Bromide-mediated paired electrosynthesis of hydrogen and epoxide

Abstract

Water electrolysis struggles to compete with the low cost of hydrogen produced from methane or coal. We propose a bromide-mediated paired electrosynthesis system in which an aqueous electrolyte is combined with olefins in an inert organic co-solvent to simultaneously generate hydrogen and epoxides. In this process, bromide electrolysis first produces gaseous hydrogen and solution-confined bromine; the bromine is then used to partially oxidize olefins into high-value epoxides while being regenerated back to bromide. By replacing low-value oxygen with valuable epoxides, this approach increases the economic viability of water electrolysis and can be applied to a wide range of olefins. We demonstrate the concept using styrene, converting water and styrene into hydrogen and styrene oxide in a membraneless system with high efficiency, high rate, and high product selectivity.

The authors present a bromide-mediated paired electrosynthesis system in which an aqueous electrolyte is combined with olefins in an inert organic cosolvent to simultaneously generate hydrogen and epoxides.

Introduction

Green hydrogen from water electrolysis powered by renewable electricity is a key energy carrier and feedstock for decarbonizing hard-to-electrify sectors, including heavy transportation, ammonia and steel production, and refining, as well as for enabling long-duration storage of solar and wind energy¹. However, as of 2024, green hydrogen accounted for only 0.3% of the global hydrogen demand. This is primarily due to the cost gap between green hydrogen ($3.90–$9.20 per kg H₂, in 2024) and gray hydrogen ($1.00–$4.30 per kg H₂, in 2024) produced from fossil fuels¹. Closing this gap is a grand challenge.

Towards this end, paired electrosynthesis aims to increase the economic viability of water electrolysis by replacing the anodic oxygen evolution reaction (OER), a difficult, energy-intensive reaction producing oxygen, a low-value byproduct commonly wasted to the atmosphere, by alternative oxidation reactions that supply electrons and protons for the cathodic hydrogen evolution reaction (HER) and produce high-value products instead of oxygen². The alternative anodic reaction should ideally generate products with both high economic value and substantial market demand to ensure parity with hydrogen demand. Previous studies on paired electrosynthesis have shown that anodic partial oxidation of olefins to epoxides, i.e., epoxidation, can effectively pair with cathodic HER, using water and olefin as feedstock materials to produce hydrogen and epoxides³. The key advantage of this pairing lies in the synergistic utilization of water as a source of protons for hydrogen evolution and hydroxide ions for the epoxidation of olefins to value-added epoxides. The global market for epoxides is currently (2025) valued at approximately $78 billion, roughly one-third of the global hydrogen market, and is experiencing rapid growth⁴. By replacing oxygen evolution with value-added products, this process provides a markedly higher economic potential than standard water electrolysis⁵.

Electrolytic epoxidation has been demonstrated through three distinct approaches⁶. The first approach employs hydrogen peroxide (H₂O₂) produced by the oxygen reduction reaction (ORR) for the partial oxidation of olefins to epoxides, as practiced in the industrial hydrogen peroxide to propylene oxide (HPPO) process⁷. Since the ORR is a cathodic reaction, this process does not produce hydrogen. The second approach utilizes reactive oxygen species (ROS) generated by anodic water oxidation to convert unsaturated C = C bonds into epoxides, mediated by either heterogeneous catalysts at the anode or homogeneous catalysts in solution⁸. This process often leads to over-oxidation of the olefin molecules, evolving CO₂ and O₂ as byproducts⁹. Over-oxidation reduces the yield of epoxides and the hydrogen gas purity, which is critical for safe operation without H₂/O₂ mixing. Thus, membranes are used to prevent H₂/O₂ crossover and cathodic side-reactions¹⁰, substantially elevating electrolyzer capital and maintenance costs. The third approach is to oxidize halide ions (e.g., chloride and bromide) instead of water, thereby mitigating over-oxidation of water to O₂ and of the organic substrate to CO₂^3,10. The oxidized halogen species (e.g., chlorine and bromine) react with olefins to yield halohydrin intermediates, which are subsequently transformed to epoxides under alkaline conditions, with concurrent regeneration of the initial halide ions. This indirect pathway improves product yield and selectivity.

Chloride-mediated epoxidation benefits from the high chloride content of seawater (2–3 wt%, corresponding to ~0.5–0.6 M)¹¹. However, its practical implementation faces several challenges. First, the oxidation product, chlorine (Cl₂), is gaseous and therefore readily escapes from solution. As a result, it must be separated from the co-produced hydrogen, typically necessitating gas-tight separation of the anodic and cathodic compartments using a membrane or diaphragm. Second, membrane-divided operation introduces additional technical constraints. In the chloro-alkali process, for example, membrane lifetimes are typically 3–5 years, and annual maintenance is required¹². Moreover, membrane-divided cells require highly purified brine, with calcium and magnesium concentrations below 20 ppb, to avoid membrane clogging by precipitated calcium and magnesium hydroxides¹³. Third, chloride electro-oxidation is energetically costly, as its standard electrode potential (E^Ө = 1.36 V_SHE) is higher than that of oxygen evolution (E^Ө = 1.23 V_SHE)¹⁴, increasing energy consumption and promoting competition with the OER.

To overcome these limitations, we propose bromide-mediated epoxidation as the anodic reaction paired with cathodic hydrogen evolution. Bromide electro-oxidation has a lower standard electrode potential (E^Ө = 1.09 V_SHE) than both chloride oxidation and OER, and generates solution-confined bromine (Br₂) rather than gaseous chlorine¹⁵. This feature lowers the thermodynamic barrier for halogen-mediated oxidation and reduces the risk of gas-phase mixing between the oxidized halogen and hydrogen. Recent studies have demonstrated bromide-mediated epoxidation in membrane-divided and membraneless cell configurations^16–18. More recently, bromine-mediated electrochemical styrene valorization was extended beyond epoxidation by converting a brominated styrene intermediate into styrene oxide, styrene glycol, and styrene carbonate, further highlighting the synthetic versatility of halogen-assisted styrene functionalization¹⁹. However, these studies primarily focused on olefin oxidation or product diversification, rather than on the deliberate integration of bromide-mediated epoxidation with efficient cathodic H₂ evolution in a high-current-density membraneless electrolyzer. This distinction is important because membrane-divided cells face durability challenges in bromine-containing media, while membraneless operation introduces the possibility of cathodic and anodic product crossover. In such a configuration, electrode selectivity becomes essential to suppress back reactions, including hydrogen oxidation at the anode and reduction of oxidized bromine species at the cathode. This work addresses these coupled challenges by developing a membrane-free paired electrosynthesis system that simultaneously produces hydrogen and an epoxide at high current density, while maintaining high Faradaic efficiency, product selectivity, and yield. Achieving this balance requires selective electrodes and an electrolyte environment that promotes bromide-mediated olefin epoxidation while preserving efficient cathodic H₂ evolution. To this end, we combine a dimensionally stable anode (DSA), an industrial electrode specifically engineered for selective halogen oxidation in the chlor-alkali process²⁰, with an acetonitrile-containing electrolyte designed to enhance olefin availability and promote bromide-mediated epoxidation. In addition to improving substrate solubility, acetonitrile provides a wide intrinsic electrochemical stability window under nonaqueous conditions (6.1 V)²¹, although the practical operating window in the present aqueous/acetonitrile bromide electrolyte is ultimately defined by bromide oxidation, hydrogen evolution, and electrode selectivity.

