP, in the current work, the production is 5 times higher than previously reported, achieving a significative higher faradaic efficiency as well, revealing the high potential of the optimized system for scale-up H₂O₂ production via 2e⁻ WOR using alternative, cost-effective materials such CFP vs. BDD electrodes.

Table 3 Comparative analysis of anodic production of H₂O₂ in flow cells

We material	Jµmax (mA cm^−2)	Flow rate (L min^−1)	Electrolyte solution	µmax(µmol cm^−2 min^−1)	FEmax (%)	µmax sampling time (min)	Ref.
BDD	300	0.01	K2CO3, 2M Na2SiO3 90 mM (pH = 12.6)	36.6	12	60	18
CFP	100	0.1	K2CO3, 2M Na2SiO3 90 mM (pH = 12.6)	4.5	14.3	150	12
CFP	200	2.34	K2CO3, 2M Na2SnO3 90 mM (pH = 11.5)	23.3	38.23	60	This work

---

Conclusions

The potential of a cost-effective electrode material, CFP, has been evaluated as a viable alternative to the expensive BDD electrodes for the anodic production of H₂O₂. A comparison between the two materials highlighted the greater importance of controlling operational parameters, particularly pH and temperature when using CFP due to its high surface reactivity. In contrast, BDD performance was less sensitive to reaction conditions, with the addition of stabilizing agents being the primary factor influencing H₂O₂ stability. When pH and temperature were precisely controlled, CFP exhibited comparable or even superior performance to BDD in terms of H₂O₂ production and process selectivity.

Furthermore, this study demonstrates that the use of Na₂SiO₃—to the best of our knowledge, the only additive reported for the electrochemical generation of H₂O₂ via the 2e⁻ WOR—can be detrimental under certain operating conditions, limiting its suitability for long-term operation. Among the potential studied alternatives to Na₂SiO₃, Na₂SnO₃ has, for the first time in an electrochemical system, been demonstrated to effectively stabilize the anodic production of H₂O₂. Comprehensive surface characterization demonstrated that Na₂SiO₃ induces electrode passivation through the formation of a silica network, while Na₂SnO₃ stabilizes the system by remaining predominantly soluble and preventing the development of a blocking layer on the electrode surface, avoiding passivation.

The combination of pristine CFP with Na₂SnO₃ as an additive, together with precise control of pH and temperature, enabled H₂O₂ production rates of up to 23.3 µmol cm⁻² min⁻¹ and a faradaic efficiency of 38.23% at 2 kA m⁻² for more than 6 h in a flow cell. To the best of our knowledge, these values represent the highest H₂O₂ production rates and selectivity reported to date for flow-cell systems operating with bare CFP for this reaction pathway under similar conditions. Moreover, the developed system operates at a larger and more industrially relevant scale than comparable studies, which are usually performed at much smaller laboratory scales or require complex substrate modifications.

This work also represents the only reported flow-cell study using bare CFP electrodes that includes long-term electrode recycling experiments, achieving a total operational time of 18 h and thereby providing a further step toward practical scalability. Overall, these results surpass those reported for other systems employing uncoated CFP electrodes and are competitive with systems based on more sophisticated modified carbon electrodes and even BDD, confirming the feasibility of anodic H₂O₂ production using a simple, cost-effective, and scalable electrochemical platform.

Author contributions

Alicia Ruiz-Marín: investigation, formal analysis, validation, writing – original draft. José I. Lozano: conceptualization, supervision, writing. Antonio J. Fernández-Ropero: investigation, supervision, writing – review & editing.

Conflicts of interest

There are no conflicts to declare.

Data availability

The authors declare that all the data supporting the findings of this study are available within the article supplementary information (SI), or available from the corresponding author on reasonable request. Supplementary information is available. See DOI: https://doi.org/10.1039/d5ra09601d.

Acknowledgements

The research conducted here has been funded by the European Union's Horizon Europe under Grant Agreement No. 101058578 (WATERPROOF project). Views and opinions expressed are, however, those of the author(s) only and do not necessarily reflect those of the European Union or the European Health and Digital Executive Agency. Neither the European Union nor the granting authority can be held responsible for them.

