Introduction

Metal hydrides play important roles in a wide range of energy and catalysis processes1,2. The ability to induce a concerted and efficient hydrogen atom transfer (HAT) process renders metal hydrides a hot spot in numerous catalytic reactions involving hydrogenation, hydrosilylation, and so on3,4,5. The advantage of being a hydride can be traced back to abundant enzyme activities in nature, which have evolved over billions of years to achieve the highest efficiency6. Countless energy transfer processes in enzyme activities involve metal hydrides as intermediates, coupled with complex proton and electron transfer7 (Fig. 1a). To date, various strategies have been reported for constructing artificial metal hydride centers that resemble enzymatic sites to realize robust chemical reactions8,9,10. Generally, the generation of intermediate metal hydrides requires highly reductive hydrogen donors11,12 (Fig. 1b). Toxic silanes, phosphines, and boranes are commonly used in tightly controlled environments. Otherwise, high-pressure gaseous hydrogen was adopted13,14. Precisely designed organic ligands (e.g., thiolate and cyclopentadienyl) have been applied to provide a specific local environment, ensuring consecutive hydride formation and transformation1,2,12. However, the sensitivity of the ligands to air and moisture results in great restrictions on the synthetic conditions. An alternative way to produce metal hydrides through a catalytic manner by adopting mild proton donors such as water and alcohols is much preferred (Fig. 1b). These processes are generally photocatalytic or electrocatalytic and meet the goal of being green and sustainable for synthetic chemistry10,15,16,17. Nevertheless, well designed mediators are required to accelerate the electron and proton transfer processes17. These delicate molecular structures are usually difficult to reproduce for high-throughput production at considerable current densities, and deliberate control of the proton content is necessary to minimize hydrogen evolution18,19. Separation and recycling issues also hinder their practical application20. Thus, constructing heterogeneous metal hydride sites that can potentially overcome these drawbacks remains a challenge.

Fig. 1: Pressure-induced generation of heterogeneous electrocatalytic metal hydride surfaces for sustainable hydrogen transfer.
Fig. 1: Pressure-induced generation of heterogeneous electrocatalytic metal hydride surfaces for sustainable hydrogen transfer.The alternative text for this image may have been generated using AI.
Full size image

a Representative enzymatic metal hydride sites in nickel-iron hydrogenases. The shadow image representing a typical nickel-iron hydrogenase was plotted using the structure deposited in the Protein Data Bank (ID: 5ADU). Hydride formation generally involves complex electron transfer channels (e.g., chains of iron-sulfide clusters) and proton transfer channels (e.g., various amino acids). b Representative synthetic strategies for constructing homogeneous metal hydrides. A molecular metal hydride site is generally obtained by directly reacting precursors with toxic and reductive silanes, phosphines, or boranes. Recently, mediator-mediated hydride generation via photo- or electro-catalytic routes was also reported. c Representative synthetic strategies for fabricating bulk metal hydrides. The synthesis of bulk metal hydrides generally requires elevated temperature and pressure to facilitate hydrogen (H2) dissociation and penetration. In addition, mechanochemical ball-milling methods adopting lithium or barium hydrides as precursors have also been reported. d Schematic illustration of the sustainable generation of heterogeneous electrocatalytic metal hydride surfaces via a facile high-pressure electrosynthesis methodology. More importantly, the obtained metal hydride surface is stable under ambient conditions and thus enables ambient-pressure electrocatalytic hydrogen processes.

Although several bulk metals or alloy hydride systems, such as lithium-, magnesium-, palladium-, lanthanum-, and titanium-based hydrides, have been developed under elevated pressures and widely applied in hydrogen storage (Fig. 1c)21,22,23,24,25, the reserved interstitial hydrogen of these bulk hydrides tends to form gaseous hydrogen rather than react with substrates during catalytic reactions26. To this end, few bulk hydride systems have been explored for catalytic applications. Very recently, alkali earth metal hydrides have been reported to have great potential in ammonia synthesis27,28. The participation of metal hydrides significantly reduced the temperature and pressure required for hydrogenation processes. However, high-energy inputs are still needed for hydride synthesis, which involves multistep chemical looping procedures for the regeneration of surface hydrides during operation. The ability to form and consume hydrides in situ in an electrocatalytic manner under ambient conditions while suppressing hydrogen evolution renders palladium the only ideal heterogeneous metal-hydride-mediated hydrogenation system to date29,30,31. The lack of elegant methods for achieving the fabrication of heterogeneous metal hydrides in a sustainable manner severely limits their application in catalytic chemistry and hinders the discovery of new catalytic mechanisms.

Herein, we demonstrate a universal strategy for generating a cost-effective heterogeneous electrocatalytic metal hydride surface by using a pressure-induced methodology (Fig. 1d). After pressure is introduced into the electrolysis process, surface-adsorbed hydrogen (H*) produced via the Volmer step is found to penetrate into the interstitial sites of the metal electrode. Surprisingly, the obtained metal hydride surface remained stable after depressurization and served as an efficient catalyst for subsequent electrocatalytic processes under ambient conditions. As a proof-of-concept, we demonstrated the formation of a zinc hydride (ZnHx) surface after high-pressure electrolysis by surface-sensitive mass spectrometry analysis. We further showcased the capability of the ZnHx surface in breaking the traditional catalyst design rules and bringing alternative mechanisms for carbon dioxide (CO2) electroreduction. We observed that upon the formation of the ZnHx surface, the well-known carbon monoxide (CO)-selective Zn gradually shifted to a formic acid (HCOOH)-selective catalyst with significantly increased intrinsic activity. The maximal HCOOH Faradaic efficiency of Zn was enhanced by ~fivefold to 83% under ambient conditions. Detailed mechanistic studies revealed that, unlike the distal hydrogenation route for CO2 to CO and hydrocarbons on pristine Zn catalysts, pressure-induced formation of the ZnHx surface enabled direct surface hydrogenation at the carbon site of CO2 to form formate (HCOO-) through a concerted homogeneous-type HAT pathway. This distinct hydrogenation behavior led to an unconventional inverse kinetic isotope effect, as revealed by isotope-labeled measurements, which well explained the high intrinsic reactivity of the hydride surface. Furthermore, by performing isotope cross-labeling experiments, we directly proved the participation of subsurface hydrogen in the reaction and illustrated the renewability of the hydride surface with hydrogen generated from the Volmer process. We also observed the same switch in hydrogenation behavior over other CO-selective metals, such as nickel, silver, and gold, demonstrating the wide applicability of our strategy.