Previous studies have shown that adding a water-miscible organic co-solvent creates an environment that improves olefin availability and favors epoxidation over undesired side reactions^18,19,22,23. Specifically, a water/acetonitrile mixed solvent (1:1, v/v) was reported to facilitate high epoxide yield and Faradaic efficiency in a one-pot membraneless system¹⁸. These studies focused on achieving high product selectivity, whilst other benchmarks such as durability, efficiency and rate (i.e., current density) were overlooked. These benchmarks are critical for industrial electrolysis and hydrogen production at scale because they significantly influence the levelized cost of hydrogen. In this work, we present a path forward to achieving these benchmarks in a membraneless electrolytic cell that separates the hydrogen produced at the cathode and the epoxide produced in the solution without dividing the cell into separate compartments. Building on our previous work on decoupled water electrolysis²⁴, particularly on bromide-mediated electrolysis that produces hydrogen and oxygen in separate cells without membranes²⁵, and on our subsequent mechanistic analysis of the operating window and electrolyte trade-offs of the bromide/bromate cycle²⁶, we sought to redirect bromide-mediated anodic chemistry from oxygen evolution toward value-added olefin epoxidation. This work presents an integrated membrane-free paired electrosynthesis system that combines a halogen-selective DSA anode with an aqueous NaBr electrolyte containing acetonitrile as a water-miscible organic cosolvent, enabling simultaneous cathodic hydrogen evolution and bromide-mediated styrene epoxidation.

Results and discussion

The overall process comprises electrochemical and chemical reactions. Electrochemical reactions generate gaseous hydrogen rxn. (1) and solution-confined liquid bromine rxn. (2) at the cathode and anode, respectively, after which bromine and hydroxide ions spontaneously oxidize olefins to epoxides in the solution. The overall reaction consumes water and olefin (and electrical energy) and produces hydrogen and epoxide, as illustrated in Fig. 1a.

$$Cathode:2H_{2}O+{2e}^{-}\to H_2+{2OH}^{-}$$ 1

$$Anode:{2Br}^{-}\to {Br}_2+{2e}^{-}$$ 2

Fig. 1. Paired electrosynthesis.

a Photograph of the membraneless paired-electrosynthesis cell before operation in a stirred 1:1 (v/v) water/acetonitrile (ACN) mixed solvent containing 0.3 M NaBr and 0.02 M styrene. The proposed bromide-mediated epoxidation pathway and the competing direct-bromination pathway are illustrated below. b iR-corrected anodic (top) and cathodic (bottom) polarization curves of current density versus potential measured in a 1:1 (v/v) water/acetonitrile mixed solvent containing 0.3 M NaBr in the aqueous phase and 0.02 M styrene in the organic phase (black); 0.3 M NaBr at pH 7 (red); 0.3 M NaNO₃ at pH 7 (magenta); and 0.3 M NaBr with 0.5 M H₂SO₄ at pH 0 (green). The curves were obtained from chronoamperometric measurements (Supplementary Fig. S1) at constant potentials held for 30 s under stirring (350 rpm), except for the complementary measurement in 0.3 M NaBr without stirring (blue). All measurements were performed at room temperature using freshly prepared solutions. Error bars represent standard deviations.

We begin by presenting cathodic and anodic chronoamperometry and linear sweep voltammetry (LSV) measurements carried out in a three-electrode configuration with an Ag/AgCl reference electrode to record the cathode and anode potentials, followed by electrolysis measurements in a two-electrode configuration to measure the overall cell voltage. All the electrochemical measurements were carried out in undivided electrolytic cells containing water/acetonitrile mixed solvent (1:1, v/v) with 0.3 M NaBr in the aqueous phase (pH 7)²⁷, unless otherwise specified. In keeping with the conventional practice in water electrolysis²⁸, we present the iR-corrected cell voltage, cathodic and anodic potentials, corrected by subtracting the current (i) × resistance (R) product from the voltage or potential applied by the potentiostat, as detailed in the Methods section. The raw values, without iR correction, are presented in the Supplementary Information.

The electrochemical process pairs cathodic HER with the anodic bromine evolution reaction (BER). Cathodic and anodic polarization curves, extracted from chronoamperometry measurements presented in Supplementary Fig. S1, are shown in Fig. 1b. In still electrolyte (without stirring), the cathodic polarization presents a typical exponential Tafel curve with an onset potential of −1.14 V vs Ag/AgCl at 10 mA cm⁻² (Fig. 1b, blue). Hydrogen bubbles evolved at the cathode, and the solution turned yellowish at the anode (Supplementary Video 1), indicating bromine evolution²⁹. The yellow liquid sank to the bottom of the cell. The onset potential of the anodic reaction was 0.87 V vs Ag/AgCl at 10 mA cm⁻². Unlike the cathodic curve, the anodic curve has a shoulder at 40 mA cm⁻² (blue), indicating mass transport limitation. No gas bubbles evolved at the anode (Supplementary Video 1). These observations, especially the shoulder, indicate that the anodic reaction was BER rather than water oxidation, since water is prevalent, which excludes mass transport limitations at this current density. To confirm this, a control experiment with 0.3 M NaNO₃ instead of NaBr was carried out (Fig. 1b, magenta). The onset potential of the anodic reaction, at 10 mA cm⁻², increased from 0.87 V vs Ag/AgCl in NaBr to 1.45 V vs Ag/AgCl in NaNO₃, corresponding to bromine and oxygen evolution reactions, respectively. Thus, under these conditions, the overpotential for the onset of bromine evolution is lower by 0.58 V than for oxygen evolution, showcasing one of the advantages of replacing OER by BER.

When stirring the solution (Fig. 1b, red), the cathodic curve shifted anodically by about 0.1 V, and the anodic shoulder disappeared. Stirring reduces both cathodic and anodic overpotential losses at current densities above 30 mA cm⁻². However, the iR-corrected cell voltage, i.e., the difference between the anodic and cathodic potentials, was still high, reaching 1.80 V at 10 mA cm⁻², corresponding to an overvoltage of 0.71 V with respect to the reversible cell voltage, 1.09 V. Under these conditions, almost all the overpotential is caused by the cathodic reaction (Supplementary Fig. S2). This demonstrates the difficulty of evolving hydrogen in neutral-pH electrolytes, a well-known challenge in seawater and mild alkaline water electrolysis^30,31. To examine the performance limits, in terms of cathodic polarization, of our system, we added 0.5 M H₂SO₄ to the solution (Fig. 1b, green) to create favorable acidic conditions for hydrogen evolution. Consequently, the cathodic curve shifted anodically by ~0.80 V, reaching 10 mA cm⁻² at −0.23 V vs Ag/AgCl, whereas the anodic curve shifted slightly to higher overpotentials. The substantial reduction in cathodic polarization loss indicates considerable potential for further improvement by promoting water dissociation (H–OH bond cleavage), and/or promoting H₂O*/OH* intermediates at the electrode/electrolyte interface, as reported elsewhere^32–34. The increase in the anodic overpotential can be assigned to the competition between sulfate and bromide anion adsorption³⁵.