References

1. Y. Deng and R. Zhao, Curr. Pollut. Rep., 2015, 1, 167–176.
2. R. Hage and A. Lienke, Angew. Chem., Int. Ed., 2006, 45, 206–222.
3. S. L. Simanjuntak, D. Zakiawati and N. Nur’aeny, Eur. J. Med. Health Res., 2025, 3, 102–117.
4. W. Kopacz, A. Okninski, A. Kasztankiewicz, P. Nowakowski, G. Rarata and P. Maksimowski, FirePhysChem, 2022, 2, 56–66.
5. T. Nishimi, T. Kamachi, K. Kato, T. Kato and K. Yoshizawa, Eur. J. Org. Chem., 2011, 2011, 4113–4120.
6. S. Li, J. Ma, F. Xu, L. Wei and D. He, Chem. Eng. J., 2023, 452, 139371.
7. Z. Gao, Y. Cao, Q. Zhu, C. Wang, W. Bai and J. Zhu, J. Environ. Chem. Eng., 2024, 12, 111960.
8. H. Cao, G. Chen, Y. Yan and D. Wang, ChemSusChem, 2025, 18, e202401100.
9. S. Mavrikis, M. Nieuwoudt, M. Göltz, S. Ehles, A. Körner, A. Hutzler, E. Fossy, A. Zervas, O. Brai, M. Wegener, F. Doerrfuss, P. Bouwman, S. Rosiwal, L. Wang and C. Ponce de León, Adv. Energy Mater., 2024, 14, 2304247.
10. S. Mavrikis, S. C. Perry, P. K. Leung, L. Wang and C. Ponce de León, ACS Sustain. Chem. Eng., 2021, 9, 76–91.
11. C. Xia, S. Back, S. Ringe, K. Jiang, F. Chen, X. Sun, S. Siahrostami, K. Chan and H. Wang, Nat. Catal., 2020, 3, 125–134.
12. D. Pangotra, L.-I. Csepei, A. Roth, C. Ponce de León, V. Sieber and L. Vieira, Appl. Catal., B, 2022, 303, 120848.
13. S. Y. Park, H. Abroshan, X. Shi, H. S. Jung, S. Siahrostami and X. Zheng, ACS Energy Lett., 2019, 4, 352–357.
14. K. Sasikumar and H. Ju, Molecules, 2025, 30, 789.
15. S. Mavrikis, M. Göltz, S. Rosiwal, L. Wang and C. Ponce de León, ACS Appl. Energy Mater., 2020, 3, 3169–3173.
16. J. F. Moulder, J. Chastain and R. C. King, Handbook of X-Ray Photoelectron Spectroscopy : a Reference Book of Standard Spectra for Identification and Interpretation of XPS Data, Physical Electronics, 1995.
17. https://www.xpsfitting.com/.
18. D. Pangotra, L.-I. Csepei, A. Roth, V. Sieber and L. Vieira, Green Chem., 2022, 24, 7931–7940.
19. S. Mavrikis, M. Göltz, S. Rosiwal, L. Wang and C. Ponce de León, ChemSusChem, 2022, 15, e202102137.
20. J. Yu and C. Lyu, Sep. Purif. Technol., 2024, 349, 127809.
21. L. Wang, Y. Wang, G. Wu, W. Feng, T. Zhang, R. Yang, X. Jin and H. Shi, Int. J. Chem. React. Eng., 2013, 11, 265–269.
22. W. D. Nicoll and A. F. Smith, Ind. Eng. Chem., 1955, 47, 2548–2554.
23. I. Huerta, P. Biasi, J. García-Serna, M. J. Cocero, J.-P. Mikkola and T. Salmi, Green Process. Synth., 2016, 5, 341–351.
24. P. J. M. Cordeiro-Junior, J. Lobato Bajo, M. R. de V. Lanza and M. A. Rodrigo Rodrigo, Ind. Eng. Chem. Res., 2022, 61, 10660–10669.
25. K. Wenderich, B. A. M. Nieuweweme, G. Mul and B. T. Mei, ACS Sustain. Chem. Eng., 2021, 9, 7803–7812.
26. P. J. Espinoza-Montero, P. Alulema-Pullupaxi, B. A. Frontana-Uribe and C. E. Barrera-Diaz, Curr. Opin. Solid State Mater. Sci., 2022, 26, 100988.
27. N. Jaiswal, H. Khan, S. John, S. Singh, A. Tricoli, B. Liu and K. Ramanujam, J. Alloys Compd., 2025, 1036, 181723.
28. J. Baek, Q. Jin, N. S. Johnson, Y. Jiang, R. Ning, A. Mehta, S. Siahrostami and X. Zheng, Nat. Commun., 2022, 13, 7256.