Results

ZnHx surface synthesis and characterization

Polycrystalline Zinc (Zn) foil (≥99.99%, 1 centimeter × 1 centimeter) was used as the electrode and was electrolyzed at the cathode side in an H-type cell pressured with 60 bar of argon (Ar) under −100 milliampere per square centimeter (mA cm−2) for 1 h (Supplementary Fig. 1). After the facile high-pressure charging process, the ZnHx surface was formed, and the obtained foil was denoted as Pressure-Zn (see details in the supplementary information). Compared with the smooth surface of the pristine Zn foil, the pressure-Zn foil exhibited obvious surface reconstruction with cracks and voids (Supplementary Fig. 2a, b). Figure 2a, b shows high-resolution transmission electron microscopy (HRTEM) images of scratched surface pieces of pristine Zn foil and Pressure-Zn. Lattice fringes were observed for pristine Zn and Pressure-Zn foils with spacings of 2.47 and 2.49 Å, respectively, corresponding to the Zn (002) facet. Notably, the oxygen susceptibility of Zn leads to spontaneous surface oxidation in the presence of air or moisture, which is well known and inevitable during sample preparation, even inside a glovebox32,33,34. We thus observed trace zinc oxide (ZnO) phases on the surfaces and simultaneously the same lattice expansion of Pressure-Zn for the ZnO (100) facet over pristine Zn (Supplementary Fig. 3). The expansion of the lattice can also be evidenced by the shift of the X-ray diffraction (XRD) peaks to smaller angles (Fig. 2c-d). Lattice expansion can generally be induced by the incorporation of heteroatoms in a multicomponent system or twin structures in a single-metal system. However, we did not observe apparent twin structures in the HRTEM images (Fig. 2a, b, Supplementary Fig. 3), indicating that this expansion should be ascribed to the former. In our case, hydrogen (H), oxygen (O), chloride (Cl), and potassium (K) atoms are potential candidates for the incorporated heteroatoms. O and Cl atoms are in the form of hydroxyl (OH−) and chloride anions (Cl−), respectively, which means that they are repulsive from the cathode side, as influenced by the electric field. Alkali ions such as K+ are hydrated in aqueous solutions and physisorbed on the cathode35,36. Chemisorption of alkali ions on the surface of the electrode, including charge transfer, is unlikely owing to their very negative reduction potentials, let alone their incorporation into the bulk phase37,38. In contrast, water (H2O) molecules or hydroniums (H3O+) dissociate during electrolysis and produce surface-adsorbed H* species with electron transfer through the well-known Volmer process39,40. Therefore, it is rational to ascribe the expansion to the penetration of surface-adsorbed H*.

Fig. 2: Synthesis of the ZnHx surface and characterization.
Fig. 2: Synthesis of the ZnHx surface and characterization.The alternative text for this image may have been generated using AI.
Full size image

a HR-TEM image of pristine Zn foil. b HR-TEM image of a Pressure-Zn foil. c XRD patterns of pristine Zn and Pressure-Zn foils. d Enlarged XRD patterns indicating the shift of the peak. e TOF-SIMS analysis of pristine Zn and Pressure-Zn foils. f TPD-MS analysis of pristine Zn and Pressure-Zn foils. H2 signals (m/z = 2) were detected. Solid lines were drawn to guide the eyes. Source data for this figure are provided in the Source Data file.

Since it is difficult to experimentally observe the interstitial hydrogen of hydride surfaces by electron microscopy41, surface-sensitive mass spectrometry-based techniques were employed. We conducted time-of-flight secondary ion mass spectrometry (TOF-SIMS) to qualitatively identify the (sub)surface hydrogen. To acquire sufficient hydrogen signal responses for analysis, a negative ion detection mode was applied as generally reported42,43,44. However, the detection of metallic Zn components under a negative ion detection mode is unfavorable. In addition, hydroxyl species, as mentioned above, also contribute to the hydrogen signal. To circumvent this dilemma and interference, we tracked the hydrogen content as a function of depth through the H−/ZnO− ratio. Since the presence of hydrogen species beyond the metal hydride is strictly limited in the form of hydroxyl groups, a fixed ratio between the hydrogen and the corresponding zinc oxide can be expected, considering the high mobility and thus high uniformity of the protons in the oxide matrix45,46. As shown in Fig. 2e, the H−/ZnO− ratio for the pristine Zn foil remains almost unchanged at approximately 0.2 when the sputtering time increases, indicating that hydrogen stably coexists with oxygen-related species. For the Pressure-Zn foil, the value of the H−/ZnO− ratio resembles that of the pristine Zn foil for the outermost layer and gradually increases to a value of ~5.0 when the sputtering time increases. The change in the H−/ZnO− ratio demonstrates the existence of non-hydroxy hydrogen species inside the subsurface layers. In addition, a distinct peak at ~325 °C for gaseous hydrogen with m/z = 2 was observed via temperature-programmed desorption-mass spectrometry analysis of the Pressure-Zn foil. In contrast, no obvious peaks were observed for the pristine Zn foil (Fig. 2f). The release of gaseous hydrogen at an elevated temperature further confirmed the existence of a hydride surface in the Pressure-Zn foil. We then sought to perform quantitative elemental analysis by thermal desorption techniques, which gave O/H atomic ratios of 1:1.2 and 1:1.5 for pristine Zn and Pressure-Zn foil, respectively, verifying a higher hydrogen content for the Pressure-Zn foil. Since it is difficult to quantify the absolute H and Zn contents in the ZnHx layer, we estimated a stoichiometric ZnH0.19 phase on the basis of the strain analysis established in the Pd-H system (Supplementary Note 1). We also performed X-ray photoelectron spectroscopy (XPS) measurements and deconvolutions to explore the surface chemical information for the two foils. As shown in Supplementary Fig. 4, metallic Zn0 and oxidized Zn2+ co-existed over pristine Zn foils, and metallic Zn0 was present in greater proportions. In stark contrast, Pressure-Zn exhibited a greater proportion of the oxidized Zn2+ component. A comparison of the O 1 s spectra revealed a similar or slightly decreased proportion of lattice oxygen over Pressure-Zn compared with that of the pristine Zn foil. This result indicated that the increased oxidization over Pressure-Zn should not be caused by a higher oxygen content. Therefore, we ascribe the increased oxidization to electron transfer from Zn to H, which is also observed in palladium hydride (PdHx) and Titanium hydride (TiHx) systems29,47,48.

Indeed, the phenomenon of hydrogen penetration has been recognized and studied as a hot topic for centuries in the field of electroplating and corrosion science49,50. The penetrated hydrogen fosters the formation and propagation of cracks, leading to catastrophic failure in terms of mechanical strength. However, in the field of catalysis, to our knowledge, the hydrogen penetration phenomenon has only been considered for palladium and titanium because of the inevitable formation of stable palladium and titanium hydrides, whereas other metals or metal hydrides are rarely reported to have similar properties51,52. Nevertheless, these metal hydrides generally show enhanced activities in hydrogen evolution, rendering the palladium system the only one capable of hydrogenation53,54. The lack of effective approaches to increase the hydrogen solubility and simultaneously stabilize the metastable hydride surface greatly hinders the exploration of unknown phenomena. Although few studies in corrosion fields have revealed the capability of cathodic hydrogen charging and storage over metallic Zn or Zn alloys55,56 and some have indicated the facilitation of hydrogen diffusion into the metallic matrix under high hydrostatic pressure over iron-based metals57, the catalytic advantages of these hydride materials remain unexplored.

