Direct electrosynthesis of pure aqueous H2O2 solutions up to 20% by weight using a solid electrolyte

See allHide authors and affiliations

Science  11 Oct 2019:
Vol. 366, Issue 6462, pp. 226-231
DOI: 10.1126/science.aay1844

A direct route to pure peroxide

Despite the widespread use of hydrogen peroxide as an oxidant and disinfectant, its commercial synthesis still requires inefficient concentration and purification steps. Xia et al. now report an electrochemical approach to synthesizing pure peroxide solutions straight from hydrogen and oxygen. Using a solid-state electrolyte, they avoid contamination of the product solution by extraneous ions. Varying the flow rate of water through the electrochemical cell tunes the final concentration over a range from 0.3% to 20% by weight.

Science, this issue p. 226


Hydrogen peroxide (H2O2) synthesis generally requires substantial postreaction purification. Here, we report a direct electrosynthesis strategy that delivers separate hydrogen (H2) and oxygen (O2) streams to an anode and cathode separated by a porous solid electrolyte, wherein the electrochemically generated H+ and HO2 recombine to form pure aqueous H2O2 solutions. By optimizing a functionalized carbon black catalyst for two-electron oxygen reduction, we achieved >90% selectivity for pure H2O2 at current densities up to 200 milliamperes per square centimeter, which represents an H2O2 productivity of 3.4 millimoles per square centimeter per hour (3660 moles per kilogram of catalyst per hour). A wide range of concentrations of pure H2O2 solutions up to 20 weight % could be obtained by tuning the water flow rate through the solid electrolyte, and the catalyst retained activity and selectivity for 100 hours.

Hydrogen peroxide (H2O2) is a nexus chemical for a variety of industries, currently produced through the indirect, energy-demanding, and waste-intensive anthraquinone process (1, 2). This traditional method usually generates H2O2 mixtures with concentrations of 1 to 2 weight % (wt %), necessitating further costly purifications and distillations to reach concentrations appropriate for commercial use (3). The overall process requires centralized infrastructure and thus relies heavily on transportation and storage of bulk H2O2 solutions, which are unstable and hazardous (4). Direct synthesis of H2O2 from a hydrogen (H2) and oxygen (O2) mixture (Fig. 1A) provides an alternative route for small-scale on-site generation (57). Catalyst development for this reaction has progressed over the past decade (812), exemplified by a palladium-tin catalyst with high selectivity (>95%) and productivity [61 moles per kilogram of catalyst (kgcat) per hour] for H2O2 synthesis (9). However, a drawback of this route is the inherent flammability hazard associated with mixing high-pressure H2 and O2 (13). In practice, the H2 feedstock must be heavily diluted using CO2 or N2 carrier gas, substantially lowering the yields of H2O2. In addition, the use of methanol solvent leads to extra purification costs in the preparation of pure aqueous H2O2 solutions.

Fig. 1 Schematic illustration of the two different H2O2 synthesis methods using H2 and O2.

(A) Synthesis of H2O2 using diluted H2 and O2 under high pressure. Methanol used to improve the solubility of the reacting gases in the medium (44) must then be removed downstream. Other studies that avoid alcohols have been performed in acidic solutions of either HCl or H2SO4, with NaBr or NaCl as promotors (44). (B) Electrosynthesis of H2O2 using pure H2 and O2 streams separately introduced to the anode and cathode, respectively. SE represents a solid electrolyte, which consisted in this study of either functionalized styrene–divinylbenzene copolymer microspheres or inorganic CsxH3-xPW12O40 (see materials and methods). Electrochemically generated cations (H+) and anions (HO2), driven by the electric field, cross in the porous SE layer and recombine to form H2O2. DI water flowing through the porous SE layer then dissolves the H2O2 with no impurities.

By contrast, electrosynthesis of H2O2 can decouple the H2/O2 redox exchange into two half-cell reactions (Eqs. 1 and 2), followed by the ionic recombination process (Eq. 3):

2e O2 reduction reaction (2e-ORR): O2 + H2O + 2e → HO2 + OH(1)H2 oxidation reaction (HOR): H2 → 2H+ + 2e(2)H2O2 formation: HO2 + H+ → H2O2(3)

In the electrochemical process, O2 and H2 can be kept safely separated and introduced in pure form to accelerate the reaction. The synthesis can proceed under ambient conditions for on-site H2O2 generation and could potentially even output electricity. Although there have been selective catalysts such as noble metals or carbon materials developed for the 2e ORR pathway (1418), the H2O2 product has typically been generated in a mixture, with solutes in traditional liquid electrolytes ranging from acidic to alkaline pH. Extra separation processes to recover pure H2O2 solutions were therefore required. Other designs including the use of deionized (DI) water or a polymer electrolyte membrane as the ion-conducting electrolyte have been explored on a preliminary basis for obtaining pure H2O2 solutions, but they generally suffered from low reaction rates, product concentrations, or Faradaic efficiencies (FEs) (supplementary text, note 1) (1921).

