Critical role of hydrogen sorption kinetics in electrocatalytic CO2 reduction revealed by on-chip in situ transport investigations

Precise understanding of interfacial metal−hydrogen interactions, especially under in operando conditions, is crucial to advancing the application of metal catalysts in clean energy technologies. To this end, while Pd-based catalysts are widely utilized for electrochemical hydrogen production and hydrogenation, the interaction of Pd with hydrogen during active electrochemical processes is complex, distinct from most other metals, and yet to be clarified. In this report, the hydrogen surface adsorption and sub-surface absorption (phase transition) features of Pd and its alloy nanocatalysts are identified and quantified under operando electrocatalytic conditions via on-chip electrical transport measurements, and the competitive relationship between electrochemical carbon dioxide reduction (CO2RR) and hydrogen sorption kinetics is investigated. Systematic dynamic and steady-state evaluations reveal the key impacts of local electrolyte environment (such as proton donors with different pKa) on the hydrogen sorption kinetics during CO2RR, which offer additional insights into the electrochemical interfaces and optimization of the catalytic systems.


2) In
-the authors attribute the large hysteresis in the electrical transport measurements (and therefore in hydrogen absorption) to be due to the slow kinetics of hydrogen reduction from water in a neutral/alkaline electrolyte. Did the authors perform any additional measurements to corroborate this? Namely, were CV's and ETS measurements performed at varying scan rates? If the hysteresis is scan rate dependent, this would pretty strongly support it is a kinetic issue. (Are the results in Figure S21 enough to corroborate this?).
3) Does the absorption of hydrogen into the Pd4Ag alloy cause a segregation of the constituents (dealloying)? 4) Paper could use minor copy-editing to improve English writing.
Reviewer #2 (Remarks to the Author): The manuscript titled "Critical role of hydrogen sorption kinetics in electrocatalytic CO2 reduction revealed by on-chip in situ transport investigations" Mu et al. probes the hydrogen sorption properties of Pd and Pd4Ag alloys using a novel electron transport spectroscopy method. The authors detail the role of different electrolytes on the hydrogen sorption properties of these materials as well as how CO2 influences hydrogen sorption and connect their observations to trends in electrolyte pKa, pH, and CO2 content. Overall, this investigation is interesting and their methods should be broadly applicable to many different systems of high complexity, however, possible diffusion-related artifacts as well as a lack of computational experiment to substantiate their claims prevent me from recommending this publication in Nature Communications. In addition, the manuscript has several language issues, which, in addition to the somewhat unfocused nature of the text, make the article difficult to read.
Major Points: (1) Diffusion of the electrolyte to the surface of the electrode is potentially problematic for the data presented. For example, significant weight of the authors' claims is given to the absolute electrochemical potentials of the Habs and Hads, however, the authors clearly show in figure S9 that due to diffusion issues, these potentials shift by hundreds of mV depending on the electrode thickness-not a material-dependent property. In addition, the authors present time-resolved ETS in different electrolytes with and without CO2 present. Both the 'equilibrium potentials' and the reaction kinetics (as defined by the half-life of a transient current signal) change based on the electrolyte and greatly influence the authors' claims. Given the sensitivity of measurement on the sample thickness and presumably sample morphology, the authors need to show that these ETS and CV measurements reflect the Pd and Pd4Ag, and not the morphology or structure of the sample. This can be accomplished by supplementing their current data with additional ETS and CV curves of different sample thicknesses in the H2CO3 and H3PO4 electrolytes similar to data presented in Figure S9. In addition, scan-rate dependence data on the peak positions of hydrogen absorption desorption would be valuable.
(2) The authors make several claims that the electronic structure of the Pd4Ag alloy is responsible for both the reduced CO poisoning and the reduced H absorption capacity. Based on their data, however, a simple geometric effect can explain their observations. The stoichiometry of Pd4 Ag1 implies that, in a homogeneous alloy, every palladium atom in an fcc lattice is in contact with a Ag atom. Ag is known to both rapidly evolve CO (due to the low CO binding energy) and does not absorb hydrogen into it's lattice. A silver atom near every Pd may therefore inhibit CO poisoning and H absorption-based only on it's proximity and not on orbital hybridization. If the authors are going to claim electronic structure is the origin of their observation, they need to supplement their data with additional computational evidence showing such an effect.