Hydride-induced selectivity switching in CO2 electroreduction

In CO2 electroreduction, the production of CO or HCOOH is strictly governed by the adsorption configuration of CO2 molecules, i.e., the specific metal category (silver, gold, zinc for CO, and lead, bismuth, indium for HCOOH)58. Jumping out of existing categories and turning a CO-selective catalyst into an HCOOH-selective catalyst is essentially impossible. However, hydride catalysts may violate traditional catalyst design rules27, offering an alternative method for selectivity regulation. As a representative case study, we compare the CO2 electrochemical reduction performance of bulk Zn with or without the ZnHx surface. As shown in Fig. 3a and Supplementary Fig. 5a, pristine Zn exhibited an intrinsically high selectivity for CO of over 70% under ambient conditions, in accordance with previous reports and traditional rules59,60,61. In stark contrast, Pressure-Zn switched its selectivity towards HCOOH with a Faradaic efficiency (FEHCOOH) greater than 80% (Fig. 3b and Supplementary Fig. 5b). Compared with the pristine Zn foil, the Pressure-Zn foil exhibited a greatly increased HCOOH partial current density and suppressed CO partial current density (Fig. 3c and Supplementary Figs. 5c-f). In addition, we found that a high current density and high pressure are simultaneously needed for generating hydride surfaces and initiating such a selectivity switching; otherwise, CO remains the dominant product (Fig. 3d, e, Supplementary Figs. 6a–d). This result was consistent with the structural evaluations of the spent pristine Zn foil after ambient CO2 electroreduction (Supplementary Fig. 7). Although the penetration of hydrogen also occurs during ambient CO2 electrolysis, in contrast to Pressure-Zn foil, a significantly lower hydrogen content and less-strained surface were observed for the pristine Zn foil, which could be the reason for its low selectivity for HCOOH production. We also observed the stable maintenance of a significant content of hydrogen over spent Pressure-Zn after ambient CO2 electrolysis (Supplementary Fig. 8). We suspect that the synergetic effect of high hydrostatic pressure and high current density can be the driving force for deep hydrogen penetration. As revealed by previous studies in corrosion sciences, high hydrostatic pressure can directly regulate the surface chemistry of H species by inducing a decrease in the thickness of the Helmholtz layer, facilitating the adsorption of hydrogen atoms, and restraining the hydrogen combination57,62,63. Moreover, a high current density guarantees sufficient hydrogen coverage at the surface. This speculation was further supported by a clear structure-selectivity relationship, as shown in Fig. 3f. Under an external CO2 pressure of 60 bar, the selectivity of the pristine Zn foil gradually switched from CO to HCOOH at higher current densities (Supplementary Figs. 6e, f and 9). After depressurization to ambient conditions, the high-pressure CO2 electrolyzed Zn foil at a high current density retained a similar selectivity of ~80% for HCOOH (Fig. 3g and Supplementary Figs. 6g, h), indicating the irrelevance of hydride surface formation on the gas type. Moreover, we did not observe a further increase in HCOOH selectivity by increasing the applied pressure and current density to 70 bar and −120 mA cm-2, respectively, which might be attributed to the intrinsically high activity of the inherent structures (such as grains) towards CO or H2 (Supplementary Fig. 10).

Fig. 3: Ambient and high-pressure CO2 electrolysis.
Fig. 3: Ambient and high-pressure CO2 electrolysis.The alternative text for this image may have been generated using AI.
Full size image

a FEs of 1-bar CO2 electrolysis over pristine Zn foil. b FEs of 1-bar CO2 electrolysis over Pressure-Zn foil obtained after 60-bar Ar electrolysis at −100 mA cm−2 for 1 h. c Partial current density of HCOOH and CO over pristine Zn and Pressure-Zn foils at 1 bar of CO2. d FEs of 1-bar CO2 electrolysis over Pressure-Zn foil obtained after 60-bar Ar electrolysis at −10 mA cm−2 for 1 h. e FEs of 1-bar CO2 electrolysis over Zn foil obtained after 1-bar Ar electrolysis at −100 mA cm−2 for 1 h. f FEs of 60-bar CO2 electrolysis over pristine Zn foil. g FEs of 1-bar CO2 electrolysis over Pressure-Zn foil obtained after 60-bar CO2 electrolysis at −100 mA cm−2 for 1 h. h 200-h stability test at −10 mA cm−2 for Pressure-Zn foil at 1 bar of CO2. The Pressure-Zn foil obtained after 60 bar Ar electrolysis at −100 mA cm−2 for 1 h was adopted for the stability test. 100% iR correction was applied to report the applied potential. The solution resistance (R) for (a–c) can be found in figure caption of Supplementary Fig. 5. The solution resistance (R) for (d–g) can be found in figure caption of Supplementary Fig. 6. The solution resistance (R) for (h) is around 4.5 Ω. Corresponding voltammograms without iR correction are also provided in Supplementary Figs. 5 and 6. The error bars correspond to the standard deviation of three independent measurements. Source data for this figure are provided in the Source Data file.

When we employed sodium chloride (NaCl) or potassium bicarbonate (KHCO3) instead of potassium chloride (KCl) as the electrolyte during the high-pressure electrolysis process, the obtained Pressure-Zn foils still maintained >70% HCOOH selectivity (Supplementary Fig. 11), further confirming that chloride or potassium incorporation is not the key reason for the observed lattice expansion and selectivity switching. We also show that the intrinsically high formate selectivity of Pressure-Zn is insensitive to cations or anions during ambient CO2 electrolysis (Supplementary Fig. 12). It is noteworthy that the surface zinc oxide/hydroxyls we have observed by ex situ HRTEM and TOF-SIMS analysis do not directly participate in the catalytic process, as well-established previously by various in situ techniques and calculations that under CO2 electroreduction conditions, surface or bulk ZnO is readily reduced to metallic Zn64,65,66,67. In addition, the standard reduction potential of Zn/Zn+ is -0.53 V vs. RHE under the reaction conditions, further ruling out the existence of ZnO species on the electrode surface68. Such a standard reduction potential is more positive than the potential range we studied, further indicating the metallic nature of surface Zn for CO2 conversion. As shown in our results, pristine Zn exhibited intrinsically high CO selectivity, which follows this trend well. On the other hand, the Pressure-Zn foil retained a high selectivity towards HCOOH. We ascribe this result to the underlying hydride skeleton, which facilitates the transformation of the freshly reduced Zn surface to ZnHx with abundant flexible H atoms, ensuring the integrity of the hydride surface. We also ruled out the influence of changes in surface physical properties such as porosity and hydrophobicity on the selectivity switching with systematic studies (Supplementary Figs. 13–21, Supplementary Table 1, Supplementary Notes 2–3).

Furthermore, a long-term stability test of the Pressure-Zn foil at a conversion rate of −10 mA cm−2 under ambient conditions was performed. The Pressure-Zn foil maintained an FEHCOOH of over 80% after 200 h of electrolysis, verifying the tenability of the selectivity switching (Fig. 3h, Supplementary Fig. 22, and Supplementary Note 4). Moreover, a total amount of 62.4 mmol of HCOOH was produced during the stability test, indicating the catalytic nature of the electrochemical reaction. Otherwise, only stoichiometric production of 4.4 mmol of HCOOH is achieved if the zinc hydrides are irreversibly consumed, even if we assume an exaggerated atomic Zn:H ratio of 1:2 (Supplementary Note 5). We also demonstrated a 5-h-long selective production of HCOOH at a higher current density of -100 mA cm-2 with a sufficient CO2 supply of 60 bar, further demonstrating the robustness of the Pressure-Zn catalyst (Supplementary Fig. 23, Supplementary Note 4).