Here, we report a porous solid electrolyte design to realize direct electrosynthesis of pure H2O2 solutions. As illustrated in Fig. 1B and figs. S1 and S2, independent H2 and O2 streams were respectively delivered to HOR and 2e-ORR catalysts coating gas diffusion layer (GDL) electrodes. The anode and cathode “sandwiched” the cation exchange membrane (CEM) and anion exchange membrane (AEM) layers (see materials and methods for details) to avoid flooding by direct contact with liquid water. In the center, a thin and porous solid electrolyte layer facilitated ionic recombination of H+ and HO2 ions crossing from the anode and cathode with small ohmic losses; a flowing DI water stream confined to this middle layer could then dissolve the pure H2O2 product with no introduction of ionic impurities. By tuning the HO2 generation rate or the DI water flow rate, a wide range of H2O2 concentrations (from hundreds of parts per million to tens of percent) could be directly obtained with no need for further energy-consuming downstream purification.

To deliver efficient conversion, electrocatalysts with high activity and selectivity for 2e-ORR and HOR are a prerequisite. We chose the state-of-the-art platinum on carbon (Pt/C) catalyst for HOR at the anode, which affords high H2-to-H+ conversion rates at small overpotentials (2224). For the cathode, however, electrocatalysts with high activity and selectivity for 2e-ORR toward H2O2 have been much less thoroughly explored than the extensively studied fuel-cell catalysts for 4e-ORR to H2O. Recent studies on noble metal catalysts such as Au-Pd or Pd-Hg (14, 25), as well as carbon materials such as graphene, carbon nanotubes, or porous carbon (15, 16, 2629), have demonstrated high selectivity toward the 2e pathway. Nevertheless, practical current densities (hundreds of milliamperes per square centimeter) with high FEs, particularly at neutral pH for the purpose of pure H2O2 generation, have not yet been achieved. We chose commercial carbon black as the starting material because of its low cost, its high surface area (fig. S3) for high mass activity, and, especially, its nanoparticulate morphology (fig. S3) to facilitate O2 diffusion from the GDL (layer-by-layer stacking of graphene nanosheets, by contrast, can hinder gas transport). Surface functional groups such as ethers (C–O–C) and carboxylic acids (HO–C=O) have previously been posited to activate the adjacent carbon atomic sites for selective 2e-ORR (15, 16). Hence, we treated the carbon black nanoparticles with nitric acid to introduce such oxidized functionality (see materials and methods and supplementary text, note 2). No morphological changes were observed for these carbon black nanoparticles after acid treatment (fig. S3); however, high-resolution x-ray photoelectron spectroscopy (XPS) (fig. S4) confirmed that acid treatment enriched the particles with oxygen-containing functional groups, including C–O–C/C–OH and HO–C=O, as deconvolved from carbon and oxygen 1s signals.

We found that surface oxidation strongly correlated with H2O2 selectivity and activity (fig. S5). The selectivity rose from <80% for the unoxidized particles to ~95% for even relatively low surface oxygen coverage (2.11%). Although the H2O2 selectivity was similar upon further increasing the surface oxygen coverage from 2.11 to 11.62% as shown in fig. S5B, we found that the 2e-ORR catalytic activity gradually improved (fig. S5C), which we ascribe to the increased concentration of active sites. After optimization, we selected carbon black with ~10% surface oxygen coverage (CB-10%) as the cathode catalyst for further development of the full cell. We first used a standard three-electrode rotation ring-disc electrode (RRDE) system in neutral pH (0.1 M Na2SO4) to evaluate the intrinsic activity of CB-10% for benchmark comparisons (see materials and methods). The catalyst presented an impressive H2O2 generation performance, with a maximal H2O2 selectivity of ~98% and an onset potential of 0.438 V versus reversible hydrogen electrode (RHE), to deliver a 0.1 mA cm−2 H2O2 generation current (fig. S6, A and B). A flow-cell system with GDL electrodes and traditional liquid electrolytes was further used to test the catalyst’s performance without O2 gas diffusion limits in both neutral and alkaline electrolytes (fig. S6C). With a wide potential window to deliver high H2O2 selectivity (>90%) in both neutral and alkaline solutions, the catalyst reached maximal FE of 98 and 99%, respectively (fig. S6D), in good agreement with RRDE tests. Furthermore, H2O2 partial currents of ~300 mA cm−2 were achieved, whereas high FEs were still maintained in neutral solutions, better than the highest O2-to-H2O2 conversion rates yet observed (30).