Figure 1b and Supplementary Fig. S2b show the cathodic and anodic chronoamperometric and LSV measurements obtained in a 1:1 (v/v) water/acetonitrile mixed solvent containing 0.3 M NaBr and 0.02 M styrene. The addition of styrene induces small anodic shifts in both the cathodic and anodic polarization curves, by 0.05 and 0.10 V, respectively. These shifts are attributed to styrene adsorption on the corresponding electrode surfaces. Styrene adsorption at the cathode enhances HER activity by modulating the proton binding energy on the Pt surface³⁶, whereas adsorption at the anode blocks surface active sites. The increase in anodic potential confirms that styrene is not directly oxidized at the anode, as direct oxidation would be expected to lower the potential³⁷. Overall, the decrease in cathodic polarization partially offsets the increase in anodic polarization. An iR-corrected cell voltage of 2.18 ± 0.04 V was obtained from the difference between the anodic and cathodic potentials at a current density of 50 mA cm⁻², which is nearly identical to the value measured in the absence of styrene (2.12 ± 0.05 V).

Steady-state galvanostatic measurements were performed in a water/acetonitrile mixed solvent containing 0.3 M NaBr in the aqueous phase and 0.02 M styrene in the organic phase over a wide current-density range of 5–500 mA cm⁻² (Supplementary Fig. S3a). Figure 2 shows the iR-corrected cell voltage as a function of current density, together with data from previous studies on paired electrosynthesis for the co-production of hydrogen and epoxides^{8–10,16,18,38–48}. These literature reports are grouped into membrane-based and membraneless systems (blue and green, respectively). The previously reported results are widely scattered; our system outperforms all but one report, which employed a diaphragm-divided cell and achieved a product selectivity of 75%⁴⁵, substantially lower than that obtained here (see below). Cell voltages of 1.89 ± 0.08 V at 10 mA cm⁻², 2.34 ± 0.06 V at 100 mA cm⁻², and 3.27 ± 0.20 V at 500 mA cm⁻² were achieved. These values remain higher than those of state-of-the-art PEM and alkaline water electrolyzer modules, which typically operate at 1.8–2.0 V at current densities of 1–3 A cm⁻² and 0.5–0.8 A cm⁻², respectively²⁴. This voltage gap most likely arises from the sluggish HER in our neutral-pH mixed solvent⁴⁹. Nevertheless, the difference is not large, and further advances in electrode design and process optimization could reduce the cell voltage and improve overall efficiency. Overall, our results surpass most previous reports on paired electrosynthesis and approach practically competitive voltages compared with water electrolysis.

Fig. 2. Comparison with previous studies.

iR-corrected cell voltage as a function of steady-state current density, obtained from galvanostatic measurements at current densities of 5–500 mA cm⁻² (red circles). The data points were extracted from chronoamperometric measurements (Supplementary Fig. S3a) and are presented as mean values ± standard deviations (error bars) from four independent measurements at each current density (Supplementary Table S3). For comparison, literature data on paired electrosynthesis are shown in blue and green for membrane-based and membraneless cells, respectively, and with different symbols for bromide-, chloride-, and ROS-mediated epoxidation pathways^{8–10,16,18,38–48}. The data from Ref. ¹⁰ (marked by ⌧ and connected by a dashed line) represent the anodic half-cell voltage only and therefore exclude cathodic losses. The secondary y-axis (right) shows the voltage efficiency (1.48 V divided by cell voltage) relative to the higher heating value (HHV) of hydrogen.

Bromide-mediated epoxidation proceeds through a sequence of electrochemical and chemical reactions (Fig. 1a)⁴². Electrolysis produces hydrogen gas at the cathode and solution-confined bromine at the anode (Supplementary Video 1). Under stirring (Supplementary Video 2), bromine undergoes hydrolysis in water to form hypobromous acid (HBrO), rxn. (3), as confirmed by UV-vis spectroscopy (Supplementary Section S2). The electrophilic HBrO then reacts with the olefin via bromohydroxylation to form a bromohydrin intermediate, rxn. (4). Finally, the bromohydrin is deprotonated and cyclized by the hydroxide ions (OH⁻) generated at the cathode rxn. (1), yielding the corresponding epoxide, rxn. (5). This bromohydrin-mediated pathway is consistent with a recent bromine-mediated styrene valorization study, in which styrene was first converted to a brominated intermediate, predominantly 2-bromo-1-phenylethanol (BPE, styrene bromohydrin), which was subsequently transformed into styrene oxide under alkaline conditions or diverted to styrene glycol and styrene carbonate by tuning the downstream reaction conditions¹⁹. Overall, coupling the electrochemical steps rxns. (1–2) with the epoxidation sequence reactions rxns. (3–5) enables the net conversion of water and olefin into hydrogen and epoxide, rxn. (6), as illustrated in Fig. 1a.

$${Br}_2+H_{2}O\to HBrO+HBr$$ 3

$${RCH=CH}_2+HBrO\to RCH\left(OH\right){CH}_{2}Br$$ 4

$$RCH\left(OH\right){CH}_{2}Br+{OH}^{-}\to RCH{CH}_{2}O+H_{2}O+{Br}^{-}$$ 5

$${RCH=CH}_2+H_{2}O\to {RCHCH}_{2}O+H_2$$ 6

Paired electrosynthesis of hydrogen and epoxide was carried out at applied currents of 50–500 mA using a 1 cm² Pt foil cathode and a 2 cm² DSA anode (Supplementary Fig. S4a–e), corresponding to current densities relevant to practical hydrogen production^24,30. Upon polarization of the initially clear and colorless water/acetonitrile electrolyte, hydrogen evolution commenced at the cathode, with bubbles released both into the headspace and into the stirred solution (Supplementary Video 3). Quantitative ¹H NMR analysis of the electrolyte after electrosynthesis (Supplementary Section S3) confirmed efficient styrene oxide production across the full current range (Table 1) using 1 cm² DSA anode. Epoxide yields of 81–89% were obtained, together with high anodic Faradaic efficiencies of 94–98%, including an epoxide yield of 89 ± 2% and an anodic Faradaic efficiency of 98 ± 2% at 50 mA cm⁻² and an electrolyte volume 50 mL. For experiments performed at higher applied currents, larger electrolyte volumes of 100 or 500 mL were used to maintain constant NaBr (0.3 M) and styrene (0.02 M) concentrations. The consistently high yields and Faradaic efficiencies indicate that back reactions associated with redox shuttling, including cathodic reduction of Br₂, HBrO, or styrene oxide, as well as anodic hydrogen oxidation, are effectively suppressed despite the use of a membraneless cell without separate cathodic and anodic compartments. This suppression is attributed to the complementary selectivity of the electrodes: the DSA anode preferentially promotes bromide oxidation^20,50, while the Pt cathode selectively drives hydrogen evolution in water/acetonitrile mixed solutions^51–53. This interpretation is further supported by the hydrogen evolution measurements discussed in the following section. The modest loss in epoxide yield is attributed primarily to direct bromination by electrogenerated bromine, as previously reported¹⁹. Overall, these results demonstrate the robustness of the system, which maintains efficient operation at high current density despite the low substrate concentration, reaching a normalized anodic current density as high as 12.5 mA cm⁻² mM⁻¹ styrene, 12.5-fold higher than previously reported¹⁹.