Mechanistic insight into the hydride-induced selectivity switching

We thoroughly investigated the reaction mechanism and kinetics to provide insights into the regulatory rules behind the unconventional selectivity switch. We conducted in situ Raman measurements to reveal the surface intermediates for CO and HCOOH during electroreduction at different potentials (Supplementary Fig. 24). For the pristine Zn foil, a broad band at approximately 2100 cm−1 corresponding to C ≡ O stretching was observed, and no peaks corresponding to C-H stretching were found in the range of 2800–3000 cm−1 (Fig. 4a, b and Supplementary Fig. 25)69,70,71. This result is consistent with observations over other reported CO-selective catalysts and implies the formation of CO through a carbon-bound COOH* intermediate72,73. In sharp contrast, for the Pressure-Zn foil, no bands were detected for C ≡ O stretching, while a significant band at approximately 2850 cm−1 assigned to the C-H stretching vibration arose at −0.7 V vs. the reversible hydrogen electrode (RHE) (Fig. 4c, d, Supplementary Fig. 25). This observation suggests that CO2 undergoes a protonation step at the carbon site rather than at the oxygen site to form HCOO−, and an OCHO* intermediate can be inferred. As such, we concluded that COOH* and OCHO* serve as intermediates for CO and HCOOH production over pristine Zn and Pressure-Zn foils, respectively. The hydride surface is assumed to steer the reaction pathway and switch the selectivity of Zn.

Fig. 4: Mechanistic insights into different hydrogenation pathways.
Fig. 4: Mechanistic insights into different hydrogenation pathways.The alternative text for this image may have been generated using AI.
Full size image

a, b In situ Raman spectra of pristine Zn foil. Dashed lines were drawn to guide the eye. c, d In situ Raman spectra of the Pressure-Zn foil. Arrows were drawn to guide the eye. 100% iR correction was applied to report the applied potential. The solution resistances (R) of the in situ Raman tests were around 30 Ω and 25 Ω for pristine Zn and Pressure-Zn foil, respectively. e Tafel plots for CO over pristine Zn and Pressure-Zn foils. f Tafel plots for HCOOH over pristine Zn and Pressure-Zn foils. g 3rd harmonics generated from FTacV measurements or simulations for pristine Zn and Pressure-Zn foils. The potentials are not iR-corrected. h Schematic illustration of the electron transfer and chemical reaction steps deduced from FTacV simulations. The value of the formal potential (E0, vs. Ag/AgCl saturated) is derived by simulations in (g). Source data for this figure are provided in the Source Data file.

We also carried out kinetic analysis to further understand the reaction pathway for Pressure-Zn. As shown in Fig. 4e, pristine Zn and Pressure-Zn foils exhibited comparable Tafel slopes of 243 and 259 millivolts per decade (mV dec−1) for CO formation, respectively, thereby indicating that they share a similar rate-limiting step (RLS) in the initial CO2 activation. It should be noted that the slopes (~250 mV dec−1) were larger than the frequently mentioned Tafel slope of 118 mV dec−1 for CO2 activation RLS, which could be attributed to the numerical difference in the symmetry factor (β = 0.5 for 118 mV dec−1) due to the more complicated electron transfer and chemical processes under realistic conditions (Supplementary Table 2, see supplementary information for more discussion). On the other hand, a distinct difference in the Tafel slope was observed for the HCOOH pathway. The pristine Zn foil has a Tafel slope of 219 mV dec−1, whereas that of the Pressure-Zn foil is 145 mV dec−1, suggesting a different reaction pathway for the Pressure-Zn foil (Fig. 4f). To uncover the underlying electron transfer processes, we sought to perform Fourier transform alternating current voltammetry (FTacV) experiments (Fig. 4g, Supplementary Figs. 26–29; see the supplementary information for a detailed discussion)74,75. The experiments were performed under either Ar-saturated conditions or deuterated conditions to exclude the potential influence of hydrogen evolution, and fortunately, the H2 selectivity of the Zn-based catalysts was quite low (Supplementary Figs. 26–28). Typically, a threefold peak with good symmetry would be detected if a reversible one-electron transfer process occurs (Supplementary Fig. 29), as reported previously74. In our case, a coupled fourfold peak was acquired over the pristine Zn foil, and a centralized threefold peak was detected over the Pressure-Zn foil. To rationalize these findings, we speculate that the peaks can be attributed to the coupling of two consecutive electron transfer-chemical reaction steps, and our simulation results are in line with our experimental results (Fig. 4g, h, Supplementary Table 3; see the supplementary information for a detailed discussion). According to the simulations, pristine Zn and Pressure-Zn demonstrated very similar formal potentials (E0) for the first electron transfer step (−1.192 vs. −1.190 V), and the E0 for the second electron transfer step of Pressure-Zn was more positive than that of the pristine catalyst (−1.194 vs. −1.267 V). Notably, the 73-mV variation in the formal potential is sufficient to reflect different pathways and demonstrates that the hydride surface promoted intrinsic electron transfer kinetics76,77. Together, the Tafel and FTacV analyses indicated an improved and concerted electron transfer process for HCOOH production on Pressure-Zn foil once CO2 was activated.

Theoretical simulations further highlighted the superiority of the hydride surface. Three types of Zn(001) surfaces were compared: an ordinary clean surface, a surface with 1% lattice tensile strain, and a surface with 1% strain and subsurface hydrogen (Hsub) (Fig. 5a, Supplementary Data 1). As revealed by previous works58 and the mechanistic studies in this work, carboxyl intermediate (COOH*) and formate intermediate (OCHO*) are key intermediates that lead to the production of CO and HCOOH, respectively. For the CO pathway, the oxygen atom in the adsorbed CO2 molecule takes a proton from water and forms COOH* (Supplementary Fig. 30). However, the production of HCOOH complies with the OCHO* pathway, where the carbon atom instead of the oxygen atom in CO2 is hydrogenated (Fig. 5b). In our case, since CO2 is less viable to directly react with water to form OCHO* due to steric hindrance78, the hydrogenation process is more likely to occur through surface hydrogenation: an adsorbed H* is first formed through a Volmer reaction and then reacts with adsorbed CO2* to form OCHO*. Therefore, it can be deduced that the CO pathway is mainly limited by the formation of COOH*, while the formation of H* regulates the HCOOH pathway. Figure 5c compares the limiting energies of the two pathways on the three Zn (001) surfaces. The formation of CO dominated over that of HCOOH for both unstrained and strained Zn (001) in the absence of Hsub. Upon the introduction of sufficient Hsub, however, the OCHO* pathway becomes more apparent, and its selectivity against CO is inverted, demonstrating the importance of Hsub rather than pure lattice strain to induce the shift in Zn selectivity. The free energy diagram for the formation of HCOO− under the limiting potential of strained Zn (001) with Hsub is further shown in Fig. 5d, indicating that the presence of Hsub not only tunes the reaction path but also lowers the barrier for HCOOH formation. Once H* forms, it can readily hydrogenate C atoms in the adsorbed and negatively charged CO2 molecules and produce HCOO−. This result is consistent with the Tafel and FTacV analyses, further confirming a homogeneous-type HAT process. The revealed reaction pathway also affects the kinetic behaviors for the Pressure-Zn catalyst, which can be demonstrated by an inverse kinetic isotope effect (iKIE, kH/kD = 0.97) or an inverse equilibrium isotope effect (iEIE, kH/kD = 0.45) for HCOOH production79,80 (Fig. 5e, Supplementary Fig. 31, Supplementary Table 4; see the supplementary information for a detailed discussion). By replacing H2O with deuteroxide (D2O) in CO2 electroreduction under ambient conditions, an iKIE value of 0.4 was observed for HCOOH production on Pressure-Zn foil at −0.85 V (Fig. 5f), whereas a normal KIE was observed for CO production (kH/kD = 2.0) and hydrogen evolution (kH/kD = 3.8). These results strongly support the proposed direct surface hydrogenation mechanism and highlight the critical role of the ZnHx surface.

