High-quality hydrogen peroxide production through electrochemical reduction of anthraquinone in ambient environment

Abstract

At present, the traditional anthraquinone method is the main method for producing high-quality and high-concentration H₂O₂, but it involves the use of H₂, resulting in high carbon emissions and certain risks. The double-electron oxygen reduction reaction has become a more environmentally friendly alternative method for producing H₂O₂. However, at high overpotentials, the production and conversion rates of H₂O₂ are still limited. Here, we report an electrochemical method that can convert H₂O and formaldehyde into high-concentration H₂O₂ and high-value formate salts with high product purity under environmental conditions. At a current density of 30 milliamperes per square centimeter, continuous operation for 5 hours can produce 6.522% (m/v) H₂O₂, and the product does not contain acid, alkali, or excessive impurities. Faraday efficiency reaches 83%. The chronopotentiometry method shows stable current performance of the electrode after 360 hours of operation. This low-carbon, safe, and sustainable approach provides a promising alternative to traditional H₂O₂ production.

An electrochemical process for producing high-purity hydrogen peroxide has been proposed.

INTRODUCTION

Hydrogen peroxide (H₂O₂) is a fundamental chemical widely used in industry and households (1–3). The traditional anthraquinone process is currently the most mature method for producing high-quality H₂O₂, which involves thermal catalytic hydrogenation followed by extraction, purification, and concentration (4). However, this method has notable drawbacks such as high energy consumption and operation under high temperature and pressure, and the use of hydrogen can easily cause explosions (5). These challenges have sparked a demand for developing cost-effective and sustainable alternatives (6). The most widely studied method for producing H₂O₂ through the two-electron oxygen reduction reaction involves a competitive reaction called four-electron ORR, which reduces the selectivity of H₂O₂ (7). Therefore, there is an urgent need for a method that can retain the advantages of traditional anthraquinone methods while addressing their high carbon and hazardous issues.

In the traditional anthraquinone process, anthraquinone hydrogenation is the most critical step (8). It is also the step with the highest energy consumption and carbon emissions (9). To address this issue, we hope to replace the hydrogenation step in traditional anthraquinone methods with low-carbon and safe electrochemical methods (10). By using the synergistic contribution of the anode and cathode to H* generation, the reaction efficiency of the system has been improved. This study proposes a process for self-sufficient production of H₂O₂ through electrocatalytic hydrogenation of anthraquinone (Fig. 1). This process uses active hydrogen to hydrogenate unsaturated substrates, H₂O is the reductive hydrogen source in the cathodic electrolysis cell to generate H*, and HCHO is the oxidative hydrogen source in the anodic electrolysis cell to generate H*, achieving simultaneous in situ production of H₂O₂ at both the anode and cathode (11, 12). Anthraquinone was considered an optimal redox center for H₂O₂ synthesis because anthrahydroquinone, its reducing form, can be directly oxidized by O₂ to produce H₂O₂, and the process is thermodynamically spontaneous (13–15)

$${\mathrm{H}}_2\mathrm{O}+{\mathrm{e}}^{-}\to {\mathrm{H}}^{*}+{\text{OH}}^{-}$$ (1)

$$\text{HCHO}+2{\text{OH}}^{-}\to {\text{HCOO}}^{-}+{\mathrm{H}}_2\mathrm{O}+{\mathrm{H}}^{*}+{\mathrm{e}}^{-}$$ (2)

$$\text{HCHO}+{\text{OH}}^{-}\to {\text{HCOO}}^{-}+2{\mathrm{H}}^{*}$$ (3)

Fig. 1. Schematic illustration of the H₂O₂ production pathway.

Traditional anthraquinone processes use trioctyl phosphate (TOP) and solvent naphtha (SN) as organic extraction phases, but there are related drawbacks such as high volatility, toxicity, and flammability (16). Separation and purification of H₂O₂ from organic extraction phases usually involve procedures like distillation and reverse osmosis, both of which are costly and demand substantial energy input (17, 18). To further replace the traditional anthraquinone method with the electrochemical method and achieve high-quality production of H₂O₂, the selection of the organic extraction phase is particularly crucial. Ideal organic extraction can effectively promote the diffusion of H*, which has maximum solubility for both anthraquinone and anthrahydroquinone while exhibiting low solubility for H₂O₂ (19, 20). It should be immiscible with water and have a substantially different density from water for separation and enrichment of H₂O₂ through water extraction (21). In addition, the organic extraction phases should demonstrate excellent chemical stability to ensure operational safety (22–24). Ionic liquids (ILs) are considered as “design solvents” with adjustable physical and chemical properties and are potential candidates for ideal electrolytes in electrocatalytic hydrogenation (25–27). In this work, the selection of ILs was studied through computational simulations and experimental data screening. [BMIM][PF₆] has been identified as the best choice.