Table 1.

Summary of paired electrosynthesis performance under different conditions

Entry	Solvent	[NaBr] (M)	Stirring	Current (mA cm−2)	Cell voltage (iR corrected)	Anodic Faradaic efficiency	Selectivity	Epoxide Yield
1	H2O/ACN	0.3	350 rpm	50	2.14 ± 0.03 V	98 ± 2%	91 ± 1%	89 ± 2%
2				100	2.25 ± 0.03 V	95 ± 2%	87 ± 1%	83 ± 1%
3				200	2.31 ± 0.03 V	95 ± 2%	86 ± 1%	82 ± 1%
4				300	2.62 ± 0.05 V	95 ± 2%	86 ± 1%	82 ± 1%
5		0.15			4.10 ± 0.10 V	94 ± 1%	96 ± 1%	90 ± 2%
6		0*			2.21 ± 0.05 V	9%	2%	0.2%
7	H2O	0.3			2 to 8 V	7 ± 1%	29 ± 1%	2 ± 1%
8	H2O/ACN			400	2.68 ± 0.07 V	94 ± 2%	86 ± 1%	81 ± 1%
9				500	2.73 ± 0.10 V	94 ± 2%	87 ± 1%	82 ± 1%
10			0 rpm		3.12 ± 0.20 V	96 ± 3%	74 ± 3%	71 ± 5%

^* NaBr replaced by 0.3 M Na₂SO_4.

To further assess the effects of electrolyte composition and operating conditions, several parameters were systematically varied. Lowering the NaBr concentration to 0.15 M (entry 5, Table 1) slightly improved product selectivity to 96 ± 1%, but this improvement was accompanied by a marked increase in cell voltage to 4.1 V, indicating higher energy consumption. In contrast, replacing NaBr with Na₂SO₄ (0 M bromide, entry 6) caused a dramatic decrease in anodic Faradaic efficiency (9%), product selectivity (2%), and epoxide yield (0.2%), underscoring the essential role of bromide as a redox mediator. The importance of acetonitrile was revealed by performing electrosynthesis in an aqueous bromide electrolyte without acetonitrile (entry 7), which resulted in inefficient epoxide production, with a Faradaic efficiency of only 7 ± 1%, selectivity of 29 ± 1%, and epoxide yield of 2 ± 1%. Under these conditions, substantial fluctuations in cell voltage were observed (Supplementary Fig. S5), along with visible fouling of the DSA anode and formation of flocculent precipitates in the electrolyte after electrosynthesis (Supplementary Fig. S6). High-resolution SEM analysis further revealed pronounced morphological changes on the anode surface (Supplementary Fig. S6), indicating surface degradation. These results show that acetonitrile is critical not only for promoting selective bromide-mediated epoxidation but also for stabilizing the electrochemical environment and preserving the DSA anode during operation.

Acetonitrile not only stabilizes the electrode surface but also modulates the bromine-mediated reaction pathway. To probe this effect, we performed time-resolved UV–Vis transmission measurements during electrolysis in a water/acetonitrile mixed solvent (1:1, v/v) and in a purely aqueous electrolyte under otherwise identical conditions (0.3 M NaBr, 100 mA, 200 s, membraneless cell equipped with a 1 cm² Pt cathode and a 1 cm² DSA anode, without styrene). The two media showed markedly different bromine speciation. In water/acetonitrile, absorbance bands appeared at ~269 and ~386 nm, consistent with HBrO/Br₃^– and Br₂, respectively (Supplementary Fig. S7 and Supplementary Section S2)⁵⁰. In contrast, in the absence of acetonitrile, the ~269 nm band diminished, the ~386 nm Br₂ band disappeared, and a new feature emerged at ~337 nm, corresponding to BrO⁻. Because BrO⁻ is not an effective epoxidation mediator under these electrochemical conditions unless additionally activated by photoirradiation⁵², these results indicate that acetonitrile suppresses the conversion of electrophilic bromine species into BrO⁻. We therefore propose that acetonitrile slows the reaction of HBrO with cathodically generated OH⁻, thereby preserving electrophilic bromine species that promote selective epoxidation. This interpretation is consistent with a recent bromine-mediated styrene valorization study showing that solvent composition strongly governs the competition between bromohydroxylation and direct bromination: water/acetonitrile mixtures favor formation of the bromohydrin intermediate, whereas excessive acetonitrile promotes direct Br₂ addition to styrene and formation of dibrominated byproducts¹⁹. Accordingly, the lower Faradaic efficiency and epoxide yield observed in purely aqueous electrolyte (entry 7, Table 1) can be attributed to a speciation shift toward unproductive BrO⁻-associated pathways.

The role of mass transport was further examined by performing electrosynthesis without stirring. Omitting stirring caused only a modest decrease in product selectivity and yield relative to the stirred case, while the anodic Faradaic efficiency remained largely unchanged (entries 9 and 10, Table 1). Collectively, these results indicate that bromide directs the epoxidation pathway, while acetonitrile plays a dual role: stabilizing the electrochemical environment and electrode surface, and modulating bromine speciation toward electrophilic species that favor selective epoxidation. Together with the membraneless cell architecture and electrode selectivity, these features enable efficient operation at high current density.

System durability was evaluated over 10 h of electrosynthesis by monitoring the cathodic Faradaic efficiency for hydrogen evolution, calculated from the measured hydrogen volume relative to the total charge passed through the cell (current × time). The hydrogen volume was determined by the water-displacement method²⁵ using a sealed three-neck flask (100 mL) configured as a membraneless electrolytic cell with an undivided Pt foil cathode and DSA anode, each with a geometric area of 1 cm² (Fig. 3a). Galvanostatic electrosynthesis was performed at a constant current density of 100 mA cm⁻². Owing to the limited cell volume (100 mL) and the low solubility of styrene in the water/acetonitrile mixed solvent, the electrolyte solution, consisting of a 1:1 (v/v) mixture of 0.3 M NaBr aqueous electrolyte and 0.02 M styrene in acetonitrile, was refreshed every hour. Accordingly, the 10 h durability test was divided into ten consecutive 1 h segments (Fig. 3b). Across all measurements, the apparent cathodic Faradaic efficiency, calculated from the collected gas volume without correcting for water vapor and gas purity, was 101 ± 5% (Fig. 3c) and remained stable throughout the experiment, with no observable temporal drift.