Minor Points: (1) Figure 2, panel a,b,c. Please clearly indicate which y-axis corresponds to the plotted data. As it currently stands, it is very difficult for the reader to interpret these plots.
(2) Figure S15 panels a and c. The arrows seem to be going in opposite directions with increasing scan. Please alter the figure to clearly indicate the scan direction.
(3) In the text and figure 4a, the authors switch notation between the empirical chemical formula and an abbreviated name. Specifically, H2CO3 and 'PBS'. To improve the readability of the figures and text, the authors should choose either the chemical formula or abbreviations, but not both.
The authors of this study combine voltammetry with electrical transport measurements to add to our understanding of the role of adsorbed and absorbed hydrogen and solution phase proton-donors in carbon dioxide electroreduction. I believe this work is well performed, timely, and provides a very interesting set of results to the electrocatalysis community. However, I would encourage the authors to address the following concerns to improve their paper.
Reply: We greatly appreciate the reviewer's positive comments on our study, and thank for the insightful suggestions that have inspired us to conduct key measurements on the subject and to improve the clarity in writing this manuscript. Per reviewer's suggestions, we have added additional data along with more concise and clarified discussions, which we believe have further improved the quality of this paper. Reply: We thank the reviewer for pointing out the critical issue of unclear presentation of results in the manuscript. In this study, we have indeed devoted a great deal of space in the text to explain what things are occurring, partially due to the reason that this is the first time ETS is conducted for investigating H sorption processes and CO2RR. We now realize that it is a bit too much and makes the main text difficult to read. Per Pd4Ag respond differently to these conditions. These data lead to the introduction of the critical role of H sorption kinetics in Pd-catalyzed CO2RR processes.
Second, in new Figure 3d, we further establish the connections between the H sorption kinetics and the corresponding ETS signals in the extended electrolytes (starting from KHCO3), which also serves as an effort to address the question #2 from the reviewer. On this basis, in new Figure 3e-h, we demonstrate the ETS studies under CO2RR conditions to elucidate the H sorption kinetics during CO2RR and rationalize the distinct CO2RR performance of different Pd-based catalysts in KHCO3. Specifically in this big section, we also rearranged the original discussion paragraphs in the revised manuscript: we first conclude the competition between CO2RR and hydrogen sorptions in KHCO3 directly from the ETS results, and then discuss the possible reactions on the surface that is "in the context with prior work". The ETS results in a different electrolyte (K2HPO4/KH2PO4) were then moved forward in the revised manuscript as new Figure 3i-l, as an effort to first "discuss more of our results collectively". Consequently, new Figure 3m and new Figure 3n were added in the revised manuscript as the suggested "summarizing figure which highlights the key conclusions", for the intuitive understanding of interfacial H sorption processes during CO2RR, different M−H interactions within the catalysts, and protondonating capacities of electrolytes as concluded from electrochemical and ETS investigations. Overall, thanks for the valuable and detailed instructions by the reviewer, we believe the revised manuscript now have more clarified logic, with the addition/rearrangement in the new Figure 3 and the correspondingly updated discussions, we sincerely thank the reviewer for these suggestions.
Next, we moved original Figure S21  "Figure 3b further shows that Pd4Ag has higher formate production rate than Pd at all tested potentials. The above results indicate an overall better CO2RR performance after the alloying with Ag atoms." "Interestingly, when the electrolyte was switched to K2HPO4/KH2PO4, the overall formate FEs on pure Pd were reduced ( Figure 3a). In comparison, Pd4Ag showed even more complex changes in CO2RR performance. The formate FEs also experienced an obvious decrease at low overpotentials (−0.15 to −0 VRHE), yet a unique increase by ~8% was observed at high overpotential region (−0.4 to −0.2 VRHE). Overall, the different CO2RR performances in KHCO3 and K2HPO4/KH2PO4 indicate distinct interfacial processes sensitive to the proton-donating electrolytes, which are presumably connected to the phase transition level of catalysts, CO poisoning/site blocking, formate production rates ( Figure 3b) and HER kinetics (Figure 3c). It should be noted that the XRD after chronoamperometric studies reveal that H sorption processes during CO2RR alloy does not cause a segregation (or any other structural or compositional change) in the Pd4Ag ( Figure S17), indicating its reversibility during the reaction.