To avoid interference from electrolytes on the product, a four-compartment electrochemical device was designed and built in this work using a Pd/Pd_NP membrane cathode and anode (Fig. 2) to realize dual hydrogenation of anthraquinone. This device comprises an anodic electrolysis tank (AE), a cathodic electrolysis tank (CE), an organic anodic reduction tank (OAR), and an organic cathodic reduction tank (OCR). Water is the hydrogen source for CE. HCHO (0.6 M) in 1.0 M KOH is the hydrogen source for low voltage oxidation in AE. 2,6-Disodium anthraquinone disulfonate (AQDSNa₂) was used as the redox center for electrochemical hydrogenation because of its great solubility in [BMIM][PF₆]. [BMIM][PF₆] with AQDSNa₂ was filled in both OCR and OAR. The H* generated from CE and AE would migrate through the Pd/Pd_NP membrane cathode and anode entering OCR and OAR, where AQDSNa₂ is reduced to AQDSH₂²⁻. After hydrogenation reaction, the AQDSH₂²⁻-[BMIM][PF₆] mixture was transferred to an L-shaped oxidation-extraction tank. Water was introduced to the tank from the bottom for extraction. The volume of H₂O is one-fifth of the volume of [BMIM][PF₆]. Oxygen was bubbled in to oxidize AQDSH₂²⁻ to form H₂O₂, which was then extracted by water. The H₂O₂ aqueous solution was collected from the top of the reactor. Without using much energy, this process is more cost-effective and greener compared with distillation or reverse osmosis used in traditional H₂O₂ manufacturing. After oxidation, the [BMIM][PF₆]-AQDSNa₂ mixture was transferred back to OCR and OAR for the next cycle of use. High-quality H₂O₂ was produced at ambient temperature and pressure using this method.

Fig. 2. Schematics of the 4-H electrocatalytic cell and H₂O₂ production through oxidation and subsequent manual extraction.

RESULTS

A linear sweep voltammogram (LSV) was obtained after several cathodic scans between 0.0 and −1.0 V versus reversible hydrogen electrode (RHE) until the Pd membrane was saturated with hydrogen. Pd was active toward H₂O reduction, and a current density of 194.46 mA cm⁻² could be reached at −1.0 V versus RHE (Fig. 3A). Monitored by gas chromatography, H₂ produced on the cathode increased linearly throughout the electrolysis when constant currents of 10, 20, and 30 mA were applied. The quantities agreed with the theoretical values and confirmed that the Faraday efficiency (FE) of hydrogen evolution on the Pd membrane cathode was close to 100% (fig. S3). The calculated FEs of H₂ production were all less than 30% on a Pd membrane anode. Such a marked difference in H₂ production efficiency implies that effective hydrogen oxidation (H* → H⁺ + e⁻) occurs on Pd when it acts as an anode (1). When using the Pd membrane as an anode, a rapid increase in positive current was observed when the scanning potential was more positive than 0.25 V (versus RHE) in a KOH solution. Further positive scanning resulted in O₂ bubbles on the surface of the anode. When adding HCHO in AE, an immediate current increase was observed because of HCHO oxidation. The LSV showed that HCHO oxidation occurred earlier than the OER reaction, which helped avoid anode side reactions. Increasing the KOH concentration from 0.1 to 1.0 M resulted in a steady increase in anode current (fig. S4). The cation effect of the supporting electrolyte was also probed: K⁺ exhibited the highest anodic current followed by Li⁺ and Na⁺ (fig. S5). Unless otherwise specified, 1.0 M KOH is used as the optimal supporting electrolyte for all subsequent experiments. The anodic current was highly dependent on the HCHO concentration with 0.6 M HCHO producing the highest current density of ~200 mA cm⁻² at 0.5 V versus RHE (fig. S6). Further increasing the HCHO concentration led to a decrease in the anodic current, likely due to the disproportionation of HCHO at high-concentration alkaline media. To assess the impact of the disproportionation products on the oxidation of HCHO, chronoamperometry experiments were conducted with addition of 0.6 M HCHO followed by 0.1 M HCOOH or 0.1 M CH₃OH. The addition of CH₃OH and HCOOH did not result in any change in the observed current, suggesting that CH₃OH and HCOOH would not be oxidized on the Pd membrane electrode at this applied voltage and did not suppress the oxidation of HCHO either (fig. S7). Compared with single-side hydrogenation with water oxidation on a Pd membrane anode or a nickel anode at the same current of 10 mA, HCHO oxidation on a Pd membrane electrode not only produced H* but also saved around 1.0- or 0.375-V voltage input to reach the same current (fig. S8). Free radical analysis by electron spin resonance (ESR) with 5,5-dimethyl-1-pyrroline N-oxide indicates that H* can be detected on both sides of the cathode and anode, confirming the migration of H* across the Pd/Pd_NP membrane electrode from the electrolysis cell to the hydrogenation cell (Fig. 3B). Using [BMIM][PF₆] in a hydrogenation chamber is critical for H* diffusion. H* was not detected in either of the reaction chambers (OCR and OAR) when [BMIM][PF₆] was replaced with a traditional anthraquinone cosolvent (SN and TOP).