The porous solid electrolyte layer comprises either an anion or cation solid conductor, which can consist of ion-conducting polymers with different functional groups (31), inorganic compounds (32), or other types of solid electrolyte materials such as ceramics, polymer–ceramic hybrids, or solidified gels (33). Among these different solid conductors, polymer ion conductors have been widely used for electrochemistry applications because of their fast ion conduction at room temperature, high reliability, and ease of processing (34). Because proton conduction is generally faster than anion conduction (35), here, we chose to use styrene-divinylbenzene copolymer microspheres (fig. S7), functionalized with sulfonic acid groups for cation (H+) conduction (36), as a representative solid electrolyte layer for demonstration. Those copolymer microspheres, once packed together in the middle layer, allow for H+ conduction along their interconnected surfaces; in addition, the micrometer pores formed between these stacked spheres allow for DI water flow and product release (fig. S7). We first studied the impact of the solid electrolyte on H2O2 selectivity by the CB-10% catalyst in a standard three-electrode setup (fig. S8A), with potentials calibrated to the RHE scale. The results (fig. S8B) indicate that there were no obvious negative or positive impacts on H2O2 selectivity of the CB-10% catalyst when switching from traditional liquid electrolyte to our solid electrolyte. Next, we systematically investigated the H2O2 production performance of CB-10% using a two-electrode cell with porous solid electrolyte, as shown schematically in Fig. 1B. Figure 2A shows the current-voltage (I-V) curve of a CB-10%//SE//Pt-C cell with O2 and H2 gas streams delivered to the cathode and anode, respectively. The DI water flow rate was fixed at 27 ml hour−1 for this 4-cm2 electrode cell to prevent substantial product accumulation, particularly under large currents. H2O2 was readily detected starting from a cell voltage of –0.54 V, suggesting an early onset considering the equilibrium voltage of –0.76 V (37). The H2O2 selectivity remained >90% across the entire cell voltage range, reaching a maximum of 95% (Fig. 2B). An H2O2-generation current of ~30 mA cm−2 (0.53 mmol cm−2 hour−1) could be obtained under 0 V (no external energy input). Moreover, a potential of only 0.61 V was required to deliver a current density of 200 mA cm−2 with an H2O2 FE of ~90%. This current represents an H2O2-generation rate of 3.4 mmol cm−2 hour−1, or 3660 mol kgcat−1 hour−1 considering both cathode and anode catalyst (see materials and methods; a comparison with literature benchmarks is given in Table 1 and fig. S9). No H2 byproduct (potentially from H2 evolution at large overpotentials) was detected from the cathode side under such a high current density (fig. S10A), indicating exclusive selectivity for ORR. Other types of solid electrolyte with different material properties, including polymeric conductors for HO2 conduction and inorganic CsxH3–xPW12O40 for cation conduction, were also demonstrated to be effective for pure H2O2 solution generation (fig. S11), which suggests the wide tunability and versatility of our solid electrolyte design.

Fig. 2 Direct electrosynthesis of pure H2O2 using H2 and O2 with porous solid electrolyte.

(A) I-V curve of CB-10%//SE//Pt-C cell with an H+-conducting porous solid electrolyte. We define the cell voltage as negative when the cell can output energy during the production of H2O2. The positive cell voltage therefore indicates that energy input is required for the reactor. The cell voltages were iR (current × resistance)compensated (see materials and methods). (B) Corresponding FEs and production rates of H2O2 under different cell voltages. (C) Dependence of H2O2 concentration on the DI water flow rate at an overall current density of 200 mA cm−2. Up to 20 wt % pure H2O2 solutions could be continuously generated for immediate use. The data points in (A) to (C) each represent the mean of two independent measurements. (D) Removal of TOC in Houston rainwater using the H2O2 solution generated at a fixed current density of 200 mA cm−2 and a fixed DI water flow rate of 27 ml hour−1 in our 4-cm2 electrode device. A high rainwater treatment rate of 0.88 liters hour−1 (0.22 liters cm−2electrode hour−1 or 2200 liters m2electrode hour−1) was achieved to meet the drinking water standards (TOC < 2 ppm according to the Texas Commission on Environmental Quality). (E and F) Stability tests for continuous generation of pure H2O2 solutions with concentrations >1000 and 10,000 ppm, respectively. No degradation of cell voltage or H2O2 concentration was observed over the 100-hour continuous operation. The cell currents and DI flow rates were (E) 60 mA and 27 ml hour−1 and (F) 120 mA and 5.4 ml hour−1, respectively.