Fig. 3. Durability test.

a Electrosynthesis setup used to measure the volume of H₂ evolved during electrosynthesis. b Measured H₂ evolution during ten consecutive 1 h electrosynthesis segments at a constant current density of 100 mA cm⁻² (colored curves), compared with the theoretical H₂ amount calculated from the total charge passed (current × time), assuming a two-electron process per H₂ molecule (black curves). c iR-corrected cell voltage as a function of time during the test (colored curve, left y-axis) and the corresponding cathodic hydrogen Faradaic efficiency for each segment (diamond symbols, right y-axis). The durability test was performed in a sealed three-neck flask (100 mL) containing a 1:1 (v/v) water/acetonitrile mixed solvent containing 0.3 M NaBr and 0.02 M styrene, using DSA anode and Pt foil cathode, each with a geometric area of 1 cm². Each 1 h segment was conducted with freshly prepared electrolyte.

To quantitatively determine the cathodic hydrogen Faradaic efficiency, the gaseous products generated during electrosynthesis at 100, 300, and 500 mA cm⁻² were swept from the cell (Supplementary Fig. S8) using an Ar flow, collected in sealed gas-sampling bags under identical electrolysis and sampling conditions, and analyzed by gas chromatography equipped with a thermal conductivity detector (GC-TCD). The GC-TCD chromatograms showed a well-defined H₂ peak at a retention time of 1.35 min, which was quantified using a calibration curve (Supplementary Figs. S9, 10). The corresponding cathodic hydrogen Faradaic efficiencies were 98 ± 2%, 98 ± 2%, and 94 ± 1% at 100, 300, and 500 mA cm⁻², respectively (Supplementary Table S1; corresponding chromatograms are shown in Supplementary Fig. S11), confirming highly efficient hydrogen evolution over the investigated current-density range.

To further evaluate the gas composition, samples collected at 100 and 500 mA cm⁻² were analyzed by gas chromatography–mass spectrometry (GC–MS). In both cases, partially overlapping peaks were observed at retention times of 0.9–1.0 min (Supplementary Figs. S12, 13). These contributions were assigned by extracting the characteristic ion signals of O₂, N₂, H₂O, and Ar at m/z = 32, 28, 18, and 40, respectively. The total concentration of minor gaseous impurities, excluding Ar and H₂, was estimated to be 1.7% and 3.4% for samples collected at 100 and 500 mA cm⁻², respectively. The minor impurities are primarily attributable to residual moisture and traces of air left after Ar gas purging and sampling. Importantly, only trace O₂ was detected, with concentrations of 0.4–0.8 vol%, which is well below the 3 vol% O₂-in-H₂ emergency shutdown threshold specified in ISO 22734-1:2025 for hydrogen generators and far below the oxygen concentration corresponding to the hydrogen-rich flammability limit of H₂/O₂ mixtures, indicating that oxygen evolution makes, at most, a minor contribution under these conditions. This conclusion is consistent with the LSV measurements, which showed that the onset potential for OER is substantially more positive than that for bromide oxidation/bromine evolution (Fig. 1b). Importantly, no Br₂ was detected in the gas phase under any of the investigated conditions.

Because water vapor and minor gaseous impurities contribute to the total gas volume measured by water displacement, the gas volume obtained from the water-displacement measurements (Fig. 3b) was corrected by 2% prior to calculating the cathodic hydrogen Faradaic efficiency (Fig. 3c). This correction yielded an average cathodic hydrogen Faradaic efficiency of 99 ± 5% during the 10 h durability test at 100 mA cm⁻² (Fig. 3), in excellent agreement with the independently calibrated GC-TCD value of 98 ± 2% obtained at the same current density (Supplementary Fig. S11).

The high cathodic Faradaic efficiency is consistent with the high anodic Faradaic efficiency determined by quantitative ¹H NMR analysis (Table 1). Moreover, it indicates that competing cathodic reactions, such as styrene or acetonitrile reduction, are effectively suppressed, in agreement with the NMR analysis showing no detectable reduction products (Supplementary Figs. S38–40). Consistent with the stable Faradaic efficiencies, the cell voltage also remained largely stable throughout the ten 1 h electrosynthesis segments (Fig. 3c), increasing only slightly from 2.43 V at the beginning to 2.54 V at the end of the 10 h test. This voltage drift began in the sixth segment and ceased after the ninth, while the other segments remained stable.

The stability of the electrodes was evaluated by SEM, EDS, XRD, XPS, and Raman spectroscopy analyses of the DSA anode, which is the more vulnerable of the two electrodes, before and after the 10 h durability test (Supplementary Section S4). No discernible changes were observed in electrode morphology, elemental composition, phase composition, or the oxidation states of Ti and Ru (Fig. 4 and Supplementary Fig. S50), demonstrating the stability of the anode under the tested conditions.

Fig. 4. Electrode stability.

SEM micrographs and corresponding EDS elemental maps of Ti and Ru (magenta and green, respectively) of the DSA anode before a and after b the durability test. c Raman spectra of the DSA anode before (black) and after (red) the durability test, obtained in mapping mode from the selected regions indicated in the optical micrographs in (d). In (d), the optical micrographs outlined in black and red correspond to the anode before and after the durability test, respectively. e XRD patterns of the DSA anode before (black) and after (red) the durability test; the intensity difference (∆) between the two patterns is shown by the blue curve.

Our paired electrosynthesis system achieved high Faradaic efficiencies for both the cathodic and anodic reactions, reaching 98 ± 2% and 98 ± 1%, respectively, while operating at high current densities of up to 500 mA cm⁻² and low cell voltages of 1.9 V at 10 mA cm⁻², 2.3 V at 100 mA cm⁻², and 3.3 V at 500 mA cm⁻² (all iR-corrected). Similarly, the epoxidation of styrene to styrene oxide proceeded with high selectivity and yield, both reaching ~80–90%. These performance metrics surpass those reported in previous studies on bromide- and chloride-mediated epoxidation in both membrane-divided and membraneless cell configurations (Fig. 5). The durability of the system was demonstrated over a 10 h electrosynthesis test, during which high Faradaic efficiency was maintained (Fig. 3b). The slight voltage drift observed in only four of the ten segments (Fig. 3c) is likely attributable to an experimental artifact rather than electrode degradation, as no discernible changes in electrode microstructure and composition were detected after the durability test (Fig. 4). Notably, these promising results were obtained using commercially available DSA and Pt foil electrodes together with low-cost NaBr electrolyte. In addition, the process was conducted in a membraneless electrolytic cell without separate cathodic and anodic compartments, providing a simple and flexible system architecture and operation²⁴, while avoiding membrane stability challenges that are especially severe in halide-containing electrolytes⁵⁴. These findings highlight the practical potential and scalability of the proposed electrosynthesis system.

Fig. 5. Benchmark radar chart.