Fig. 3. 4-H electrocatalytic dual hydrogenation.

(A) LSVs of a blank Pd membrane electrode as the anode in 1.0 M KOH in the absence (OER) and presence (FOR) of 0.6 M HCHO and the LSV of a H-saturated Pd membrane as the cathode in H₂O (hydrogen evolution) using a H-cell in a three-electrode configuration, in which the Pd membrane anode (1.0 M KOH), Pd membrane cathode, and Ag/AgCl were the working, counter, and reference electrodes, respectively. (B) ESR spectra of 5,5-dimethyl-1-pyrroline N-oxide spin capture electrocatalytic H* in SN, TOP, OAR, AE, CE, and OCR. a.u., arbitrary units. (C) Analysis of visible spectra of AQDSNa₂ and AQDSH₂Na₂ in the OCR chamber using UV-vis spectrophotometry. h, hours. (D) Amount of H₂O₂ produced over time during electrolysis under an applied current density of 30 mA cm⁻² and a 1:5 volume ratio of H₂O and [BIIM][PF₆]. (E) Chronopotentiometry test of the Pd/Pd_NP membrane electrode. The Pd/Pd_NP membrane electrode maintains its catalytic activity for 360 hours without any degradation. (F) Comparison of H₂O₂ yield, electrode stability, and FE at 30 mA cm⁻² between this work and the previously reported electrocatalytic technology.

The AQDSNa₂ absorbance in the cathode side hydrogenation chamber (OCR) decreased gradually with the elapsing of electrolysis time, and the absorbance of AQDSH₂²⁻ increased in the meantime (Fig. 3C), suggesting H* hydrogenation of AQDS²⁻ to AQDSH₂²⁻_. After 5 hours of electrolysis, the absorbance of AQDSNa₂ at the cathode hydrogenation chamber decreased from 3.14 to 1.31, and the absorbance of AQDSH₂²⁻ increased from 0.19 to 1.53. In the anode side, as expected, AQDSH₂²⁻ resulting from H* hydrogenation was detected in the reaction chamber (OAR) too. The decrease in concentration of AQDSNa₂ indicates the hydrogenation of AQDS²⁻ to AQDSH₂⁻² in OCR and OAR (fig. S9). Besides, the sum of AQDSNa₂ and AQDSH₂²⁻ in OCR and OAR remains relatively constant. This electrocatalytic dual-hydrogenation strategy substantially saved voltage input and improved the AQDSH₂²⁻ yield and total FE (total FE = 83%) when compared with single-side cathodic hydrogenation (FE = 59%) with water oxidation on a Pd/Pd_NP membrane anode at the same energy input. After hydrogenation reaction, AQDSH₂²⁻-[BMIM][PF₆] collected from the OCR and OAR was transferred to the oxidation-extraction tank and followed by bubbling of O₂ for 30 min at a flow rate of 35 ml min⁻¹ (fig. S10). H₂O₂ is highly miscible with water, so it could be easily extracted and enriched from [BMIM][PF₆] by water. The effect of the H₂O-to-[BMIM][PF₆] ratio on extraction was investigated, and the optimum ratio was 1:5. The recyclability of AQDSNa₂ was analyzed by performing ultraviolet-visible (UV-vis) analysis on the extracted water layer. We observe complete retention of AQDSNa₂ in [BMIM][PF₆], with no transfer of reduced or oxidized quinone species to H₂O (fig. S11). AQDSNa₂ was not detected in CE and the H₂O₂ product (fig. S12). In the ¹H nuclear magnetic resonance (NMR) spectrum, besides the water peak, another peak appeared at 10.94 ppm (parts per million), confirming that we did produce H₂O₂ during the experimental process (fig. S13). In the ¹⁷O NMR spectrum, because of the highly overlapping oxygen peaks of H₂O and H₂O₂, a high-intensity peak appears at 0.397. No presence of other oxidizing species was detected. In summary, we have great confidence in the purity and yield of H₂O₂ production. The H₂O₂ concentration in the aqueous phase was quantified using cerium sulfate spectrophotometry (fig. S14) and iodometry (fig. S15) for cross-validation. Under the optimum electrocatalytic condition, after electrolysis for 5 hours, the cumulative concentration of H₂O₂ produced from the cathode side reached 6.522% (m/v), and the cumulative concentration of H₂O₂ produced from the anode side reached 2.6% (m/v) (Fig. 3D). Although H₂O₂ can also be produced when the anode and cathode operate separately (fig. S16), the combined hydrogenation of the same substrate can achieve maximum yield benefits. Cyclic electrolysis measurements were performed to evaluate the stability of H₂O₂ production. In eight electrolysis test cycles, the yields of H₂O₂ in OCR and OAR were 6.079 and 2.23%, respectively. The yield remains basically stable (fig. S17A). Compared with the IL before reaction, the ¹H NMR spectrum of the IL after eight cycles exhibits no notable chemical shifts. Neither the positions of the characteristic peaks nor the integral ratios show obvious differences, indicating that the chemical structure of the IL remains stable (fig. S17B). The recovery rate of AQDSNa₂ was found to be above 90%, which indicates that anthraquinone can be recycled (fig. S17C). The Pd/Pd_NP membrane electrode and [BMIM][PF₆] performed stably over the 360-hour chronopotentiometry scan (Fig. 3E). [BMIM][PF₆] before and after 360 hours of electrolysis was sampled for Fourier infrared spectroscopy. After 360 hours of electrolysis, the chemical bonds of [BMIM][PF₆] remain completely unchanged, which indicates that the properties of [BMIM][PF₆] under electrolysis conditions are stable and can be reused (fig. S18). We conducted chronoamperometry tests at a fixed potential. The results, presented in fig. S19, confirm that our electrode maintains stable performance for more than 300 hours, thereby strongly supporting its durability in practically relevant operating scenarios. The H₂O₂ yield reached 430.7 mM hour⁻¹. This work showed superior H₂O₂ yield (Fig. 3F) and great stability. Compared with commercial H₂O₂ (figs. S20 and S21), the H₂O₂ produced through electrochemical hydrogenation (0.448 mg liter⁻¹ K⁺ and 2.064 mg liter⁻¹ Na⁺) is purer (fig. S22 and tables S1 and S2).

