In situ small-angle X-ray scattering study during CO2 RR. We first synthesized and characterized monodisperse Cu NPs as a model electrocatalyst with an average radius (ravg) of 3.55 ± 0.48 nm, confirmed by transmission electron microscopy (TEM, Fig. 1a). The as-synthesized Cu NPs were directly deposited onto graphite foil with a loading amount of ~ 0.25 mg/cm2, generating a catalyst layer with a thickness of 1 ~ 2 µm (Supplementary Fig. 1a-c). TEM and grazing incidence X-ray diffraction (GIXRD) analysis demonstrated that the deposited Cu NP is a mixture of metallic Cu, Cu2O, and CuO (Supplementary Fig. 1d-f). The resulting Cu NPs electrode was incorporated into a custom-built electrochemical flow cell that is designed to enable in situ SAXS measurements under realistic CO2RR conditions (Supplementary Fig. 2a,b). The thickness of the electrolyte pathway (6 mm) is minimized to mitigate attenuation of the X-ray transmission by the electrolyte. The minimized electrolyte pathway can cause bubbles to block the catalyst layer during the gas-evolving CO2RR and hydrogen evolution reaction (HER). This blockage leads to discontinuous current density profiles due to limited reactant availability (Supplementary Fig. 2c). To facilitate the detachment of gas bubbles from the electrode surface, we added anionic hydrotropes into the electrolyte that can decrease the bubble size and residence time by reducing the surface tension.25 After adding the hydrotropes, stable current densities were recorded under CO2- and Ar-purged electrolyte up to -0.9 V (Supplementary Fig. 2d,e). Although the addition of hydrotropes induced a decrease in the Faradaic efficiency of CO2RR by increasing the rate of HER, the partial current densities for C2+ hydrocarbons were comparable to that of the sample without the hydrotropes (Supplementary Fig. 2f,g).
To monitor the structural transformation of Cu NPs, we collected in situ SAXS spectra for 1 hr of electrolysis with an interval of 1 min at -0.65 V under CO2RR conditions (Fig. 1b). The global unified function was employed to fit the SAXS pattern,26 assuming that the Cu structure, derived from the initial Cu NPs, consisted of two distinct structural levels: primary Cu NPs (Fig. 1c) and secondary aggregates (Fig. 1d). This fitting method allows us to estimate the radius of gyration (Rg) and the power-law slope (P) of primary NPs (Rg1 and P1, Fig. 1e) and of secondary aggregates (Rg2 and P2, Fig. 1f), respectively. The Rg value is related to the size depending on the structural model, and the P value reflects the morphology of the underlying structure.27 The calculated radius of the initial Cu NPs at OCV was 3.01 ± 0.16 nm in the sphere model, which is in agreement with the size derived from the TEM (Supplementary Fig. 3a). The fitting result also yields the number density and volume fraction of the secondary aggregates by calculating the Porod invariant Q (Fig. 1g). The detailed theory and application of the unified fit are discussed in the Supplementary Information. The P2 values remained between 3 and 4 during the whole reaction time, which represents surface fractal morphology (Fig. 1f). The P1 values before 6 min were close to 4, corresponding to spherical particles, whereas they showed a deviation from the value of 4 after 6 min, implying changes in the structure, such as particle elongation or surface roughening (Fig. 1e).26 We left the P1 value as a free parameter during the fitting process to reflect the changes in the particle morphology and polydispersity. The size distributions can be calculated under the assumption of spherical particles (P1 = 4) and log-normal distributions.26,28 Similar Rg values were found between the fitting processes with a fixed P1 = 4 and with P1 as a free parameter (Supplementary Fig. 3b,c).