Comparison of anodic and cathodic Faradaic efficiencies, product selectivity, and voltage efficiency (relative to the higher heating value of hydrogen) at 100 and 300 mA cm⁻². The symbol shapes and colors are consistent with those used in Fig. 2: red circles denote this work; blue and green symbols represent membrane-divided and membraneless cells, respectively; and squares and circles indicate chloride- and bromide-mediated processes, respectively. The data from Ref. ¹⁰ (blue dashed line) and Ref. ¹⁸ (green dashed line) correspond to the anodic half-cell voltage only and therefore exclude cathodic losses.

Although demonstrated here using a liquid olefin (styrene) that was partially oxidized to a liquid epoxide (styrene oxide), the system can, in principle, be adapted to gaseous olefins such as ethylene and propylene, yielding the corresponding epoxides, ethylene oxide and propylene oxide. Upon introduction of a gaseous olefin, bromine generated in solution reacts to form a bromohydrin intermediate, which is subsequently deprotonated and cyclized by hydroxide ions produced during cathodic HER to yield the corresponding epoxide. In this way, the active mediator HBrO can react homogeneously with the olefin in the electrolyte, analogous to the industrial HPPO process⁷. Because ethylene oxide and propylene oxide have relatively low boiling points (10.7 and 35 °C, respectively), the dissolved epoxide products can be readily transferred to the gas phase by mild heating. The same principle also enables straightforward separation of the epoxide product from hydrogen by cooling below the boiling point of the epoxide. Overall, the process consumes only water and olefin, offering the prospect of low-maintenance operation based primarily on replenishment of these feedstocks. Future scale-up to less reactive or more hydrophobic substrates, such as ethylene and propylene, will likely require further optimization of substrate delivery, electrolyte composition, and reactor hydrodynamics, particularly under flow-cell operation.

Our paired electrosynthesis process offers compelling economic advantages over conventional water electrolysis. For example, under a realistic scenario assuming nearly 100% hydrogen yield and 90% ethylene oxide yield at a current density of 100 mA cm⁻², the process would consume 66.5 kWh per kilogram of hydrogen produced. Although this value is 25–33% higher than that of state-of-the-art water electrolyzers (50–53 kWh per kg H₂)²⁴, the added value derived from co-producing ethylene oxide instead of oxygen is expected to substantially enhance the overall process economics. Based on market values of approximately $2.4 kg⁻¹ for hydrogen and $1.3 kg⁻¹ for ethylene oxide, the net value of the paired process is estimated to be about $18.5 per kilogram of hydrogen produced (Supplementary Section S5). By contrast, conventional water electrolysis generates hydrogen together with low-value oxygen, which is typically vented to the atmosphere, giving a total product value of only about $2.4 per kilogram of hydrogen. Other epoxides can provide even higher-value products (Supplementary Table S9), although their market demand is lower than that of ethylene oxide. More broadly, the global epoxide market is valued at approximately $78 billion. In addition to epoxides, this approach could also be extended to the synthesis of other value-added anodic products, such as glycols and cyclic carbonates, increasing the potential total addressable market to $130 billion, equivalent to 58% of the hydrogen market (Supplementary Fig. S52).

Paired electrosynthesis has the potential to improve the economic viability of green hydrogen production by co-producing hydrogen and value-added chemicals instead of oxygen. By tuning the electrolyte conditions, including pH and temperature, and introducing an olefin-containing organic solvent, bromide-mediated water electrolysis²⁵ can be directed to produce hydrogen and high-value epoxides rather than hydrogen and low-value oxygen. This strategy enables membraneless electrosynthesis with high rates, efficiencies, and product selectivities. Membraneless operation is particularly advantageous in bromide- and chloride-containing electrolytes, which are known to cause rapid membrane degradation. Future efforts should focus on developing low-cost, durable cell components and translating this proof of concept into industrially relevant electrolyzer systems. Another key challenge will be aligning the green hydrogen and epoxide value chains, which will require rigorous assessment of levelized cost, environmental impact, and process circularity. The volatility of bromine and acetonitrile, both hazardous substances, may affect the process's environmental impact and economics. However, these concerns could be mitigated through closed-system operation, condensation-based recovery, efficient gas handling, off-gas treatment and reactor designs with strong electrolyte circulation and mass transfer. A comprehensive hazard and operability (HAZOP) analysis of bromine and solvent management during prolonged operation will also be essential. These issues are beyond the scope of the present study, which is intended as a process-level demonstration of high-performance paired electrosynthesis using commercial components in a membraneless cell architecture.

Methods

Chemicals

Double-distilled water (DDW; Direct-Q 3 UV, Merck), acetonitrile (HPLC grade, Sigma-Aldrich), sodium bromide (NaBr, 99 + %, Alfa Aesar), sodium nitrate (NaNO₃, 99 + %, ACS grade, Acros), and sulfuric acid (H₂SO₄, 95–98%, analytical reagent grade, Bio-Lab) were used to prepare the water/acetonitrile mixed solutions. Styrene (99%, Thermo Scientific) was used as the model olefin substrate in this study. After electrosynthesis, styrene oxide and other side products were dissolved in deuterated chloroform (chloroform-d, 99 + %, Thermo Scientific) for NMR analysis using toluene (99%, analytical reagent grade, Bio-Lab) or tetramethylsilane (99.9%, AA Blocks) as the internal standards. Authentic styrene (99%, Thermo Scientific) and styrene oxide (98 + %, AA Blocks) were used as the reference samples. Sulfuric acid was also used to clean the electrochemical cells and Pt electrodes before use.

Electrodes

Pt foil electrodes (10 × 10 × 0.1 mm; Gaoss Union) were used as cathodes in the electrosynthesis measurements and, in selected control experiments, as anodes. Before use, the Pt electrodes were cleaned by immersion in sulfuric acid and then thoroughly rinsed with DDW. Commercial RuO₂–TiO₂/Ti dimensionally stable anodes (DSAs; Ti substrate thickness, 1 mm; mixed-oxide layer thickness, ~20–30 μm; DSA15K, De Nora) were used as anodes. In most experiments, the exposed geometric areas of both the cathode and anode were 1 cm². For experiments used to determine the anodic Faradaic efficiency, the exposed DSA anode area was increased to 2 cm² while the Pt cathode area remained 1 cm². The exposed DSA area was defined by masking the electrode with Kapton tape, leaving either a 1 × 1 cm² or 1 × 2 cm² region exposed at the end of the electrode (Supplementary Fig. S16). Before use, the DSA anodes were cleaned for 10 min in a 1:1 (v/v) mixture of DDW and ethanol in an ultrasonic bath (MRC ultrasonic cleaner, 120 W) and then thoroughly rinsed with DDW. An Ag/AgCl electrode (saturated KCl aqueous solution, 99.9%, Acros) was used as the reference electrode in three-electrode measurements. The reference electrode was immersed in the electrolytic cell 15 min before the measurements were initiated.

Electrode characterization

The microstructure, elemental composition, and phase composition of the DSA anode were characterized before and after the 10 h durability test by scanning electron microscopy (SEM; Zeiss Ultra Plus FEG-SEM), energy-dispersive X-ray spectroscopy (EDS; Oxford X-MAX system equipped with a silicon drift detector with an 80 mm² active area and 127 eV energy resolution), X-ray diffraction (XRD; Rigaku SmartLab diffractometer, 9 kW), and Raman spectroscopy (WITec alpha300 Raman microscope equipped with a 532 nm laser operated at 10 mW and an 1800 g mm⁻¹ grating).