Anthraquinone was considered an optimal redox center for H₂O₂ synthesis (Fig. 4A). An electrolyte plays a critical role not only in electrocatalytic hydrogenation but also in subsequent H₂O₂ separation and enrichment. ILs were considered because of their effectiveness in facilitating H* diffusion, as well as their ability to dissolve AQDSNa₂. Eighteen ILs were evaluated in this work (table S3).

Fig. 4. Interactions between AQDSNa₂ and ILs.

(A) Hydrogenation and oxidation process of AQDSNa₂. (B) Schematic comparison of polarity, donor number, and binding energy for AQDSNa₂ and ILs. (C) UV-vis spectra of AQDSNa₂ in [BMIM][PF₆], [BMIM][Sb₆], and [BMIM][FSI]. (D) The solubility of three ILs in water varies with time.

The solubility of AQDSNa₂ in the IL can be related to the solvent’s polarity and electron donor ability. Interaction energy (E_T ) values are widely used to measure solvent polarity on the basis of the intramolecular charge transfer p-p^* absorption of Reichardt’s dye in different solvents. In general, strongly polar molecules have larger E_T values, and vice versa. As shown in Fig. 4B, [BMIM][PF₆] showed greater E_T values (54.87 kcal mol⁻¹) than [SbF₆⁻]-based ILs and [FSI⁻]-based ILs (table S4). According to “like-dissolve-like,” the polar AQDSNa₂ is more soluble in the [PF₆⁻]-based IL. For the electron donor ability, the number of donors was determined using NMR measurements. The [PF₆⁻]-based IL showed higher donor values (3.96 kcal mol⁻¹) than [FSI⁻]- and [TFSI⁻]-based ILs, suggesting its stronger interaction with AQDSNa₂ and a higher solubility of AQDSNa₂ in [BMIM][PF₆]. To reveal the dissolution behavior of AQDSNa₂ in ILs, the interaction energies between ILs and AQDSNa₂ were simulated as well. The electrostatic potentials of [BMIM][FSI], [BMIM][PF₆], [BMIM][SbF₆], H₂O₂, and AQDSNa₂, together with the binding energy and charge transfer between abovementioned ILs and H₂O₂ or AQDSNa₂, were calculated using the DMol3 module in Materials Studio (MS) 2020 (fig. S23; specific calculation in the Supplementary Materials). AQDSNa₂ tends to dissolve in ILs with higher binding energy. Of the three ILs, [BMIM][FSI] and [BMIM][PF₆] showed minimum and maximum E values of 1.56 and 1.90 kJ mol⁻¹, respectively, suggesting that AQDSNa₂ had the lowest solubility in [BMIM][FSI] and the highest solubility in [BMIM][PF₆]. The solubility of AQDSNa₂ in different ILs was determined by a UV-vis spectrophotometer. Absorption in the range of 250 to 350 nm is a characteristic peak for AQDSNa₂. The peak intensity of AQDSNa₂ in the [PF₆⁻]-based IL was higher than those in the [FSI⁻]- and [SBF₆⁻]-based ILs. The calculated charge-transfer results (method S5) agreed well with the experimental data (Fig. 4C). The solubility of ILs in water was evaluated by measuring total organic carbon (TOC) in water. Despite the fact that TOC content in the aqueous extraction of all three ILs increased over time (Fig. 4D), [BMIM][Pf₆] exhibited the lowest water solubility (97.9 mg liter⁻¹). To obtain high-purity H₂O₂, we suggest completing the extraction process within 15 min.