ETS measurements were further conducted to elucidate the H sorption kinetics and rationalize the distinct CO2RR performance of different Pd-based catalysts in different electrolyte environments. First, we have compared the H sorption kinetics by ETS in Ar-saturated KHCO3 and HClO4 at varying scan rates (Figure 3d&S18). In HClO4, significantly larger ETS hysteresis loops were observed to achieve complete H adsorption/desorption when the scan rate was increased from 10 to 80 mV/s, which confirms that the H sorption process is kinetic-dependent. In addition, while the degree of phase transition (corresponding to the ETS current level) kept unchanged in HClO4, lower degree of phase transition (dashed arrow in Figure 3d) was observed in KHCO3 with the increasing scan rate. These results suggest the slow H sorption kinetics in near neutral electrolyte, which is reasonable due to the low concentration of hydronium ion and slow kinetics of water reduction in neutral/alkaline electrolytes 46-49 . Figure 3e depicts the CV and ETS curves of Pd under CO2RR conditions (in 0.1 M CO2-saturated KHCO3). An obvious alternation in the H sorption hysteresis loop in ETS can be observed (marked with blue colored area and dashed arrow in Figure 3e) after the introduction of CO2, which may originate from the change of proton source and competitive surface reactions. Specifically, the addition of CO2 increases the concentration of H2CO3 in the electrolyte…" "The nonlinear variation of ISD and phase transition with the negative shift of potential therefore strongly indicate the competition between CO2RR and H sorption at potentials <0 VRHE in KHCO3. Two parallel pathways were generally proposed for Pd-catalyzed CO2RR, leading to formate or CO products 24 : in the formate or formic acid pathway: in the CO pathway: CO 2 +e − +H + + * → * COOH (7) The formate mechanism involves a proton-coupled electron transfer (PCET) process, during which the proton transfer and electron transfer occur in a same elementary step, and the M−H bond is formed on the surface 52-53 .
Although H adsorption on the surface is thermodynamically more favorable (by 0.33 eV) compared to subsurface H absorption (which leads to hydride formation), H could diffuse into the subsurface and then bulk fcc Pd lattice at more negative potentials 18, 54-55 . The overall hydrogen sorption process within Pd and Pd4Ag systems can be described by 56 : where the diffusion between Hads and Hsubsurface follows the equilibrium that is determined by the chemical potentials of H atoms in each phase (μH ads vs. μH subsurface ). During active CO2RR, the C1 intermediates occupy the Pd sites and inhibit the production of Hads, shift the equilibrium between Hads and Hsubsurface, and eventually alter the level of hydride formation. In addition, the more favorable formate pathway (eq 4-6) will largely consume Hads and slow down the kinetics of subsurface H diffusion and phase transition process. Importantly, these H-involving processes during CO2RR can be reflected on the ETS signals corresponding to the H sorptions. On this basis, as the unusual change in ISD (Figure 3g) is in well correspondence with the high formate FE (>70%) in the potential range (>−0.2 VRHE), our results therefore confirm that…" On page 16, line 9: "As summarized in Figure 3m, while the H sorption kinetics is inhibited by CO2RR-related H consumption and/or CO poisoning, it can be promoted by the local proton-donating species including H2CO3 in equilibrium with CO2. It should also be noted that the H sorption is essentially a kinetic-dependent process, and the phase transition potentials may vary under different test conditions (potential scan rates, electrode geometries, etc.). Additional ETS tests indeed show positive shifts of the phase transition potential at smaller film thickness or slower scan rates ( Figure S21), emphasizing the importance of consistence in test conditions. To this end, the ETS measurements conducted with 25 nm thin film thickness and 10 mV/s scan rate represent the experimental condition that gives close-to-intrinsic properties of the electrode materials, where the impact from the insufficient electrolyte diffusion to the sub-layer nanowires within the thin film device was minimized.   The pKa of 0, 10.33, 7.21, 6.35 correspond to H3O + , HCO3 − , H2PO4 − and H2CO3, respectively. d Summary of the time for 50% and 90% level of maximum phase transition in Pd and Pd4Ag. e Proposed mechanism for proton donation, H sorption and CO2RR on Pd-based materials in near neutral conditions. "M-int" in e represents the surface adsorbed intermediates during CO2RR.