The growth of Cu NPs derived from quantitative SAXS analysis showed three distinct periods (Fig. 1h). Until 6 min of reaction, the Rg1 value showed a stepwise increase and reached up to ~ 83 Å (Fig. 1e). The aggregates also started to grow in this period, but the increase in the Rg2 value did not exceed 20% (Fig. 1f). The volume fraction and number concentration of aggregates rapidly decreased in this period, possibly indicating the particle detachment from the electrode as well as migration out of the irradiated volume (Fig. 1g). This result suggests that initial aggregation of neighboring Cu NPs during the early stages of the reaction occurred through the PMC mechanism. Previous research has shown that the electromigration of the metal can serve as a driving force for the aggregation of nanostructures in response to an applied electric field and the flow of current.7,29 In the next period (6 ~ 12 min), the Rg2 value rapidly increased up to ~ 553 Å, whereas the Rg1 value started to decrease slightly down to 72 Å. The rapid growth of aggregates without any prominent size decrease in primary Cu NPs indicates that an increase in Rg2 occurs through the sintering of pre-existing aggregates, indicating PMC as the dominant agglomeration mechanism. The OR mechanism can also contribute to the growth of aggregates since the Rg1 value slightly decreased with a broadening size distribution (Supplementary Fig. 3d). Assessing the individual contribution of each process to the overall agglomeration rates poses a considerable challenge. However, the predominant agglomeration mechanism transitioned from the PMC during the early stages of the reaction to a combination of PMC and OR as the agglomeration process decelerated.
During the last period (13 ~ 60 min), a declustering behavior was observed in both the primary Cu NPs and secondary aggregates. This phenomenon is characterized by a reduction in particle size resulting from the breaking apart of particles into smaller ones, as has been previously reported.22 The Rg1 values decreased rapidly in this period and the Rg2 values also began to decrease after 30 min. The decrease in both values could be explained by the dissolution of transient Cu species and redeposition onto the large particles, according to the OR process.30 We observed the formation of micrometer-sized CuOx agglomerates via post-mortem ex situ scanning electron microscopy (SEM) with energy-dispersive X-ray spectroscopy (EDS) analysis (Supplementary Fig. 4a-d). These large particles can keep growing by consuming the Cu NPs even after the PMC mechanism for agglomeration is no longer dominant. We also found evidence of the emergence of Cu clusters with a radius of about 2 nm (Supplementary Fig. 4e,f), further indicating that the OR process was dominant. Thus, the results indicate that a mechanistic transition from PMC to OR process took place as the CO2RR proceeded at -0.65 V, resulting in a hierarchical Cu structure ranging from micrometer-sized agglomerates to nanoscale clusters (Fig. 1h).
Structural transformation of Cu NPs depending on the applied potentials. To investigate the structural evolution of Cu NPs depending on the applied potentials during the CO2RR, we collected in situ SAXS data (Supplementary Fig. 5) and conducted post-mortem microscopy on the samples after reaction under CO2-purged electrolyte (Supplementary Figs. 7–9). By using the unified fit, we tracked and compared the size evolution of secondary aggregates (Fig. 2a) and primary Cu NPs (Fig. 2b) with the applied potentials from − 0.45 V to -0.9 V. At -0.45 V, the growth of aggregates was not evident except for a ~ 10% increase in the early period of the reaction (< 5 min), and the initial size of primary Cu NPs remained constant until 15 min. The Rg1 value slightly decreased from 30 Å to 25 Å with a broadening size distribution after 15 min, indicating that the OR process took place (Supplementary Fig. 6a). The SAXS analysis was correlated to the ex situ SEM, EDS, and TEM data analysis. We observed micrometer-sized particles on top of the catalyst layer, which turned out to be organic residues derived from ligands used in the preparation of the Cu NPs (Supplementary Fig. 7a-c). The organic residues were the only micrometer-sized structures observed for the sample run at -0.45 V, with no evidence of micrometer-sized Cu agglomerates (Supplementary Fig. 7d). The phosphonic acid-based ligands utilized in this study have previously been demonstrated to desorb from the surface of Cu NPs at negative potentials.31 The rate and extent of desorption exhibited an increasing trend with more negative applied potentials and with time. Notably, the detached ligands exhibit an organized ligand layer formation near the NP surface, thereby potentially establishing catalytic hotspots during the CO2RR, rather than undergoing complete loss into the electrolyte solution.32 We confirmed the initiation of ligand desorption at approximately 0 V (Supplementary Fig. 8a,b), and even after applying a potential of -0.9 V under both CO2 and Ar conditions, a residual amount of ligands appears to remain on the electrode (Supplementary Fig. 8c). We found that the final Cu structure after the CO2RR at -0.45 V exhibited a mixture of polycrystalline aggregates attributable to PMC and individual Cu NPs transformed through the OR process (Supplementary Fig. 7e,f). These results suggest that the driving forces for both PMC and OR processes were weak due to the low applied overpotential and the protective role of surface ligands toward agglomeration between the Cu NPs.