To increase the number of active sites on the electrode surface (28), Pd nanoparticles were electrodeposited on one side of the Pd membrane (Fig. 5, A and B) to produce a Pd/Pd_NP membrane electrode (see method S1 for details). Powder x-ray diffraction (XRD) results revealed that this Pd/Pd_NP membrane electrode retained the same crystal phase structure as the original Pd membrane electrode (Fig. 5C). The original Pd membrane electrode exhibits characteristic XRD peaks at 40.1°, 46.5°, and 68.0°, corresponding to the (111), (200), and (220) crystal planes of the Pd phase crystal structure, respectively (JCPDS card number 46-1043). The lattice parameters of the original Pd/Pd_NP membrane electrode were calculated to be 3.890 Å, which closely matches the literature value of 3.889 Å. The Pd lattice expands after absorbing hydrogen atoms (14). A new β-PdH_x phase was formed after the absorption of hydrogen atoms with a lattice parameter of 4.054 Å, an increase of 4.21% relative to the Pd/Pd_NP membrane electrode (fig. S24).

Fig. 5. Deposition of the Pd catalyst to enhance the active sites on the electrode surface.

(A) Schematic diagram of an electrochemical reactor without a deposited Pd catalyst. (B) Schematic diagram of an electrochemical reactor with a deposited Pd catalyst. (C) Powder XRD patterns of the Pd membrane electrode, electrodeposited Pd nanoparticles on pristine Pd membrane (Pd/Pd_NP), a Pd/Pd_NP membrane cathode, and Pd/Pd_NP membrane anode after electrolysis at 30 mA cm⁻² and standard pattern of the cubic phase crystal structure of Pd (bottom). (D) SEM images of pristine Pd membrane anode, Pd membrane anode after saturated hydrogen absorption, Pd membrane cathode, and Pd membrane cathode after saturated hydrogen absorption. (E) SEM images of the Pd/Pd_NP membrane anode, Pd/Pd_NP membrane cathode (top), and high-resolution TEM image of Pd (bottom). (F) Current densities at 0.96 V versus RHE from the cyclic voltammograms collected on pristine Pd or Pd/Pd_NP in 1.0 M KOH at various scan rates.

The morphology of the prepared Pd/Pd_NP membrane electrode was observed by scanning electron microscopy (SEM) equipped with energy-dispersive x-ray spectroscopy (fig. S25) and transmission electron microscopy (TEM), and the fine structure of the crystal was observed by high-resolution TEM. SEM and TEM images show that Pd nanoparticles and nanosheets are uniformly covered on the surface of the Pd membrane and Pd crystals grow on the membrane surface like flower buds (Fig. 5D). Pd nanoparticles were uniformly distributed on the electrode surface (Fig. 5E). Compared to the original Pd membrane electrode, the electrochemical double-layer capacitance of the Pd/Pd_NP membrane electrode increased by about 40 times (20.72 μF cm⁻² versus 0.52 μF cm⁻²) (Fig. 5F and fig. S26). Operating under the same conditions, the total amount of H₂O₂ produced by the Pd/Pd_NP membrane cathode is ~1.65 times of that produced by the Pd membrane cathode (fig. S27).

We calculated carbon emissions of the H₂O₂ production processes for both the traditional anthraquinone method and electrocatalytic anthraquinone processes. The H₂ used for the traditional method typically originates from steam methane reforming, a high-temperature and capital-intensive process that is difficult to scale down (29). This energy is primarily provided by fossil fuel combustion, leaving large carbon footprint even before H₂O₂ production (30). In China, more than 2.8 million tonnes of carbon footprint was generated in H₂O₂ production in 2015 (31). For every 1 kg of H₂O₂ production (equivalent to 100% concentration), this method could reduce carbon emissions from 3.917 kg by the traditional anthraquinone method (32, 33) to 1.129 kg (method S7). We purified the anode product to obtain high-purity potassium formate by-product (fig. S28). We analyzed the purified product. The ¹H NMR spectrum only shows a single peak at 8.85 ppm (fig. S29), attributed to formate ions. Formate, a high-value by-product obtained in this process (Fig. 6A and figs. S30 and S31), can reduce raw material costs (Fig. 6B).

Fig. 6. Additional advantages of electrocatalytic dual hydrogenation.