Table 1 Performance metrics of different H2O2 generation methods

View this table:

Under a fixed DI water flow rate of 27 ml hour−1, the H2O2 concentration from our 4-cm2 electrode cell reached ~1.7 wt % with an overall cell current of 800 mA. By speeding up or slowing down the DI water flow rate while maintaining the H2O2 generation current, we could tune product concentration over a wide range for different application scenarios (Fig. 2C). Up to 20 wt % [200,000 ppm] aqueous H2O2 solutions could be directly and continuously obtained by means of electrochemical synthesis. We noticed that the measured H2O2 selectivity decreased with increased H2O2 concentration (fig. S12A). We ascribe the observed decrease in apparent FE (98% at 0.3 wt % versus 70% at 6.6 wt %) to the following two possible processes. First, the higher concentration of H2O2 product in the solid electrolyte layer could shift the equilibrium of the 2e-ORR while enhancing the selectivity of the competing 4e pathway to H2O product, thereby lowering intrinsic H2O2 selectivity. Second, while H2O2 formation proceeds, a fraction of the generated H2O2 might not be detected, particularly at high product concentration, because of a potentially increased bimolecular decomposition rate and/or increased crossover to the anode, as frequently observed in methanol or formic acid fuel cells (3840); this would result in an apparent decrease in H2O2 selectivity. Possible impurities in the product solution, such as sodium (common in water), iron (from the device), sulfur (from the SE), and platinum (from the anode), were quantified to be at or below ppm levels determined by inductively coupled plasma atomic emission spectroscopy (ICP-OES) (table S1 and supplementary text, note 3). Therefore, the electrochemically synthesized H2O2 solutions could be put to immediate use out of the cell without any further purification, lowering cost substantially compared with other methods and simplifying the setup for the deployment of on-site generation technology. Long-term stability is another important metric for evaluating catalysis. Our device produced ~1200 and ~11,000 ppm H2O2 solutions continuously in 100-hour test runs with no degradation in activity or selectivity (Fig. 2, E and F). XPS characterization of the CB-10% catalyst after the reaction revealed that the surface oxygen functionality was robust and did not appear to have been electrochemically reduced during the operation of the ORR (fig. S10B).

As a representative demonstration of on-site application, we used the as-synthesized H2O2 from our device for rainwater purification (Fig. 2D and fig. S13). Compared with traditionally used chlorine compounds, which may produce carcinogens in the processed drinking water (41), H2O2 is safe for both human and environmental health when disinfecting and decomposing organic contaminants, typically assessed as removal of total organic carbon (TOC) (42). The use of electrochemically generated H2O2 is not only economical (see supplementary text, note 4), but also avoids the transportation and storage of hazardous bulk H2O2. We directly mixed the generated H2O2 stream (200 mA cm−2, 27 ml hour−1 DI water flow) from our 4-cm2 electrode device with the rainwater stream (feeding rate ranging from 18.32 to 0.05 liters hour−1) to optimize the purification efficiency. The TOC of the pristine rainwater collected at the Rice University campus in Houston was detected to be ~5 ppm (see materials and methods), which is above the Texas treated-water standard of ~2 ppm (43). Decreasing the rainwater feeding rate gradually lowered the TOC remaining (Fig. 2D), demonstrating the efficacy of the generated H2O2 solution in water treatment. A maximal processing rate of 0.88 liter hour−1 (0.22 liter cm−2electrode hour−1 or 2200 liters m2electrode hour−1) was achieved in lowering the TOC level to meet the drinking water standards, making our design economically and environmentally appealing for practical rainwater treatment when scaled up.