By contrast, the structural changes of Cu NPs at -0.9 V showed an immediate decrease in Rg1 value from 32 Å to 25 Å within 10 min and a size broadening of primary Cu NPs as the reaction started, and this trend continued throughout the 1 hr of electrolysis (Fig. 2b and Supplementary Fig. 6b). An increase in Rg2 value from 306 Å to 388 Å was observed up to 10 min, followed by a relatively stable value of ~ 370 Å for the rest of the reaction. The growth of aggregates without an increase in Rg1 value, and the broadening size distribution of primary Cu NPs, imply that the OR process was the prevailing mechanism throughout the agglomeration process. The post-mortem microscopic analysis revealed that dendritic CuOx particles with a size up to ~ 5 µm were formed (Supplementary Fig. 9a-d). In addition, we identified Cu clusters adjacent to the aggregates with a radius as small as ~ 1.3 nm, which is notably smaller than the original NPs with a radius of 3.5 nm (Supplementary Fig. 9e,f) and also indicative of OR. The PMC mechanism in the early period of the reaction cannot be completely excluded, due to the existence of polycrystalline aggregates comprised of fused and elongated rod-shaped NPs in the TEM images. The trend of structural changes was reproduced in another Cu NPs sample (Supplementary Fig. 10). Thus, these results strongly suggest the OR mechanism dominated the structural evolution of Cu NPs at -0.9 V under CO2-purged electrolyte. Figure 2c illustrates the morphological evolution of Cu NPs during the CO2RR depending on the applied potentials. As the negative potentials varied from − 0.45 to -0.65 V, the degree of structural reconstruction became increasingly prominent. This reconstruction was characterized by the PMC observed during the initial stages of the reaction, followed by the OR process. The emergence of PMC-driven agglomeration in the early period could be due to the accelerated electromigration rate of Cu at the more negative potential, along with the desorption of the ligands at -0.65 V. The declustering of both aggregates and NPs appears to be accelerated with more negative potential as well, which could be due to OR dominating over other agglomeration processes at -0.9 V.
Investigation of reaction parameters governing structural transformation. To study the driving force behind the structural evolution of Cu NPs in response to reaction conditions, we measured the reconstruction of Cu NPs with different applied potentials under Ar-purged 0.5 M KxH3−xPO4 electrolyte by using in situ SAXS (Fig. 3a,b) and post-mortem microscopic analysis (Supplementary Figs. 11–13). A similar trend of size changes in the primary Cu NPs and secondary aggregates from − 0.45 V to -0.9 V was observed compared to that of the samples under CO2-purged electrolyte. At a potential of -0.45 V, the growth of secondary aggregates was found to be minimal and a reduction in the size of primary Cu NPs was observed after 10 min, which was attributed to the OR process occurring during the later stages of the reaction. A potential of -0.65 V resulted in a rapid increase in both Rg1 and Rg2 values, which can be ascribed to the PMC operating during the early stages of the reaction. After 5 min, a rapid decline in Rg1 values from ~ 80 Å to ~ 40 Å was observed, indicating that the OR process had taken over. At -0.9 V, an immediate decrease in Rg1 values with a broadening of the size distribution was discovered. The Rg2 values increased from 305 Å to 400 Å for 13 min, suggesting the prevalence of the OR process throughout the entire reaction. The P2 values under both CO2 and Ar-purged electrolytes remained within the values of 3 and 4 regardless of the reaction conditions, indicating the surface fractal structure of secondary aggregates (Supplementary Fig. 14a,b). Similar behavior in the reconstruction under CO2- and Ar-purged electrolytes indicates that the applied potential is one of the main factors to drive size and morphological changes during the CO2RR.