29. J. H. Baek, T. M. Gill, H. Abroshan, S. Park, X. Shi, J. Nørskov, H. S. Jung, S. Siahrostami and X. Zheng, ACS Energy Lett., 2019, 4, 720–728.
30. S. Y. Park, H. Abroshan, X. Shi, H. S. Jung, S. Siahrostami and X. Zheng, ACS Energy Lett., 2019, 4, 352–357.
31. X. Shi, S. Back, T. M. Gill, S. Siahrostami and X. Zheng, Chem, 2021, 7, 38–63.
32. Q. Shentu, K. Xing, K. Li, N. Han and Y. Li, Adv. Funct. Mater., 2025, e18926.
33. Z. Wang, S. Liu, Y. Liu, X. Wei, N. Liang, Z. Sang, J. Jiang and B. Li, Small, 2025, 21, 2410612.
34. L. Fan, X. Bai, C. Xia, X. Zhang, X. Zhao, Y. Xia, Z.-Y. Wu, Y. Lu, Y. Liu and H. Wang, Nat. Commun., 2022, 13, 2668.
35. Z. Wang, X. Duan, M. G. Sendeku, W. Xu, S. Chen, B. Tian, W. Gao, F. Wang, Y. Kuang and X. Sun, Chem Catal., 2023, 3, 100672.
36. C. Rivera-Vera, M. A. Rodrigo-Rodrigo, C. Saez, A. Thiam and R. Salazar-González, Chemosphere, 2024, 348, 140764.
37. X. Zhao, C. Ni, Q. Chang, B. Chang, P. Zhang, F. Sun, J. Shi and C. Li, J. Mater. Chem. A, 2025, 13, 40767–40775.
38. Y. Guo, X. Tong and N. Yang, Nanomicro Lett., 2023, 15, 77.
39. A. L.-T. Pham, F. M. Doyle and D. L. Sedlak, Environ. Sci. Technol., 2012, 46, 1055–1062.
40. C.-H. Lee, E.-S. Lee, Y.-K. Lim, K.-H. Park, H.-D. Park and D.-S. Lim, RSC Adv., 2017, 7, 6229–6235.
41. P. Brosler, A. V. Girão, R. F. Silva, J. Tedim and F. J. Oliveira, Front. Mater., 2023, 10, 1020649.
42. C. Liu, C. Sun, Y. Gao, W. Lan and S. Chen, ACS Omega, 2021, 6, 19153–19161.
43. S.-J. Park and B.-J. Kim, Mater. Sci. Eng., A, 2005, 408, 269–273.
44. Y. Wang, H. Li, B. Cui, X. Xu and Y. Wang, J. Compos. Sci., 2023, 7, 231.
45. N. Vinokur, B. Miller, Y. Avyigal and R. Kalish, J. Electrochem. Soc., 1996, 143, L238–L240.
46. T. L. Read and J. V. Macpherson, J. Visualized Exp., 2016, 107, e53484.
47. F. Sopaj, M. A. Rodrigo, N. Oturan, F. I. Podvorica, J. Pinson and M. A. Oturan, Chem. Eng. J., 2015, 262, 286–294.
48. L. C. Espinoza, A. Henríquez, D. Contreras and R. Salazar, Electrochem. Commun., 2018, 90, 30–33.
49. P. C. Wegner, Hydrogen peroxide stabilizer and resulting applications, US Pat., US6815408B2, 2023.
50. T. J. Lewis and T. M. Walters, J. Appl. Chem., 1960, 10, 403–407.
51. C. Yin, J. Huang, S. Deng, C. Ma and Y. Peng, Front. Chem., 2025, 13, 1646971.
52. M. Kley, A. Kempter, V. Boyko and K. Huber, Langmuir, 2014, 30, 12664–12674.
53. Y.-Z. Lv, Y.-Z. Jin, Z.-B. Wang, Y.-F. Li, L. Wang and D.-X. Cao, RSC Adv., 2014, 4, 18074–18079.
54. D.-R. Yang, Z.-L. Wu, K. Ren, P. Dong, D. Zhang, B. Yang and F. Liang, J. Min. Metall. B Metall., 2023, 59, 1–15.
55. Bench Chem, An In-depth Technical Guide to the Hydrolysis of Sodium Stannate Trihydrate in Aqueous Solutions, [Online PDF]. Available at: https://www.benchchem.com/product/b077131#hydrolysis-of-sodium-stannatetrihydrate-in-aqueous-solutions.
56. N. Zulumyan, A. Isahakyan, H. Beglaryan and S. Melikyan, J. Inorg. Organomet. Polym. Mater., 2017, 27, 1323–1332.