(A) Quantitative analysis of products after purification in an anode electrolysis tank. (B) Comparison of carbon emissions for a 1-kg H₂O₂ yield and raw material cost with a 1-tonne H₂O₂ yield between the traditional anthraquinone process (TRA) and this work (EC). (C) Local sensitivity analysis for single-parameter scenarios in the electrocatalytic method. (D) Local sensitivity analysis for dual-factor (Pd and AQDSNa₂) synergistic scenarios in the electrocatalytic method. (E) Local sensitivity analysis for dual-factor ([BMIM][PF₆] and AQDSNa₂) synergistic scenarios in the electrocatalytic method. (F) Local sensitivity analysis for dual-factor ([BMIM][PF₆] and Pd) synergistic scenarios in the electrocatalytic method.

In the local sensitivity analysis, the impacts on output when the cycle numbers of three key materials decrease individually are evaluated. The reduction in [BMIM][PF₆] cycle numbers has the most pronounced effect on output; as its cycles decrease from 1800 to 1000, the average output decline reaches 12.88%, with a maximum decrease of 31.86%, demonstrating a clear negative response trend (Fig. 6C). This study further explores the combined impact on output when the cycle numbers of key materials decline simultaneously, aiming to examine potential synergistic sensitivity effects. Scenario 10 exhibits a decline, with an average decrease of 17.57% (Fig. 6D). Scenario 11 shows a mean output reduction of 12.93%, indicating that while anthraquinone contributes to the synergy, its sensitivity is relatively moderate (Fig. 6E). Scenario 12 demonstrates the most minor change, with an average output drop of only 4.75% (Fig. 6F). In general, the impact of dual-factor synergistic scenarios on output far exceeds the linear superposition of single-factor changes (fig. S32 and tables S5 to S10).

DISCUSSION

We demonstrated in this work green production of H₂O₂ through electrocatalytic dual anthraquinone hydrogenation followed by O₂ oxidation and water extraction. AQDSNa₂ in [BMIM][PF₆] serves as the redox center to produce H₂O₂. H* was produced through cathodic water reduction and low-potential anodic HCOH oxidation by using Pd/Pd_NP membrane electrodes, which allow H* diffusing from the electrolysis cell to the spatially separated adjacent hydrogenation cells to hydrogenate AQDSNa₂. The hydrogenation product, AQDSH₂²⁻, was oxidized by continuously bubbling in O₂, and the formed H₂O₂ was extracted from [BMIM][PF₆] by water. The [BMIM][PF₆]-AQDSNa₂ media were recovered and reused for the next cycle reaction. Dual electrocatalytic hydrogenation not only improves the FE and H₂O₂ yield but also produces formate, a high-value by-product. The use of IL electrolyte, two separate hydrogenation cells, and water extraction simplifies the process, lowers the operating cost, and eliminates the potential environmental and safety hazardous issues associated with traditional manufacturing. Compared with other reported results, the electrocatalytic method exhibits superior performance in terms of yield and stability. The electrodes used in this work demonstrated stable performance for at least 360 hours. With further optimization, electrocatalytic dual hydrogenation of anthraquinone for H₂O₂ production has great potential in large-scale applications.

MATERIALS AND METHODS

Materials and chemicals

AQDSNa₂ was obtained from Adamas and used as received. Formaldehyde (38%), formic acid, methanol, and potassium hydroxide (85%) were obtained from Aladdin and used as received. The Pd membrane electrode was obtained from Wuhan battery supplies (6 mm thickness), and Ag/AgCl reference electrodes were obtained from Wuhan Gaoshi Ruilian Technology Co., Ltd. The anion exchange membrane was manufactured by Jiangsu Helper Functional Materials, purchased from the Taobao Store, and soaked in aqueous 1.0 M NaCl when not in use. All of the ILs were purchased from Ningxia Zhuoyu New Material Technology Co., Ltd.

Catalyst preparation

Pd/Pd_NP membrane electrodes were prepared by an electrodeposition method using 15.9 mM PdCl₂ and 1.0 M HCl as the electrolyte. Specifically, a Pd membrane electrode (1.16 m²), as the working electrode, was clamped between an electrochemical chamber and a chemical chamber, and a Pt electrode and a Ag/AgCl (saturated KCl) electrode were used as the counter and reference electrodes, respectively, in the electrochemical chamber. Electrodeposition was performed at a constant potential of −0.2 V versus Ag/AgCl until 16.75 C was passed.

Characterization and analytical methods

TOC was measured using a Shimadzu TOC analyzer (TOC-VCPH, Shimadzu, Japan). The morphology and composition of samples were characterized by field emission SEM (Phenom Pro X, Shanghai, China) coupled with an energy-dispersive x-ray system. ESR analysis was performed using a Bruker ESR 5000 spectrometer. The preparation process of the activator was traced by simultaneous differential scanning calorimetry/thermogravimetric analysis (SDT/SDT-Q600, US). Powder XRD patterns were obtained by using a Lab X/6100 x-ray diffractometer (Shimadzu, Japan), equipped with a Cu Kα x-ray source (λ = 0.15406 nm) in the scanning range of 2θ = 10° to 80° with a scanning speed of 5° min⁻¹. The morphology of the samples was characterized by SEM (Zeiss Merlin Compact, Germany).