Cathodic and anodic polarization measurements

The cathodic polarization curves shown in Fig. 1, derived from the chronoamperometric data in Supplementary Fig. S1, were measured in an undivided borosilicate glass cell (60 mL, custom-made by the Technion glassblowing workshop) containing 50 mL of a mixed solution composed of DDW and acetonitrile (1:1, v/v), with 0.3 M NaBr or other electrolytes, and with or without 0.02 M styrene. A Pt foil electrode (1 cm²) was used as the working electrode (cathode), a DSA electrode (1 cm²) as the counter electrode (anode), and an Ag/AgCl electrode immersed in saturated KCl solution (99.9%, Acros) as the reference electrode. The reference electrode was positioned close to the working electrode surface to minimize errors arising from solution resistance. These chronoamperometric measurements, as well as the linear sweep voltammetry (LSV) measurements shown in Supplementary Fig. S2, were performed using a potentiostat (BioLogic SP-150) in a three-electrode configuration. The LSV sweep rate was 5 mV s⁻¹. All measured potentials were iR-corrected according to E = E_measured – iR, where i is the current and R is the series resistance, determined before each measurement by either the current-interruption (CI) method or potentiostatic electrochemical impedance spectroscopy (PEIS). Measurements were conducted at room temperature, with or without magnetic stirring at 350 rpm.

The anodic polarization curves shown in Fig. 1 were measured under the same conditions, except that the DSA electrode (1 cm²) served as the working electrode (anode) and the Pt foil electrode (1 cm²) as the counter electrode (cathode).

Electrosynthesis measurements

The electrosynthesis measurements presented in Fig. 2 were performed in an undivided borosilicate glass cell (60 mL, custom-made by the Technion glassblowing workshop) containing 50 mL of a mixed solution composed of DDW and acetonitrile (1:1, v/v), with 0.3 M NaBr and with or without 0.02 M styrene. The initial pH of the solution was 6.7. All measurements were conducted at room temperature under magnetic stirring at 350 rpm. Galvanostatic measurements were carried out using a DSA anode (1 cm²) as the working electrode and a Pt foil cathode (1 cm²) as the counter electrode, with constant applied currents ranging from 5 to 500 mA. The voltage between the two electrodes (cell voltage) was recorded in a two-electrode configuration using a potentiostat (Ivium-n-Stat, Ivium Technologies). Each measurement at a given current was conducted for 30 s, with a 5 s interval allowed for transient relaxation between successive current steps. The series resistance, $R$, was measured before and after electrosynthesis, and the average value was used for iR correction of the cell voltage according to V = V_measured - iR. Each measurement was repeated four times to determine the mean cell voltage and standard deviation. The corresponding cell voltage–time curves, measured with and without styrene, are shown in Supplementary Fig. S3.

The electrosynthesis measurements used to determine anodic Faradaic efficiency, product selectivity, and yield (Table 1) were performed at constant currents ranging from 50 to 500 mA under otherwise similar conditions. The experiments were conducted in 50, 100, or 500 mL water/acetonitrile mixed solvent containing 0.02 M styrene, using a 2 cm² DSA anode and a 1 cm² Pt foil cathode, with a BioLogic SP-150 potentiostat. All measurements were carried out at room temperature under magnetic stirring at 350 rpm.

Hydrogen evolution during the 10 h durability test

Hydrogen evolution was measured during a 10 h durability test using a custom-built sealed electrochemical system designed to quantify the amount of hydrogen generated during operation (Fig. 3a). A three-neck round-bottom flask (100 mL, 14/24/14 NS, Glassco) served as the undivided electrochemical cell and was filled with 100 mL of a mixed solution consisting of 0.3 M NaBr in DDW and 0.02 M styrene in acetonitrile, combined at a 1:1 (v/v) ratio of the aqueous and organic phases. Electrosynthesis was carried out at room temperature under magnetic stirring at 350 rpm. Two necks of the flask were fitted with a 1 cm² DSA anode and a 1 cm² Pt foil cathode, while the third was connected to a glass outlet through which the gas evolved during electrosynthesis was directed to a water-displacement system for continuous monitoring of the effluent gas volume, as described elsewhere²⁵. The water-displacement setup consisted of an inverted dropping funnel filled with water, a glass beaker to collect the displaced water, and a weighing scale connected to a LabVIEW-based data acquisition system. The mass of displaced water was recorded every second during electrosynthesis and converted to the corresponding gas volume, assuming 1 mL per g of water. Because the amount of styrene available in the cell was limited to 2 mmol, the 10 h electrosynthesis test was divided into ten consecutive 1 h segments, with the electrolyte replaced by fresh solution at the beginning of each segment.

During electrosynthesis, a constant current density of 100 mA cm⁻² was applied, and the cell voltage was recorded using a potentiostat (Ivium-n-Stat, Ivium Technologies). The measured voltage was corrected for ohmic losses according to V = V_measured – iR, where R is the series resistance determined by the current-interruption method. Hydrogen bubble evolution was observed at the Pt cathode and within the stirred solution (Supplementary Video 3). In contrast, no gas bubble evolution was observed at the anode. The evolved gas exited the electrochemical cell freely and entered the water-displacement apparatus, where it displaced water from the dropping funnel into a glass beaker placed on a weighing scale (Fig. 3a). The mass of displaced water (m_water) was continuously recorded and converted to the corresponding gas volume using a conversion factor of 1 g (H₂O) = 1 mL gas based on the density of water being approximately 1 kg L⁻¹ under ambient conditions. The measured hydrogen volume ($V_{{\mathrm{H}}_2}$) was then converted to the number of moles of hydrogen ($n_{{\mathrm{H}}_2}$) using the ideal gas law, $PV=nRT$, where $P$ is pressure, $V$ is volume, $n$ is the number of moles, $R$ is the ideal gas constant, and $T$ is temperature.

Gas product analysis

Gaseous samples were swept from the electrochemical cell by Ar gas (99.999%) into 1 L Tedlar sampling bags (Restek) and analyzed by gas chromatography (GC). GC analysis was conducted on a SCION 8500 instrument equipped with a ShinCarbon ST column, thermal conductivity and flame ionization detectors (TCD and FID, respectively), using N₂ (99.999%) as the carrier gas.

Calibration standards were prepared using bottled H₂ and Ar gases (99.999%), both of which were quantified using the TCD with N₂ as the carrier gas.Calibration mixtures containing 0, 25, 50, 75, and 100 mol% H₂ in Ar were prepared immediately prior to analysis using mass flow controllers. For each calibration point, a total gas volume of 100 μL was injected to the column, and three replicate injections were performed. The oven temperature was maintained at 100 °C throughout the analysis. Hydrogen retention time and peak area were recorded for each measurement (Supplementary Fig. S9). The average peak area from triplicate measurements was used to construct the calibration curve. The calibration curve was generated by plotting the averaged peak area as a function of hydrogen concentration (mole fraction). Linear regression analysis was applied to determine the calibration equation and correlation coefficient (R² > 0.995) (Supplementary Fig. S10).