57. R. A. El-Salamony, E. Amdeha, A. M. El Shafey and A. M. Al Sabagh, Int. J. Environ. Anal. Chem., 2023, 103, 868–883.
58. P. Deshmukh, J. Bhatt, D. Peshwe and S. Pathak, Trans. Indian Inst. Met., 2012, 65, 63–70.
59. A. F. Michels, P. A. Soave, J. Nardi, P. L. G. Jardim, S. R. Teixeira, D. E. Weibel and F. Horowitz, J. Mater. Sci., 2016, 51, 1316–1323.
60. M. Kurth and P. C. J. Graat, Surf. Interface Anal., 2002, 34, 220–224.
61. M. E. Simonsen, C. Sønderby, Z. Li and E. G. Søgaard, J. Mater. Sci., 2009, 44, 2079–2088.
62. A. U. Alam, M. M. R. Howlader and M. J. Deen, ECS J. Solid State Sci. Technol., 2013, 2, P515–P523.
63. L. Stobinski, B. Lesiak, A. Malolepszy, M. Mazurkiewicz, B. Mierzwa, J. Zemek, P. Jiricek and I. Bieloshapka, J. Electron Spectrosc. Relat. Phenom., 2014, 195, 145–154.
64. A. F. Carles and P. R. Davies, in Interfacial Science (IUPAC Chemistry for the 21st Century), ed. M. W. Roberts, Blackwell Science, Oxford, 1997, p. 77.
65. A. I. Stadnichenko, S. V. Koshcheev and A. I. Boronin, Mosc. Univ. Chem. Bull., 2007, 62, 343–349.
66. D. J. Belton, O. Deschaume and C. C. Perry, FEBS J., 2012, 279, 1710–1720.
67. M. Matinfar and J. A. Nychka, Adv. Colloid Interface Sci., 2023, 322, 103036.
68. M. S. Kronka, F. L. Silva, A. S. Martins, M. O. Almeida, K. M. Honório and M. R. V. Lanza, Mater. Adv., 2020, 1, 1318–1329.
69. S. Zhang, Z. Yue, X. Pang, M. Pan, J. Tang, X. Cheng, J. Li, Y. Liu and W. Shen, Desalination Water Treat., 2020, 179, 387–395.
70. P. Sun, J. Qiu, J. Wu, D. Wu, R. Wang, X. Yan, Y. Wan and X. Wu, Carbon Energy, 2025, 7, e70042.
71. A. Raveendran, M. Chandran and R. Dhanusuraman, RSC Adv., 2023, 13, 3843–3876.
72. W. C. Schumb, Ind. Eng. Chem., 1949, 41, 992–1003.
73. M. Bauer, S. Beratz, K. Ruhland, S. Horn and J. Moosburger-Will, Appl. Surf. Sci., 2020, 506, 144947.
74. X. Chen, C. Zhang, G.-L. Song, D. Zheng, Y. Guo and X. Huang, Materials, 2021, 14, 1758.
75. X. Liu, C. Yang and Y. Lu, Appl. Surf. Sci., 2012, 258, 4268–4275.
76. Y. Yi, G. Weinberg, M. Prenzel, M. Greiner, S. Heumann, S. Becker and R. Schlögl, Catal. Today, 2017, 295, 32–40.
77. C. Qiu, L. Jiang, Y. Gao and L. Sheng, Mater. Des., 2023, 230, 111952.
78. C. U. Pittman, W. Jiang, Z. R. Yue, S. Gardner, L. Wang, H. Toghiani and C. A. Leon y Leon, Carbon Hydrogen, 1999, 37, 1797–1807.
79. M. Zhu, Z. Ge, Z. Xu, X. Liang, L. Yan, Y. Sun, Y. Zhong and Y. Hu, Mater. Sci. Eng. R Rep., 2026, 168, 101168.
80. Z. Liu, Z. Zhao, Y. Wang, S. Dou, D. Yan, D. Liu, Z. Xia and S. Wang, Adv. Mater., 2017, 29, 1606207.
81. M. Jerigová, M. Odziomek and N. López-Salas, ACS Omega, 2022, 7, 11544–11554.
82. X. Lu, W.-L. Yim, B. H. R. Suryanto and C. Zhao, J. Am. Chem. Soc., 2015, 137, 2901–2907.
83. U. Javed, M. Tebyetekerwa, C. Tang, X. Zeng, Z. Wang, K. Sun, J. Yang, I. Marriam, L. Guo, X. Sun, A. K. Sahu, Y. Zhang, A. Zamyadi, A. Du, Q. Li, T. E. Rufford and X. Zhang, Adv. Mater., 2025, 37, 2500834.
84. C. Ling, D. Jin, R. Li, C. Li and W. Wang, Chem. Eng. J., 2023, 465, 142903.

---