Electrochemical tests

All electrochemical measurements were conducted on a Metrohm PSTAT30 (Vantone, Switzerland). All reported potentials were referenced to the RHE by calibration using Pt as the working electrode in H₂-saturated 1.0 M KOH. The potentials versus Ag/AgCl were converted to the RHE using the Nernst equation: E_RHE = 𝐸_Ag/AgCl + 0.198 + 0.059 × pH (pH = 14 for 1.0 M KOH).

For the electrocatalytic oxidation of formaldehyde on the Pd membrane electrode, a H-compartment electrochemical cell separated by an anion exchange membrane was used with a Pd membrane anode (1.0 M KOH), Pd membrane cathode, and Ag/AgCl as the working, counter, and reference electrodes, respectively. The Pd membrane anode working electrode and Ag/AgCl (1.0 M KOH) reference electrode were placed in the anodic chamber, while the Pd membrane cathode counter electrode was placed in the cathodic chamber. All linear sweep voltammetry experiments were conducted using a three-electrode configuration at a scan rate of 10 mV s⁻¹.

The electrochemically active surface area was measured by the double-layer capacitance method. In the non-Faraday interval, the cyclic voltammetry tests were implemented at the scan rates of 10, 20, 30, 40, 50, 60, and 80 mV s⁻¹ in 1.0 M N₂-saturated KOH to exclude the effect of trace oxygen in the electrolyte. Δj was plotted as a function of scan rate, the slope of which is the double-layer capacitance (C_dl). A Pd or Pd/Pd_NP membrane anode, Pd or Pd/Pd_NP membrane cathode, and Ag/AgCl were used as the working, counter, and reference electrodes, respectively.

For electrocatalytic H₂ production, a H-compartment assembly was used with two Pd membrane electrodes (area, 1.16 cm²) as the cathode and anode, respectively. LSVs were collected at a scan rate of 10 mV s⁻¹, and chronopotentiometry experiments were performed at 10, 20, and 30 mA.

Stability tests (chronoamperometry and chronopotentiometry) were operated with a O₂-saturated solution in a 4-H electrolyzer. The resistance measurement was performed in the frequency range from 0.1 Hz to 100,000 Hz. Tests were performed with the Pd/Pd_NP membrane anode (area, 1.16 cm²) in 1.0 M KOH, Pd/Pd_NP membrane cathode (area, 1.16 cm²), and Ag/AgCl as the working, counter, and reference electrodes, respectively.

For the electrocatalytic hydrogenation of AQDSNa₂, a four-compartment electrochemical device was used with two Pd membrane electrodes or two Pd/Pd_NP membrane electrodes (area, 1.16 cm²) as the cathode and anode, respectively. The Pd nanoparticle side of each Pd/Pd_NP membrane electrode faced the chemical chamber (OCR and OAR) for the hydrogenation reactions. The electrolyte in the cathodic chamber was H₂O with Na₂SO₄ (conductivity, 1.1 S m⁻¹), while the anodic chamber contained 0.6 M HCHO in 1.0 M KOH; 50 mM AQDSNa₂ in [BMIM][PF₆] was added to the hydrogenation chambers on both sides of the 4-H electrochemical device. Chronopotentiometry experiments were carried out at 30 mA cm⁻². The final product of AE (HCOOK) was obtained as a white solid after water evaporation.

Cyclic experiment

Before electrolysis, [BMIM][PF₆] is pretreated. The specific method refers to feeding [BMIM][PF₆] and deionized water at a mass ratio of 1:3, 60 ± 5°C, and a speed of 800 to 1200 rpm for 2 hours. The conductivity monitors show that the conductivity of the IL is stable at <0.37 S m⁻¹. The separated IL is used for subsequent electrolysis reactions. Then, the Pd/Pd_NP membrane is subjected to pretreatment by immersing one side of the electrode in the treated IL for 2 hours to attach a layer of IL film to the electrode surface. This process aims to form a stable prewetting film of [BMIM][PF₆] on the electrode surface, optimize the electrode/electrolyte interface contact, and form a stable contact interface. The processed IL (nitrogen saturation) and AQDSNa₂ are placed in OCR and OAR. After electrolysis is completed, we transfer [BMIM][PF₆] to the oxidation reaction unit and introduce oxygen under the liquid surface. Subsequently, [BMIM][PF₆] is transferred to the extraction separation unit, and an aqueous solution is added to the extraction unit according to different ratios. It is stirred at 1000 rpm for 1 hour and then left to stand for 2 hours.