The main difference in the structural evolution of Cu NPs under CO2- and Ar-purged electrolytes is a faster decline of Rg1 values from − 0.45 V to -0.9 V than the samples under CO2-purged electrolyte (Supplementary Fig. 14c). At a potential of -0.45 V, a decrease in Rg1 was accelerated from ~ 30 Å to ~ 16 Å after 10 min (Supplementary Fig. 15a). The declustering of primary Cu NPs was observed within 5 min at -0.65 V, which was faster than the samples under CO2-purged electrolyte. Under the potential of -0.9 V, the Rg1 values exhibited a more rapid decrease from ~ 30 Å to ~ 15 Å within 30 min, accompanied by a broadening size distribution (Supplementary Fig. 15b). This decline was observed to be faster compared to the sample under CO2 conditions. Additionally, the Rg2 values calculated at -0.65 V were found to be below 450 Å, which are smaller than those calculated for the CO2-purged electrolyte (Supplementary Fig. 14d). The enhanced declustering behavior observed under Ar conditions has been previously reported,22 and several factors, such as CO* intermediates, high local pH, and Cu dissolution, could influence the dynamics of Cu NPs reconstruction under different reaction conditions. The enhanced declustering behavior could be attributed to the high formation rates of molecular Cu species via interactions between surface Cu atoms and adsorbed H (H*) under Ar conditions, resulting in the acceleration of the OR process. However, a study suggests that surface-bound CO2RR intermediates (e.g., *CO and *C2O4) can enhance the dissolution of Cu surface atoms, which would be minimized under N2 conditions.33 A similar observation of CO2RR intermediates enhancing the degradation of the catalyst surface has been reported on Au NPs.7 These previous findings suggest that Cu dissolution is not a major factor in the enhanced declustering observed under Ar conditions.
Another factor to consider is the high local pH near the electrode surface, caused by the production of OH─ ions during the CO2RR and HER. To investigate the effect of local pH on the reconstruction of Cu NPs, we conducted measurements and analysis of in situ SAXS data of Cu NPs under Ar-purged 0.5 M KOH electrolytes (pH > 13, Supplementary Fig. 16). The fitting results demonstrated a sudden decrease in Rg1 values, reaching ~ 20 Å within 5 minutes at -0.45 V, and ~ 16 Å within 2 to 3 min at -0.65 V and − 0.9 V, respectively. Following the significant decline during the initial phase of the reaction, the Rg1 values remained relatively stable around 15 Å at -0.65 V and − 0.9 V. This rapid reduction in the size of the primary Cu NPs can be attributed to the instability of Cu NPs in pH levels higher than 13 and the anodic OCV value (~ 792 mV), causing oxidation and dissolution of the Cu NPs (Supplementary Note 1).16,18 The presence of dissolved Cu ions and/or species, which are generated during the OCV period, along with the reduction of the Cu oxide/hydroxide layer, can contribute to a high rate of small cluster formation when a cathodic bias is applied. This process ultimately can lead to a sudden reduction in the size of the primary Cu NPs. However, the electrocatalysis under Ar-purged KOH electrolytes failed to reproduce the overall trend of structural transformation of Cu NPs observed under both CO2 and Ar conditions, indicating that high local pH driven by the reaction alone does not determine the reconstruction of Cu NPs. Since neither the rate of Cu dissolution nor local pH are the main factors explaining the enhanced declustering behavior under Ar conditions, we turn our attention to the higher rate of HER under the Ar conditions compared to that of CO2 conditions. The HER generates an excessive amount of adsorbed H intermediate (H*) on the Cu surface and H2 gas close to the Cu NPs. These hydrogen atoms can penetrate solid metals and reduce the stress required for the initiation and propagation of cracks, a phenomenon known as hydrogen embrittlement.34 Copper, with its low diffusivity and solubility of hydrogen, is generally considered to be less susceptible to hydrogen embrittlement than other metals, such as iron and nickel.35 However, the electrochemical HER conditions result in the formation of trapped H2 gas inside the metallic Cu in the form of voids36 or retained in lattice defects such as grain boundaries, vacancies, and dislocations.37 The internal accumulation of hydrogen reduces the ductility of copper and increases its susceptibility to hydrogen embrittlement. Under CO2RR conditions, the intermediates can compete with the H* at the surface and suppress the rate of HER,38 thereby mitigating the hydrogen embrittlement of Cu during the reaction.
The morphological difference in the samples under CO2- and Ar-purged electrolytes was also revealed in the formation of micrometer-sized particles depending on the applied potentials. The ex situ SEM analysis demonstrated that micrometer-sized dendritic particles formed under CO2RR conditions (Fig. 3c), whereas the formation of micrometer-sized particles was partially suppressed under Ar conditions (Fig. 3d). The growth of the dendritic structure during the CO2RR has been observed previously using in situ TEM analysis.17 This structure can form easily under the wet electromigration condition that enhances the migration of metal species in the presence of electrolytes.29 Under electric bias, the dissolved metal species tend to deposit at localized needles or spikes, because the higher current density at their tips considerably enhances the probability of additional deposition.39 Due to the resulting concentration gradients, the formation of a dendritic structure can occur, which is consistent with our findings under CO2-purged electrolyte. In contrast, the suppression of dendrite growth under Ar-purged electrolytes may be explained by excessive bubble formation that can interfere with the nucleation process by increasing turbulence on the Cu surface.40 The current density profile of Cu NPs under Ar-purged electrolytes included repetitive large spikes, indicating the generation and rapid eruptions of the gas bubbles (Supplementary Fig. 2e). Therefore, various reaction parameters, including the applied potentials, surface intermediates, and bubble generation, contribute to the dynamic process of structural transformation of Cu NPs during the CO2RR at different length scales.