Fig. 5: Theoretical insights into the reaction pathways and isotopic evidence.
Fig. 5: Theoretical insights into the reaction pathways and isotopic evidence.The alternative text for this image may have been generated using AI.
Full size image

a Constructed models for an ordinary clean surface, a surface with 1% lattice tensile strain, and a surface with 1% strain and Hsub. b Schematic illustration of the possible hydrogenation pathway. Cool gray, red, white, pastel gray, and purple spheres represent Zn, O, H, C, and K atoms, respectively. Specifically, green spheres represent H atoms from water to form COOH*, and orange spheres represent H atoms from the ZnHx surface to form OCHO*. c Comparison of the limiting energies for the HCOOH and CO pathways on the three Zn(001) surfaces. d Free energy diagram of HCOO− formation under the limiting potential of strained Zn(001) with Hsub. e Qualitative energy level diagrams illustrating the inverse isotope effect. f Isotopic study over a Pressure-Zn foil at −0.85 V vs. RHE. 100% iR correction was applied to report the applied potential. The solution resistances (R): H2O, 2.9 ±  0.1 Ω; D2O, 4.2 ±  0.2 Ω. The error bars correspond to the standard deviation of three independent measurements. Source data for this figure are provided in the Source Data file.

We further performed isotope cross-labeling experiments to prove the participation of surface hydrides in the catalytic reaction with the help of 1H nuclear magnetic resonance (NMR) spectroscopy. As reported previously, 1H-NMR spectroscopy enables quantitative detection of the produced HCOO- species but is blind to the deuterated formate (DCOO-) counterparts81. As demonstrated in Fig. 6a, we obtained Pressure-Zn foil by conducting high-pressure electrolysis in H2O. The obtained Pressure-Zn foil charged with H was then thoroughly purged with Ar to remove surface adsorbed water. After that, we conducted CO2 electrolysis in the deuterium oxide (D2O) phase over the H-charged Pressure-Zn foil to determine whether HCOO- species were produced.

Fig. 6: Renewable ZnHx surface evidenced by isotope cross-labeling experiments.
Fig. 6: Renewable ZnHx surface evidenced by isotope cross-labeling experiments.The alternative text for this image may have been generated using AI.
Full size image

a Schematic illustration of the determination of the participation of hydrides during CO2 electrocatalysis by isotope labeling and 1H NMR analysis. b 1H NMR spectra acquired after ambient electrolysis at a constant conversion rate of -5 mA over Pressure-Zn foil, which is charged with interstitial H under high-pressure electrolysis. From top to bottom: electrolysis conducted in the CO2/H2O phase for 1 h with the accumulation of 16 scans, electrolysis conducted in the CO2/D2O phase for 1 h with the accumulation of 16 scans, electrolysis conducted in the CO2/D2O phase for 1 h with the accumulation of 128 scans, electrolysis conducted in the CO2/D2O phase for 2 h with the accumulation of 128 scans, and electrolysis conducted in the Ar/D2O phase for 1 h with the accumulation of 128 scans. c Amount of formate detected by 1H NMR and ion chromatography. Source data for this figure are provided in the Source Data file.

Prior to the experiment, three patterns for HCOOH production on the hydride surface were expected (Fig. 6a). If the hydride species were irreversibly consumed, which is accompanied by the hopping of Hsub to the surface site31,81, we can anticipate the observation of the characteristic peak assigned to HCOO- at 8.4 ppm17 and a selectivity switching from HCOO- to CO after the hydride was completely consumed. In contrast, an intrinsically stable or dynamically stable hydride surface enables the sustainable production of HCOOH over a long period. If the hydride structure is intrinsically stable, which means that the interstitial hydrogen is too inert to participate in the reaction, only the surface adsorbed deuterium (D*) generated by the Volmer process participates in formate production, and no peaks should be observed at 8.4 ppm in the 1H-NMR spectra. The sustainable production of formate together with the observation of HCOO- in the 1H-NMR spectra can only be achieved on a dynamically stable hydride surface, where D* species generated via the Volmer process are capable of penetrating into the interstitial site to refill the vacancies generated by reacted H species.

The results of the experiments were in line with our inference of a dynamically stable hydride surface. We conducted CO2 electrolysis at a moderate current of -5 mA to acquire a significant amount of HCOO- for analysis. Notably, CO2/H2O phase electrolysis over Pressure-Zn foil for 1 h produced a total HCOO- amount of 76.13 μmol, corresponding to an FEHCOOH of 81.6%, as determined by ion chromatography and 1H NMR spectroscopy (Fig. 6b and c). A significant peak assigned to HCOO- can be observed at 8.4 ppm (Supplementary Fig. 32). However, only a small peak can be observed in the 1H-NMR spectrum when electrolysis is conducted in the CO2/D2O phase. After increasing the scan number of 1H NMR spectroscopy (from 16 to 128), we acquired a clear view of this peak, clearly demonstrating the participation of charged internal H in the production of HCOO-. The amount of HCOO- produced was calculated to be 0.24 μmol, which is significantly less than the total amount of HCOO- and DCOO- of 87.00 μmol determined by ion chromatography, corresponding to an FEtotal formate of 91.3%. This result fits well with our previous discussion on the KIE effect over Pressure-Zn, in which a D2O environment facilitates formate generation (Fig. 5e, f, Supplementary Fig. 31, Supplementary Table 4). Further extending the electrolysis time to 2 h in the CO2/D2O phase did not change the amount of HCOO- obtained, as revealed by the 1H-NMR spectrum. Therefore, at least 0.24 μmol H is charged for Pressure-Zn and contributes to HCOOH formation. In addition, the total FEformate retained a high value of 91.0% with a total formate amount of 169.74 μmol, suggesting the continuous production of formate after the consumption of stored H and thus the gradual refilling of D species into the subsurface. We also performed ambient electrolysis in the Ar/D2O phase to eliminate the interference of impurities, and no peaks were observed at approximately 8.4 ppm. Of note, the observed facile renewability or self-healing ability of the hydride surface ensures its stability for long-term catalytic hydrogenation82,83, as revealed by the durability test (Fig. 3h, Supplementary Fig. 23). The as-revealed mechanism can also be categorized as Mars-van Krevelen-type. Although the Mars-van Krevelen-type mechanism has been reported for various electroreduction reactions catalyzed by hydrides, mostly PdHx and TiHx31,81, these reports focused mainly on enhancing catalytic activity by offering alternative ways for hydrogen supplementation. Our work not only introduced a kind of cost-effective ZnHx surface beyond Pd- and Ti-systems for catalyzing electroreduction via the Mars-van Krevelen mechanism but also revealed its ability to regulate selectivity. These results provide guidance for future catalyst design and reaction control.