The H₂ concentration in analyte gas samples was determined by using the calibration curves. Gas product analysis was performed after electrolysis at current densities of 100, 300, and 500 mA cm⁻², with total charges (Q = It) of 150, 150, and 180 C, respectively. Prior to electrolysis, the electrochemical cell (100 mL, 14/24/14 NS, Glassco) was sealed and purged with Ar (99.999%) to remove residual air. The gas outlet was connected to a water trap to isolate the electrochemical cell from ambient air (Supplementary Fig. S8). At the beginning of electrolysis, the outlet tube connected to the water trap was closed, and the valve of a 1 L gas sampling bag (Restek) was opened to collect the evolved gas for subsequent GC analysis. During electrolysis, the generated gas was continuously carried into the gas sampling bag with the assistance of an Ar flow (approximately 100 mL/min). After electrolysis, the total gas volume collected in the sampling bag was measured by water displacement. The measurement uncertainty was estimated to be ±8 mL, mainly arising from gas transfer/injection and reading of the volume marker. The collected gas sample, maintained under an Ar atmosphere, was then analyzed by GC using a TCD detector. The integrated area of the hydrogen peak (${Area}_{H2}$) was converted to the H₂ mole fraction using the calibration curve in Supplementary Fig. S10 (${mol\%}_{H2}={Area}_{H2}/580777$). Then, the net hydrogen volume ($V_{netH2}$) was calculated from the total gas volume ($V_{total}$) measured by the water displacement method (Fig. 3a) and the H₂ mole fraction measured by GC-TCD (Supplementary Fig. S11), $V_{netH2}=V_{total}\times {mol\%}_{H2}.$The cathodic hydrogen Faradaic efficiency (${FE}_{H_2}$) was determined by comparing $V_{netH2}$ with the theoretical hydrogen volume ($V_{theoreticalH2}$) calculated from the applied charge ($Q/nF$):

$${FE}_{H_2}=\frac{V_{netH2}}{V_{theoreticalH2}}\times 100\%=\frac{\left({Area}_{H2}/580777\right)\times V_{total}}{Q/nF\times RT/P}\times 100\%$$ 7

where Q is the charge during electrolysis (150, 150, and 180 C for electrolysis experiments at 100, 300 and 500 mA cm⁻², respectively); n = 2 is the number of electrons required to produce one H₂ molecule; and F is the Faraday constant (96485 C mol⁻¹).

Gaseous impurities were analyzed by GC-MS using a SCION 8700 SQ MS system coupled to a SCION 8500 GC and equipped with a Scion-624 MS column, using helium as the carrier gas. The oven temperature was initially maintained at 40 °C and subsequently ramped to 100 °C at 10 °C min⁻¹.

Organic liquid product extraction and analysis

The spent solutions collected from the ten consecutive 1 h segments of the 10 h durability test were combined and transferred to a separatory funnel (500 mL, 29/32 NS, Isolab; Fig. S17). The organic products were extracted using dichloromethane (or chloroform), chosen because they are immiscible with water and have relatively high densities (1.33 and 1.49 g mL⁻¹, respectively), forming the bottom phase. After vigorous shaking and phase separation, the lower organic layer, containing dichloromethane (or chloroform) together with acetonitrile, residual styrene, styrene oxide, and other organic products, was collected in a one-neck round-bottom flask. The aqueous phase was then re-extracted with fresh dichloromethane at least twice. The combined organic extracts were concentrated by removing dichloromethane and acetonitrile under reduced pressure using a rotary evaporator connected to a vacuum pump (MZ 2 C NT, BrandTech). The resulting crude organic residue was redissolved in chloroform-d and transferred to an NMR tube (3.5 mL, WG-5mm-ECONOMY-7, Wilmad). ¹H NMR spectra were recorded using a Bruker Avance NEO 500 MHz spectrometer.

The benchmark parameters of conversion (Table 1) were calculated as follows:

$$\mathrm{Anodic}\mathrm{FE}\%=\frac{\mathrm{mol}\mathrm{of}\mathrm{products}}{it/nF}\times 100\%$$ 8

$$\mathrm{Selectivity}\left(\%\right)=\frac{\mathrm{mol}\mathrm{of}\mathrm{epoxide}}{\mathrm{mol}\mathrm{of}\mathrm{products}}\times 100\%$$ 9

$$\mathrm{Epoxide}\mathrm{yield}\left(\%\right)=\frac{\mathrm{mol}\mathrm{of}\mathrm{epoxide}}{it/nF}\times 100\%$$ 10

where i is the applied constant current (A); t is the electrosynthesis time (s); n = 2 is the number of electrons required to produce one H₂ molecule; and F is the Faraday constant (96485 C mol⁻¹).

To ensure reliable quantitative NMR analysis, quantification was performed using a calibration protocol based on established qNMR methodologies^55,56, according to the standard relation:

$$\frac{n_x}{n_y}=\frac{I_x{N}_y}{I_y{N}_x}$$ 11

where n, I, and N are the mole, integrated signal area, and nuclei number of the compound (x or y), respectively. Toluene was selected as the internal standard due to its chemical inertness and well-separated signals from styrene oxide. A calibration curve was constructed using known molar ratios, yielding high linearity (R² = 0.9982, Supplementary Fig. S24).

Author contributions

G.R. and A.R. conceived the conceptual idea. G.R. designed and performed the experiments. G.R. and A.R. analyzed the data. G.R. and A.R. wrote the manuscript. A.R. supervised and guided the entire project.

Peer review

Peer review information

Nature Communications thanks Jinlong Gong, Nagaraj Shetti, and the other, anonymous, reviewer for their contribution to the peer review of this work. A peer review file is available.

Funding

This project has received funding from the European Research Council (ERC) under the European Union’s Horizon Europe research and innovation program (grant agreement No. 101097966). Views and opinions expressed are, however, those of the authors only and do not necessarily reflect those of the European Union or the European Research Council Executive Agency. Neither the European Union nor the granting authority can be held responsible for them.

Data availability

All data supporting the findings of this study, including numerical values underlying the main and Supplementary Figs. and tables, are provided within the Article and its Supplementary Information. The raw data of Figs. 1–5 generated in this study have been deposited in the Zenodo database under accession code https://doi.org/10.5281/zenodo.21625637.

Competing interests

Technion Research & Development Foundation Ltd. is the applicant for pending U.S. Provisional Patent Application No. 63/715,733, entitled “Systems for Continuous Generation of Hydrogen Gas and Value-Added Chemicals,” naming Guilin Ruan and Avner Rothschild as inventors. The application covers systems and methods for the continuous co-production of hydrogen and value-added chemicals, including the bromide-mediated membraneless paired electrosynthesis of hydrogen and epoxides described in this manuscript. The authors declare no other competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-76617-x.