Catalyst detachment revealed by in situ synchrotron X-ray absorption spectroscopy. As the volume fraction and number density of secondary aggregates rapidly decreased in the first 5 min of operation (Fig. 1g), we further investigated the integrity of Cu NPs on the carbon support by using in situ quick-scanning XANES measurements. This mode allowed simultaneous monitoring of Cu concentrations in the catalyst layer and oxidation states of the Cu NPs with a temporal resolution of 15 seconds. We employed an XAS flow cell that is designed to remove bubbles without adding hydrotropes to the electrolyte (Supplementary Fig. 17).41 We tracked the evolution of XANES spectra for Cu NPs deposited on the graphite foil under CO2-purged electrolyte at -0.6 V (Fig. 4a,b). Interestingly, the XANES signal drastically decreased within 2 min. The relative concentration of Cu NPs in the catalyst layer over time was estimated by using the XANES edge step height that scales with the absolute amount of Cu in the X-ray beam.42 The normalized edge step value decreased to ~ 3% in 2 min (Fig. 4c). The difference in the Cu NPs concentration calculated from SAXS (~ 78% and ~ 56% in volume and number concentration, respectively) and XANES may result from distinct hydrodynamics in the two types of flow cells employed. The geometries of cells notably influence mass transport and the elimination of bubbles from the electrode surface.43 Additionally, the presence of hydrotropes in the SAXS measurements can account for this divergence by aiding in the removal of bubbles that may drive the detachment of NPs.44 The post-mortem EDS mapping analysis of the Cu NPs further demonstrated that the Cu NPs detached from the graphite foil when a negative bias was applied (Supplementary Fig. 18a-c). The detachment of Cu NPs is also consistent with the fact that the CO2RR performance was very weakly correlated with the mass loadings (Supplementary Fig. 18d,e).
To suppress the detachment of Cu NPs, we utilized carbon black as a conductive support and Nafion ionomer as a binder (Cu NPs/CB, Fig. 4d). The individual Cu NPs were isolated on the support surface as confirmed by TEM, and high current densities were observed in the XAS flow cell under both CO2 and Ar-purged electrolytes (Supplementary Fig. 19a-c). The dispersion of Cu NPs on a high surface area carbon support improved overall activity toward CO2RR, possibly as a result of increased interparticle distance, which can facilitate the mass transport of CO2 to individual Cu NP, while simultaneously reducing NP detachment (Supplementary Fig. 19e,f). The implementation of Nafion ionomer has been reported to enhance the adhesion of NPs onto the carbon support, thereby reducing the NP detachment.45 The XANES spectra of Cu NPs/CB electrodes under CO2-purged electrolytes at -0.6 V demonstrated that the detachment of NPs was mitigated (Fig. 4e) and the edge step values retained ~ 70% of the initial values at OCV (Fig. 4f). The normalized values of the edge step decreased with more negative applied potential from − 0.25 V to -0.7 V, indicating that the cathodic bias contributes to the driving force for NP detachment (Supplementary Fig. 19d).
The mitigation of Cu NPs detachment enabled us to test the hypothesis that the oxidation states of Cu NPs play a role in structural evolution during CO2RR. We collected in situ XANES spectra as a function of applied potentials in CO2- and Ar-purged electrolytes (Supplementary Figs. 20–22). Repeated XANES spectra were collected until no further changes were observed, and the first derivatives of the XANES spectra were used to evaluate the oxidation states of Cu NPs. The qualitative XANES analysis revealed that Cu+ species were present even after 25 and 12 minutes of operation at -0.25 V under CO2- and Ar-purged electrolytes, respectively, (Supplementary Fig. 20). Within the potential range of -0.45 V to -0.7 V, which is relevant to CO2RR, the Cu NPs remained in a metallic state (Supplementary Fig. 21). The rate of reduction to a fully metallic state of Cu NPs under Ar-purged electrolytes was slower than that of CO2-purged electrolytes across the applied potentials, indicating CO2RR intermediates on the surface may play a role (Supplementary Fig. S23). The Cu NPs were fully reduced into metallic states below − 0.45 V under CO2-purged electrolytes within 2 min of the reaction. Considering that the size of Cu NPs did not significantly change in 2 min as confirmed by SAXS analysis, this result suggests that the oxidation states do not play a major role in the ensemble-averaged structural evolution of Cu NPs during the CO2RR.