As other additional efforts to make the revised manuscript more clarified and easier to read, the "INTRODUCTION" part was slightly modified to better clarify the logic of this whole work, and several sub-titles were added in the "RESULTS AND DISSCUSSION" section in the revised manuscript, as following:

Adsorption regulation
In the Introduction: "Metal−hydrogen (M−H) interactions and the correlated chemical/catalytic hydrogen processes (H adsorption, absorption, evolution and oxidation) participated in multiple applications such as hydrogen fuel cell 1-2 , hydrogen/pH sensors 3-6 , metal hydride batteries 7 , and electrocatalytic hydrogen evolution reaction (HER) and hydrogenation reactions [8][9][10] . To this end, Palladium (Pd) is one of the mostly adapted materials, which serves as a typical model catalyst for the fundamental investigation of M−H states and dynamic hydrogen transitions, owing to its unique and rich interactions with hydrogen 9-13 . Among various Pd-catalyzed electrochemical reactions, electrochemical carbon dioxide reduction (CO2RR) attracts most research attentions as it represents a sustainable means to reduce CO2 emissions by converting it into valuable chemicals and hydrocarbon fuels, providing an effective and economical approach towards carbon neutralizaiton [14][15][16][17]  Despite the significant influence on the adsorption of intermediates in CO2RR and other hydrogenation reactions, there are only few experimental approaches for the quantitative measurement of adsorbed/absorbed H atoms and corresponding H sorption kinetics in Pd-based catalysts under operando conditions. Specifically, the phase transition of Pd is buried at a solid/liquid interface, which is difficult for in situ characterizations and poses particular challenge in the study of corresponding electrocatalytic mechanisms. In most cases, in situ X-ray absorption spectroscopy (XAS) and in situ X-ray diffraction (XRD) were typically employed to characterize the Pd-Pd bond lengths and the lattice expansion during phase transition 17-18, 22, 27-28 , which successfully revealed the impact of catalyst morphologies on the potential range (with difference up to 100~300 mV) for PdHx formation.
AC-impedance 29 , quartz crystal microbalance 30 and cyclic voltammetry 31-33 are the commonly employed approaches for directly studying Pd−H interactions, however each individual methodology typically produces information on restricted dimension. To fully elucidate the comprehensive electrocatalytic mechanisms that includes interfacial chemical processes and the local environments, it is essential to bring up additional in situ approaches (better with alternative signaling mechanism) to complement existing characterization toolbox for the systematic investigation of Pd−H interactions and corresponding hydrogenation processes." In the main text, the following sub-titles were added: "Catalyst preparation and device fabrication", "ETS identification of in situ H sorption processes in perchloric acid" , "Electrochemical CO2RR performances and their kinetic dependence on H sorptions in buffered electrolytes", "Potentiostatic ETS analysis for nearequilibrium operando H sorption quantifications", "Connection between CO2RR performances and H sorption processes".
(2) In Figure 3, the authors attribute the large hysteresis in the electrical transport measurements (and therefore in hydrogen absorption) to be due to the slow kinetics of hydrogen reduction from water in a neutral/alkaline electrolyte. Did the authors perform any additional measurements to corroborate this?
Namely, were CV's and ETS measurements performed at varying scan rates? If the hysteresis is scan rate dependent, this would pretty strongly support it is a kinetic issue. (Are the results in Figure S21 enough to corroborate this?).
Reply: We thank the reviewer for this inspiring suggestion on the ETS measurements at varying scan rates. This is certainly a classic approach used in CV tests to prove the kinetic-controlled electrochemical processes. Indeed, as hydrogen sorption includes the process of surface-adsorbed hydrogen atom gradually diffusing into the lattice, it will be correlated to the scan rate of potential.