We also demonstrated that the oxidation reaction on the anode side could be flexibly modified to be coupled with the cathodic 2e-ORR for applications where H2 is not available (fig. S14). Water oxidation to O2 with concurrent proton release might be easier to access than hydrogen oxidation. A 0.5 M aqueous sulfuric acid solution on the anode side was used to lower the ionic resistance; H2SO4 was not consumed during the reaction and was continuously circulated (see materials and methods). The CEM membrane blocked crossover of the H2SO4 into the porous solid electrolyte layer, ensuring the formation of pure H2O2 solutions. This was confirmed by pH and ICP-OES measurements: The pH of the generated H2O2 solution was ~6 to 7 (pure H2O2 solutions show weak acidity), and the sulfur impurity level was <10 ppm. High H2O2 productivity of 3.3 mmol cm−2 hour−1 (3565 mol kgcat−1 hour−1) could be achieved at a cell voltage of 2.13 V (Fig. 3A), representing an electricity-to-chemical energy conversion efficiency of 22.6%. H2O2 selectivity was very close to that observed with the O2//SE//H2 design at comparable current density (Fig. 2B), ruling out any impact on the cathodic 2e-ORR reaction by the anodic water oxidation. The ultrahigh purity of the synthesized H2O2 solution was confirmed using ICP-OES. A 100-hour test continuously generated pure H2O2 solutions, confirming the robust stability of the O2//SE//H2O cell (fig. S15). To further simplify our process, we directly pumped air rather than purified O2 to the cathode side (Fig. 3C). Although higher cell voltages were required to drive the reaction owing to substantially decreased O2 concentration and activity, the air//SE//H2O cell continued to provide H2O2 selectivity of >90%. A maximal H2O2 partial current of ~123 mA cm−2 was reached at 2.71 V, corresponding to an impressive H2O2 productivity of 2.3 mmol cm−2 hour−1 (2490 mol kgcat−1 hour−1). To validate the scalability of our porous solid electrolyte design, we extended the electrode area from 4 cm2 used for performance evaluation to ~80 cm2 in one unit modular cell (Fig. 3, D to F); these could be further stacked in the future for greater capacity. A maximal cell current of >20 A was achieved, with a high H2O2 selectivity of ~80% and a production rate of ~0.3 mol hour−1. Under a fixed cell current of 8 A, our scaled-up device produced highly concentrated pure H2O2 solutions of up to 20 wt % under a DI flow rate of 5.4 ml hour−1 (Fig. 3F and fig. S12B).

Fig. 3 Electrosynthesis of pure H2O2 solutions by 2e-ORR and water oxidation.

(A) I-V curve for an O2//SE//H2O cell in which H2O is oxidized at the anode side to form protons and O2. A 0.5 M aqueous H2SO4 solution was used to improve ionic conductivity on the anode side and was not consumed during electrosynthesis. (B) Corresponding FEs of the O2//SE//H2O cell. (C) I-V curve and FEs for an air//SE//H2O cell generating pure H2O2 solutions. Pure H2O2 solutions were generated at a high production rate of 2.3 mmol cm−2 hour−1 (2490 mol kgcat−1 hour−1) using only air and water as cathode and anode feedstock, respectively. (D) I-V curve of the scaled-up unit cell module (80 cm2 electrode, no iR compensation), and (E) the corresponding H2O2 FEs. (F) Dependence of H2O2 concentration (up to ~20 wt %) on the DI water flow rate at a constant overall current of 8 A. The data points in (A) to (E) each represent the mean of two independent measurements.

Given the wide variety of liquid products amenable to electrochemical synthesis, our solid electrolyte design could in principle be extended beyond H2O2 generation to other important electrochemical applications.

Supplementary Materials

Materials and Methods

Figs. S1 to S15

Table S1

Supplementary Text

References (5467)

References and Notes

Acknowledgments: Funding: This work was supported by Rice University. H.W. is a CIFAR Azrieli Global Scholar in the Bio-inspired Solar Energy Program. C.X. acknowledges support from a J. Evans Attwell-Welch Postdoctoral Fellowship provided by the Smalley-Curl Institute. Author contributions: C.X. and H.W. conceptualized the project. H.W. supervised the project. C.X. synthesized carbon black catalysts with the help of Y.X. C.X. and Y.X. conducted the catalytic tests and the related data processing. C.X. performed materials characterization and analysis with the help of P.Z. L.F. designed the schemes. C.X. and H.W. wrote the manuscript. Competing interests: A U.S. provisional patent application (no. 62/874,176) based on the technology described in this work was filed on 15 July 2019 by C.X. and H.W. at Rice University. The other authors declare no competing interests. Data and materials availability: All experimental data are available in the main text or the supplementary materials.

Stay Connected to Science

Navigate This Article