Mitigation of degradation of Cu NPs by high surface area supports. Inspired by the mitigation of NPs detachment by using a carbon black support and Nafion ionomer, we investigated the role of support materials in the structural reconstruction of Cu NPs with in situ SAXS measurements (Supplementary Fig. 24). The changes in the Rg1 value at -0.65 V under CO2-purged electrolyte indicated that the agglomeration process through the PMC mechanism in the early period of the reaction (< 5 min) was suppressed in the case of Cu NPs/CB (Fig. 5a). The Rg2 values of Cu NPs/CB increased up to ~ 320 Å within 3 min, but remained at ~ 360 Å throughout the reaction time. This value is lower than that of Cu NPs without the supports (up to ~ 550 Å, Fig. 5b). The size distribution demonstrated that the growth of primary Cu NPs larger than their original sizes was absent in the early period of reaction (Supplementary Fig. 25a). The post-mortem ex situ SEM and TEM analysis revealed that the formation of micro-sized Cu agglomerates was prevented, but Cu clusters with a radius as small as ~ 1.3 nm were still observed (Fig. 5c,d). These results illustrate that the OR process was only partially suppressed, even though the carbon supports helped to relieve the agglomeration of Cu NPs through the PMC process.
For Cu NPs/CB at -0.9 V under CO2-purged electrolytes, the Rg1 values were stable, remaining close to their original value of ~ 32 Å throughout the 1 hr of electrolysis Fig. 5a). The changes in the size distribution of primary Cu NPs demonstrated that the NPs were resistant to rapid size broadening compared to the Cu NPs without supports (Supplementary Fig. 25b). No significant differences in the Rg2 values were observed in the first 40 min of the reaction. However, after 40 min, the Rg2 values decreased down to ~ 285 Å, which is smaller than that of the samples without the support (~ 370 Å, Fig. 5b). This trend in structural changes in the Cu NPs/CB was similar for the samples under the Ar-purged electrolytes, indicating that the carbon supports and ionomer coatings alleviate the OR process of Cu NPs regardless of reaction conditions (Supplementary Fig. 26). However, post-mortem ex situ SEM and TEM analysis revealed the generation of micro-sized particles and small clusters locally, attributed to the OR process (Fig. 5e,f). These findings suggest that incorporating carbon supports can mitigate both the PMC and OR process during the CO2RR. However, completely preventing the OR process at very negative potentials requires additional protection to reduce the corrosion and consequent dissolution of Cu NPs into electrolytes.
Based on these results, a variety of approaches can be derived to mitigate Cu nanocatalyst degradation during CO2RR. The appropriate strategy depends on the ranges of the applied potentials, which dictate the predominant degradation mechanism. The usage of Cu nanocatalysts with lower overpotentials for CO2RR may enhance durability, as structural transformation is minimal at -0.45 V. For potentials ranging from − 0.45 V to -0.9 V or beyond, support materials with optimal Cu-support interactions could mitigate both the PMC and OR process by modulating particle adhesion energy and atom binding energy with supports.13,46 For potentials more negative than − 0.9 V, schemes to prevent Cu corrosion into dissolved species should be the focus, such as physical coatings with graphene47 and N-doped carbon48 to shield the Cu surface. Alloying with Ga metal has been reported to improve the resistance toward Cu dissolution during CO2RR.49 The HER-induced hydrogen embrittlement of Cu could be mitigated through the use of ionomer coatings coupled with pulsed electrolysis, which was found to reduce the Faradaic efficiency of H2 to 4%.50 The detachment of Cu nanocatalysts can be suppressed by introducing supports, as demonstrated in this study, or patterning electrode surfaces to guide the bubble flow and limit the bubble growth, such as striped patterns and periodic cracks.51