The ETS of Pd were tested at varying scan rates in both Ar-saturated HClO4 and KHCO3, the results were added as new Figure 3d, new Figure S18 and new Figure S21 in revised MS and SI. In HClO4, with the scan rates increased from 10 to 80 mV/s, larger hysteresis loops were indeed observed, which confirms that the H sorption process is kinetic-dependent. In addition, the ISD in HClO4 can eventually reach to a same level plateau (indicating same level phase transition degree) at all investigated scan rates, while in sharp contrast, lower degree of phase transition (indicated by the arrow in new Figure S18) is observed in KHCO3 even at a more negative potential (−0.5 VRHE) within the same range of scan rates. This difference further reveals the slow H sorption kinetics in near neutral electrolyte. In addition, more systematic ETS measurements on Pd/Pd4Ag of varying film thicknesses, and with different potential scan rates, were added as new Figure S21, which further reveal the kinetic-dependent nature of the corresponding measurements (and help derive the key parameter intrinsic to the materials). Again, we thank the reviewer for the valuable suggestions. In summary, the ETS results obtained from different scan rates have been added as new Figure S18 and new Figure  &S18). In HClO4, significantly larger ETS hysteresis loops were observed to achieve complete H adsorption/desorption when the scan rate was increased from 10 to 80 mV/s, which confirms that the H sorption process is kinetic-dependent. In addition, while the degree of phase transition (corresponding to the ETS current level) kept unchanged in HClO4, lower degree of phase transition (dashed arrow in Figure 3d) was observed in KHCO3 with the increasing scan rate. These results suggest the slow H sorption kinetics in near neutral electrolyte, which is reasonable due to the low concentration of hydronium ion and slow kinetics of water reduction in neutral/alkaline electrolytes 46-49 ."   (111) and (022) diffractions of graphite from the current collector. There is no obvious peak shift or generation of new diffraction peaks. These results support that the alloy structure of Pd4Ag is well retained after electrolysis without phase segregation. The new data has been added as new Figure S17 and the corresponding discussions have been updated in the revised manuscript, as following:

New
Page 10, line 25: "Overall, the different CO2RR performances in KHCO3 and K2HPO4/KH2PO4 indicate distinct interfacial processes sensitive to the proton-donating electrolytes, which are presumably connected to the phase transition level of catalysts, CO poisoning/site blocking, formate production rates ( Figure 3b) and HER kinetics ( Figure 3c). It should be noted that the XRD after chronoamperometric studies reveal that H sorption processes during CO2RR alloy does not cause a segregation (or any other structural or compositional change) in the Pd4Ag ( Figure S17), indicating its reversibility during the reaction." New Figure S17. XRD patterns of the glassy carbon current collector (red curve), Pd4Ag powders (black curve) and Pd4Ag loaded on glassy carbon current collector after electrolysis at −0.38 V for 1 h in 0.1 M CO2-saturated KHCO3 (green curve) or K2HPO4/KH2PO4 (orange curve).
(4) Paper could use minor copy-editing to improve English writing.
Reply: We thank the reviewer for carefully reading the manuscript and pointing out the issue of English writing.
We have gone through the manuscript and indeed found out many typographical, grammar mistakes and inappropriate expressions in the original manuscript, which have been fixed and highlighted in the revised manuscript. After the editing, we believe the quality of the English writing has been improved. Reply: We greatly appreciate reviewer's positive comments on the novelty/significance of our methodology and investigations, and thank for the valuable suggestions that have inspired us to include key measurements and additional computational results on the subject. In addition, we have made major changes in the revised manuscript to fix the typographical/grammar mistakes and to present a more concise and clarified discussion, as an effort to address the language/writing issue raised by the reviewer. Overall, we hope we can convince the reviewer that the additional experimental/computational data and modified writing have improved the quality of the revised manuscript, making it up to the high standard of this journal.