Discussion

Taken together, our experimental and theoretical results herein showcase the pressure-induced generation of a heterogeneous electrocatalytic metal hydride surface for sustainable hydrogen transfer and demonstrate its ability to lead to an unconventional selectivity switching in CO2 electroreduction. We further demonstrated the feasibility of extending high-pressure treatment as a universal method for generating heterogeneous electrocatalytic metal hydride surfaces. For example, nickel (Ni) has also been previously identified as a typical catalyst that goes through the COOH* pathway and suffers from severe CO poisoning84,85. The inability of hydrogenation on the carbon site of strongly adsorbed CO molecules/COOH* intermediates prohibits the production of HCOOH and thus leads to a dominant HER during electrolysis. However, after an Ar electrolysis process under 60 bar pressure, we observed an increase in HCOOH selectivity from ~2% to ~23% over that of the nonoptimized bulk Ni foil catalyst (Supplementary Fig. 33). For other CO-selective metals, such as Ag and Au, which have lower hydrogen capacities, a selectivity switching is also observed by providing an in situ high-pressure environment (Supplementary Figs. 34–36, Supplementary Note 6). Together with contemporaneous works reported recently, we believe there is more to be achieved in developing efficient catalysts for sustainable chemical production via protocols such as pressure methodologies and phase engineering strategies86,87,88,89,90.

More importantly, the underlying hydride-mediated HAT process we illustrated here enables hydrogenation pathways that are considered “prohibited” by the substrate adsorption configuration and offers opportunities for selectivity regulation. In addition, the hydride surface can also serve as a good platform for electrocatalytic reactions involving multistep electron and proton transfers and targeting deep hydrogenation products (Supplementary Fig. 37). We believe that our strategy can potentially be coupled with traditional methods for even greater outcomes, and the high-pressure-induced hydrogen penetration effect could, in principle, be extended to other material systems, such as constructing an exceptional oxide surface for oxidation reactions and a nitride surface for nitrogen-related reactions.

Methods

Chemicals

Polycrystalline zinc (Zn), nickel (Ni), silver (Ag), and gold (Au) foils (0.2 mm thick, ≥99.99% trace metal basis) were purchased from ZhongNuo Advanced Material Technology Company Limited. Potassium bicarbonate (KHCO3, 99.9%), potassium chloride (KCl, 99.8%), sodium chloride (NaCl, 99.5%), ethanol (C2H5OH, 99.5%), formic acid ( ≥ 99%), deuterium oxide (D2O, 99.9 atom %D), and dimethyl sulfoxide (DMSO, ≥99.9%) were purchased from Shanghai Macklin Biochemical Co., Ltd. All chemicals were used without further purification. All aqueous solutions were prepared using deionized water obtained from a Milli-Q water purification system (18.2 MΩ cm). Ar (99.99%) and CO2 (99.99%) gases were purchased from Southwest Institute of Chemical Company Limited.

Preparation of pristine Zn foil

Pristine Zn foil was obtained by cutting purchased Zn foil into 1 cm × 1 cm pieces (corresponding to a geometric surface area of 2 cm2), which were then mechanically polished with sandpaper. Then, the pieces were washed with ethanol and deionized water and finally dried under an Ar flow.

Preparation of Pressure-Zn foil

The Pressure-Zn foil was obtained by subsequent hydrogen penetration of the pristine Zn foil. The process was conducted in an H-type electrolysis cell (Gaossunion) pressured with 60 bar of argon (Ar) under −100 mA cm-2 for 1 h at room temperature (Supplementary Fig. 1). A CHI electrochemical workstation (CHI660E) was used for the electrochemical treatment. The cell was separated by a Nafion 117 membrane. Pristine Zn foil was fixed by a glassy carbon electrode holder as the working electrode. A Ag/AgCl electrode with a saturated KCl filling solution and a Pt foil (1 cm × 1 cm) were used as the reference electrode and the counter electrode, respectively. 12 mL of 1 M KCl electrolyte was used for the cathode chamber, and 12 mL of 0.5 M K2SO4 electrolyte was used for the anode chamber. Before the penetration process, the cell was repeatedly charged and discharged with 5 bar of Ar 5 times to ensure that no air was left inside. After that, 60 bar of Ar was charged to the cell with the cathode and anode sides sharing the same pressure to avoid mechanical damage to the Nafion film. After the penetration process, the obtained Pressure-Zn foil was ready for CO2 electrolysis. For characterization, the obtained Pressure-Zn foil was rinsed with deionized water to eliminate surface electrolytes and dried under Ar flow, followed by immediate transfer to the glovebox and sealing.

Electrochemical measurements

Ambient CO2 electrolysis

All the electrochemical measurements were conducted at room temperature. A CH Instruments electrochemical workstation (CHI660E) was used to perform the electrochemical measurements. Typical three-electrode cell measurements were performed using a customized gas-tight H-type glass cell (The configuration of the H-cell can be found in Supplementary Fig. 22). A Ag/AgCl electrode with a saturated KCl filling solution was used as the reference electrode, and a Pt wire was used as the counter electrode. Zn-based foils were fixed by a glassy carbon electrode holder as the working electrode. 35 mL of 1 M KCl electrolyte was used for the cathode chamber, and 35 mL of 0.5 M K2SO4 electrolyte was used for the anode chamber. The chambers were separated by a Nafion 117 film. Before electrolysis, the electrolyte in the working-electrode chamber was saturated by CO2 bubbling at a flow rate of 30 standard cubic centimeters per minute (sccm, monitored by an Alicat Scientific mass flow controller). Chronopotentiometry was used for CO2 reduction tests in the H-type cell. Vigorous stirring (500 rpm) was applied during electrolysis.

For the evaluation of catalytic performance, all potentials measured against Ag/AgCl (EAg/AgCl) using a three-electrode set-up were converted to the RHE scale based on the following Eq. (1):

$${E}_{{RHE}}={E}_{{Ag}/{AgCl}}+0.197+0.0591\times {pH}$$
(1)

The applied potentials were 100% manually compensated with solution resistance (R, ohm) and noted as E – iR, where R was measured using electrochemical impedance spectroscopy analysis in open circuit potential at frequencies ranging from 0.1 Hz to 200 kHz immediately after the chronopotentiometry measurements. The intersections of the Nyquist plot with the x-axis at the high-frequency regions were used to estimate the corresponding solution resistances. i is current. For the iR-corrected voltammograms, corresponding plots without iR correction were also provided.