H₂O₂ quantification

A titration UV-vis spectrophotometer was used for quantitative determination of H₂O₂ concentration. A standard Ce(SO₄)₂ solution was prepared by dissolving 0.332 g of Ce(SO₄)₂ in 1 liter of 0.5 M H₂SO₄ solution. The initial color of the Ce(SO₄)₂ solution was yellow by Ce⁴⁺ ions, but H₂O₂ reduces Ce⁴⁺ to Ce³⁺, making it a colorless solution.

Iodometry

For the determination of the standard curve, the amount of H₂O₂ was analyzed by iodometry. To a solution of H₂O₂ (8 ml) with a known concentration, 1 ml of (0.1 M) potassium hydrogen phthalate (C₈H₅KO₄) aqueous solution and 1 ml of (0.4 M) potassium iodide aqueous solution were added. Afterward, the resulting mixture was allowed to stand for 30 min. H₂O₂ reacted with iodide anions (I⁻) under acidic conditions (H₂O₂ + 3I⁻ + 2H⁺ → I³⁻ + 2H₂O), forming triiodide anions (I₃⁻) with a distinct absorption peak at around 350 nm. The quantity of I₃⁻ was determined using UV-vis spectroscopy on the basis of the absorbance at 350 nm, allowing for the estimation of the amount of H₂O₂ produced during each reaction.

High-performance liquid chromatography methods for AQDSNa₂ detection

The concentration of AQDSNa₂ was determined on an Agilent 1260 II C 18 column (4.6 mm by 150 mm, 5 μm). The column temperature was set at 40°C, and elute was set as 0.1 ml min⁻¹. A mobile phase of water with 0.1% phosphate and methanol (30:70) was used for AQDSNa₂ determination in the sample without other probes. Detection was carried out at 248 nm.

H₂ quantification

The evolved H₂ from electrolysis was determined via gas chromatography (SRI 8610C) equipped with a Molecular Sieve 13 packed column, a HayesSep D packed column, and a thermal conductivity detector. The oven was kept at 80°C using Ar as the carrier gas.

Detection of H₂O₂ by NMR

The samples for NMR detection of H₂O₂ were prepared by mixing the aliquot (540 μl) with a solution of 2-(N-morpholino) ethanesulfonic acid (60 μl, 1 mM) in an NMR tube. For the measurement, the temperature was controlled using a cryogenic probe and set at 2°C, with a Gaussian pulse set at 3 ms and 30,720 scans in 30 min. The measurements were conducted on a Bruker Avance III 400 MHz NMR spectrometer.

Detection of HCHO/HCOOH by NMR

The identification and quantification of methanol and formic acid were determined by ¹H NMR from the calibration curves by applying standard solutions with known concentrations of commercially purchased compounds with t-butanol (10.0 mM) as an internal standard. One hundred microliters of electrolyte taken from the electrochemical cell was acidified with 20 μl of concentrated HCl (37%), followed by the addition of 500 μl of D₂O. ¹H NMR spectra were recorded on a Bruker AV 400 MHz spectrometer using a water suppression method.

UV-vis spectroscopy of AQDSNa₂

UV-vis spectra were collected on a Varian Cary 50 UV-vis spectrometer, with manual baseline correction. Quartz cells with a 1-cm path length and capped with septa were used for all measurements. For this experiment, 50 mM AQDSNa₂ in [BMIM][PF₆] was fully reduced via bulk electrolysis at 30 mA cm⁻² in the four-compartment electrochemical device described below. The spectrum of AQDSH₂Na₂ was recorded every 60 min, and then AQDSH₂Na₂ was extracted into an oxidation cell and permitted slow oxidation. The flow rate of O₂ is 35 ml min⁻¹. Full oxidation was achieved within 30 min. The reaction of AQDSH₂²⁻ with O₂ was followed by UV-vis spectroscopy.

Quantification of the high-value product

During the oxidation reaction of the anode electrolytic cell, HCHO was continuously consumed. After 5.6 hours of electrolysis, 0.6 M HCHO was almost completely consumed. To further evaluate the economic benefits of the electrolysis process, electrolysis was continuously performed for 120 hours, and 0.6 M HCHO was added every 5.6 hours. The anode electrolyte was collected after electrolysis. Then, slowly and in batches, formic acid (HCOOH) was added to the mixed solution, and stirring and cooling were continued. The neutralized solution was heated and concentrated on a rotary evaporator until a large amount of crystals precipitated. It was cooled to room temperature or in an ice water bath to promote complete crystallization. It was dried for several hours at 60° to 80°C.

Dynamics simulations

The electrostatic potentials of [BMIM][FSI], [BMIM][PF₆], [BMIM][SbF₆], H₂O₂, and AQDSNa₂ as well as the binding energy and charge transfer between the IL above and H₂O₂ or AQDSNa₂ were respectively calculated by the DMol3 module in Materials Studio (MS) 2020.

Supplementary Materials

This PDF file includes:

Supplementary Methods

Figs. S1 to S32

Tables S1 to S10

References