The produced CO and H2 were tested online by a gas chromatograph (Clarus 690, PerkinElmer) equipped with a thermal conductivity detector and a flame ionization detector. The produced HCOOH was quantified by collecting and analyzing the electrolyte using an ion chromatograph (ICS-600, Thermo Scientific). Before quantification, the gas chromatograph was calibrated by a series of standard gas mixtures, and the ion chromatograph was calibrated by a series of standard solutions with precise HCOOH concentrations. Unless otherwise mentioned, all the experiments were conducted at ambient pressure and room temperature (about 25 oC).

High-pressure CO2 electrolysis

The process was conducted in an H-type electrolysis cell (Gaossunion) pressured with 60 bar of CO2 under −100 mA cm−2 for 1 h at room temperature (Supplementary Fig. 1). A CHI electrochemical workstation was used for the electrochemical treatment. The cell was separated by a Nafion 117 membrane. Pristine Zn foil was fixed by a glassy carbon electrode holder as the working electrode. A Ag/AgCl electrode with a saturated KCl filling solution and a Pt foil (1 cm × 1 cm) were used as the reference electrode and the counter electrode, respectively. 12 mL of 1 M KCl electrolyte was used for the cathode chamber, and 12 mL of 0.5 M K2SO4 electrolyte was used for the anode chamber. Before the penetration process, the cell was repeatedly charged and discharged with 5 bar of CO2 5 times to ensure that no air was left inside. After that, 60 bar of CO2 was charged to the cell with the cathode and anode sides sharing the same pressure to avoid mechanical damage to the Nafion film. Gaseous products in the headspace of the chamber were extracted and analyzed by gas chromatography, and the liquid products were quantified by ion chromatography. After high-pressure electrolysis, the obtained Pressure-Zn foil was ready for 1-bar CO2 electrolysis. Of note, a water bath at 30 oC was used to avoid the liquidation of CO2 when conducting CO2 electrolysis at 70 bar. Notably, to acquire a j–V plot under high-pressure conditions, we derived a pH of 3.09 for potential calibration based on the dissolution equilibrium of CO2 over 60-bar CO2 electrolysis via a MATLAB code91 (Supplementary Fig. 9).

MATLAB code for the pH calculation in a CO2-pressurized KCl solution

The calculation of pH was performed in MATLAB following a reported code91:

syms a b c d x y;

%{

a = [CO2 (aq)], b = [H2CO3], c = [HCO3-], d = [CO32-], x = [H+], y = [OH-]

%}

k1 = 2.5 * 10^(-4);

k2 = 4.84 * 10^(-11);

ka1 = 4.45 * 10^(-7);

kw = 1 * 10^(-14);

kh = 0.0143;

p = 60;

%{

k1 = equilibrium constant of carbonic acid dissociation into bicarbonate

k2 = equilibrium constant of bicarbonate dissociation into carbonate

ka1 = equilibrium constant of carbonic acid and dissolved CO2 dissociation into bicarbonate

kw = equilibrium constant of water dissociation at 25 oC

kh = Henry’s constant of CO2 dissolved in 1 M KCl, in M/atm

p = CO2 partial pressure, in atm

%}

eq1 = x * c./ b - k1;

eq2 = x * d./ c - k2;

eq3 = (x * c)./ (a + b) - ka1;

eq4 = a - kh * p;

eq5 = (x + 1) - (c + 2 * d + y + 1);

eq6 = x * y - kw;

[a, b, c, d, x, y] = solve([eq1, eq2, eq3, eq4, eq5, eq6], [a, b, c, d, x, y]);

disp(vpa(a, 50));

disp(vpa(b, 50));

disp(vpa(c, 50));

disp(vpa(d, 50));

disp(vpa(x, 50));

Stability test for 1-bar CO2 electrolysis

Due to the fixed volume of electrolyte in an H-cell, a long-term stability test will lead to the accumulation of HCOOH and result in significant acidification and inevitable corrosion of the Pressure-Zn foil. To solve this problem, an external bottle of 1 M KCl electrolyte was connected to the cathode chamber with two peristaltic pumps (YZ15-13A, Huiyu-pump) to ensure that the catholyte cycled and that the produced HCOOH was maintained at a low concentration (Supplementary Fig. 22). At the same time, the external KCl electrolyte was continuously purged with sufficient CO2 gas to maintain CO2-saturated conditions. After a certain amount of electrolysis, the catholyte was extracted for the quantification of HCOOH and renewed for a longer test. For the initial 96 h, we cycled the electrolyte with 400 mL of fresh KCl electrolyte every time 3 times, and a 5-litre KCl solution was used for the final part of a 104-h run. The amount of HCOOH for each period was quantified for the FEHCOOH calculation as the electrolyte was refreshed.

Preparation of Zn foil for CO2 electrolysis via an electrochemical pulse (EP) method

We prepared two additional foils (1 cm × 1 cm pieces, corresponding to a geometric surface area of 2 cm2) with different porosities for a systematic study via a general EPOR method59. Typically, a piece of Zn foil was immersed in 0.2 M NaOH solution in an H-type cell separated by a Nafion membrane with a Pt count electrode and a Ag/AgCl reference electrode (saturated KCl filling solution inside). A symmetric 50 Hz square-wave pulse between 0.1 V and 0.7 V (vs. Ag/AgCl, KCl-saturated) was applied. The Zn foils treated for 10,000 cycles and 20,000 cycles were denoted EP-Zn-10k and EP-Zn-20k, respectively. Prior to CO2 electrolysis, these foils were electrochemically reduced in 1 M KCl at a constant current of −20 mA for 10 min.

Preparation of Ag and Au foils for CO2 electrolysis

The purchased Ag and Au foils (1 cm × 1 cm pieces, corresponding to a geometric surface area of 2 cm2) were cleaned with nitric acid, washed with ethanol and DI water, and finally dried under an Ar flow. For Ag foil, a symmetric 50 Hz square-wave pulse between 0.1 V and 2.0 V (vs. Ag/AgCl, KCl-saturated) was applied in 0.2 M NaOH solution for 20,000 cycles to increase the CO selectivity59. For the Au foil, a symmetric 50 Hz square-wave pulse between 0.1 V and 2.5 V (vs. Hg/Hg2SO4) was applied to a 0.5 M H2SO4 solution for 50,000 cycles to increase the CO selectivity92,93. Prior to CO2 electrolysis, these foils were electrochemically reduced in 1 M KCl at a constant current of −20 mA for 10 min.

Nitrate electroreduction

The electrochemical measurements were carried out in the same three-electrode H-type cell system that used for CO2 reduction test, which was separated by a Nafion membrane. A Ag/AgCl electrode and Pt foil were used as the reference electrode and counter electrode, respectively. 0.1 M KNO3 (30 mL) and 0.1 M K2SO4 (30 mL) were used as the cathode and anode electrolytes, respectively. Chronopotentiometry was used for nitrate reduction tests in the H-type cell. Vigorous stirring (500 rpm) was applied during electrolysis. Ar gas was delivered into the cathodic compartment at a rate of 10 mL min−1 to remove dissolved O2. The outlet was immersed in 0.05 M H2SO4 solution in a gas-scrubbing bottle to collect the NH3 evaporated from the cathodic compartment. NH3 in the cathodic compartment solution and in the gas-scrubbing bottle solution was detected and quantified by cation chromatography. The NO2− in the cathodic chamber was detected by anionic chromatography.

In situ Raman test

Raman analysis was performed with a Renishaw inVia Raman microscope equipped with a 785 nm laser. In situ Raman measurements were carried out using a liquid-electrolyte flow cell (Gaossunion, Supplementary Fig. 24) with a catholyte and anolyte separated by a Nafion 117 film. A Ag/AgCl electrode with a saturated KCl filling solution was used as the reference electrode, and a Pt wire was used as the counter electrode. Zn-based foils were fixed from the back by a copper tapper at the inspection window as the working electrode, with the front side exposed to the catholyte. A 1 M KCl solution presaturated with CO2 was cycled to the cathode chamber, and a 0.5 M K2SO4 solution was cycled to the anode chamber.

Fourier-transformed alternating current voltammetry (FTacV) measurements

FTacV measurements were conducted with a commercially available instrument (CHI660E). An applied sine perturbation with an amplitude (∆E) of 80 mV and a frequency (f) of 9.02 Hz was applied. The potential range was set to −0.6 to −1.8 V (vs. Ag/AgCl, saturated KCl) with a scan rate of 20 mV s−1, and total data points of 65536 (216) were acquired for analysis. The simulation was based on the MECSim software package, which is available at http://www.garethkennedy.net/MECSim.html. Based on the experimental results, an optimized model for two successive surface-confined electron transfer-chemical reaction processes was adopted for simulation. Detailed information and discussions for the simulation can be found in Supplementary Table 3.

Structural characterization

Scanning electron microscopy (SEM) images were taken on a Hitachi S-4800 field emission scanning electron microscope. High-resolution transmission electron microscopy (HR-TEM) images were taken on a JEOL ARM-200F field-emission transmission electron microscope using Cu-based TEM grids. X-ray diffraction patterns were recorded using a Philips X’Pert Pro Super diffractometer with Cu-Kα radiation (wavelength, λ  =  1.54178 Å). An Attension Theta (Biolin Scientific) was used to test the contact angles.

Time-of-flight secondary ion mass spectrometry (TOF-SIMS) measurements

TOF-SIMS depth profile analysis was conducted in negative ion mode with a TOF-SIMS 5-100 mass spectrometer (ION-TOF, GmbH). The profile was obtained with a 2 keV Cs sputtering beam and 30 keV Bi+ analysis beams. The analysis area and the sputtering area were 50 × 50 μm2 and 200 × 200 μm2, respectively. To circumvent the disturbance of surface zinc hydroxyl groups, we tracked the change in hydrogen content against depth through the H-/ZnO- ratio against the sputtering time.

Temperature-programmed desorption-mass spectrometry (TPD-MS) analysis

The thermal desorption properties of the treated Zn foils were analyzed with a TG-IR-GC/MS high-fidelity heated transfer system (TL-9000, PerkinElmer). Under an Ar flow, the samples were heated from 30 °C to 400 °C at a heating rate of 20 °C/min. H2 signals (m/z = 2) were recorded with a mass spectrometer.

1H nuclear magnetic resonance (NMR) spectroscopy

1H NMR measurements (Varian 400 MHz NMR spectrometer, Bruker AVANCE AV III 400) were carried out to detect the liquid products. Specifically, 0.8 mL of the electrolyte after electrolysis was mixed with 0.2 mL of D2O/DMSO solution. The concentration of DMSO in the D2O solution was 14.27 mM. Then, 0.7 mL of the mixed electrolyte/D2O/DMSO solution was extracted for 1H NMR analysis.

Surface roughness analysis

The surface roughness factor of the prepared Zn electrodes was estimated by measuring the electric double-layer capacitance. Cyclic voltammetry (CV) was performed on the foils at various scan rates (i.e., 20, 30, 40, 50, and 60 mV s–1) in an Ar-bubbled 0.5 M K2SO4 electrolyte. The CV potential range was selected from –0.31 to –0.41 V vs. RHE. The capacitance of each electrode was calculated with the following Eq. (2):

$$\frac{\Delta j}{2}={C}_{{{{{\rm{d}}}}l}}\times \frac{{{{{\rm{d}}}}V}}{{{{{\rm{d}}}}t}}$$
(2)

where Δj/2 is the half-charging current density difference (mA cm–2), Cdl is the capacitance (mF cm–2), and dV/dt (V s–1) is the scan rate. To measure the non-Faradaic current densities that were induced by the geometric effect of the electrode, Δj/2 was obtained at –0.36 V.

The roughness factor is calculated with the following Eq. (3):

$${R}_{f}=\frac{{C}_{{dl}}}{{C}_{{dl},{ref}}}$$
(3)

where Rf is the roughness factor, Cdl is the double-layer capacitance of foil, and Cdl,ref is the average double-layer capacitance of a smooth metal surface (Cdl,ref = 0.02 mF cm-2)94,95.

The electrochemically active surface area is determined by the following Eq. (4):

$${A}_{{ECSA}}={A}_{{geometric}}\times {R}_{f}$$
(4)

where Rf is the roughness factor, AECSA is the electrochemically active surface area and Ageometric is the geometric surface area of the foils.

Texture coefficient analysis

Based on the XRD patterns, we were able to calculate the texture coefficients (TC) of these foils via the following Eq. (5)59,70,96:

$${{Texture\; coefficient}}_{({hkl})}=\frac{{I}_{({hkl})}/{I}_{0({hkl})}}{{\sum }_{i=1}^{n}{I}_{({hkl})n}/{I}_{0({hkl})n}}\times 100\%$$
(5)

where I(hkl) is the diffraction intensity of the (hkl) facet for the foils, I0(hkl) is the standard intensity of the (hkl) facet taken from JCPDS No.04-0831, and n is the number of diffraction peaks of Zn (n = 6, here). A detailed discussion on texture coefficient can be found in Supplementary Note 3.

Density functional theory calculations

We performed structural optimizations with a periodic plane-wave implementation using the Vienna ab initio Simulation Package (VASP) code based on density functional theory (DFT)97,98. Meanwhile, we adopted the Perdew-Burke-Ernzerhof (PBE) functional based on the generalized gradient approximation (GGA) approach to incorporate exchange and correlation energies. The ion-electron interactions were described with the projector augmented wave (PAW) pseudo-potentials99,100. An energy cutoff of 500 eV was adopted. A first-order Methfessel-Paxton smearing of 0.05 eV was applied to the orbital occupation during the geometry optimization and for the energy computations.

Four-layer 3×3 supercells with the bottom two layers constrained were built to simulate the exposed metal surfaces and evaluate adsorption energies. In addition, [4×4×1] Monkhorst–Pack k-point grids were used with a convergence threshold of 10–6 eV for the iteration in the self-consistent field (SCF)101. Structure optimizations were kept until the force components were less than 0.02 eV/Å. The vibrational frequencies of free molecules and adsorbates were calculated by using the phonon modules in the VASP 5.3 code. We employed a standard thermodynamic correction to determine the free energy corrections including the effect from zero-point energy, pressure, inner energy, and entropy. The transition states were determined using the climbing image nudged elastic band (CI-NEB) method with a convergence threshold of 0.05 eV/Å102. The optimized configurations used for calculating the adsorption energy, as well as the initial/final states to obtain the transition state were included in the Supplementary Data 1.