Major Points:
(1) Diffusion of the electrolyte to the surface of the electrode is potentially problematic for the data presented. To address this issue, we first tested the ETS of Pd in HClO4 and KHCO3 with varying scan rates, as suggested by the reviewer. The corresponding results were added as new Figure 3d and new Figure S18 in revised MS and SI.
The large hysteresis in the ETS measurements (and therefore in hydrogen absorption) were linked to the slow kinetics of hydrogen reduction from water in a neutral/alkaline electrolyte in the original Figure 3. In the new scan-rate dependence measurements, with the scan rates increased from 10 to 80 mV/s, larger hysteresis loops were indeed observed in the ETS of Pd in HClO4, which further confirm that the H sorption process is kineticdependent. In addition, the ISD in HClO4 can eventually reach to a same level plateau (indicating same level phase transition degree) at all investigated scan rates, while in sharp contrast, lower degree of phase transition (indicated by the arrow in new Figure S18) is observed in KHCO3 even at a more negative potential (−0.5 VRHE) within the same scan rate range. This difference further reveals the slow H sorption kinetics in near neutral electrolyte. Overall, it should be noted that the diffusion of the electrolyte has been recognized as an important process in the typical electrochemical voltametric measurement, and our original and new results certainly indicate it is also reflected on the ETS measurements. As a result, the scan-rate dependence data (such as hysteresis loops and the phase change degree) are indeed valuable to probe the H sorption dynamics in Pd-based system.
Second, on the basis of the significant impact of diffusion processes on the ETS measurement, the reviewer is also correct on the issue of variation in the absolute value of "onset potentials" for the phase change of Pd in the ETS tests, which seems to be influenced by the diffusion of electrolyte to the electrode surface (indicated by original Figure S9) and the diffusion of H atoms from the electrode surface into the subsurface lattice. To further address this issue, we further conducted systematic ETS measurements on both Pd and Pd4Ag of varying film thicknesses, and with different potential scan rates, the new results were added as new Figure S21 in the revised SI. For pure Pd in KHCO3, the phase transition potential is positively shifted by ~120 mV when film thickness was reduced from 200 nm to 25 nm (yellow rectangles in new Figure S21a). In addition, higher degree of phase transition and more positive phase transition potentials can be observed with the decreasing scan rate from 80 mV/s to 10 mV/s (yellow rectangle in new Figure S21a). Similar phenomenon can be observed in other tests conditions (yellow rectangles in new Figure S21a-d). All these results show that hydrogen sorption process is essentially a kineticdependent process, and slower scan rates and thinner film thicknesses would benefit the interfacial diffusion of electrolyte or hydrogen atoms. We would also like to emphasize here that, when the film thickness was reduced to ~25nm, it corresponds to ~1-2 layers of nanowires, as confirmed by optical microscopy and AFM ( Figure S8).
In this case, the Pd NWs within the whole film were easily exposed to the electrolyte, and the influence from insufficient diffusion of electrolyte from top to underneath Pd NWs film was presumably reduced to a minimum level. Further reduced thickness leads to discontinuous film that is not conductive to allow for the CV and ETS measurements. In parallel, from the scan-rate dependence measurements, we also observed in the new experiments that the influence of scan rates was gradually reduced with the decreasing scan rates (due to the longer reaction times that allow for more sufficient reaction toward equilibrium states), as clearly seen in new Figure S21e-h. We can therefore conclude that 10 mV/s is a scan rate value that is low enough to minimize its influence on diffusion-controlled shifts in obtained electrochemical potentials. Scan-rates smaller than 10 mV/s led to the instability of the device, especially for the minimum thickness of ~25 nm. In summary, while we are aware of the possible impact of electrolyte diffusion on the absolute value of Habs potentials, the potential values obtained from 25nm thickness device and with 10 mV/s scan rate represent the closest value to the intrinsic properties of the electrode materials, where the impact from the insufficient electrolyte diffusion to the sub-layer nanowires in the thin film device was minimized, and the measured onset potential value for phase transition was solid.
With the data shown in new Figure S18 and new Figure S21, we have further summarized the phase transition potentials to better reveal the intrinsic properties of catalysts and electrolytes with precisely controlled experimental factors. The data was added as new Figure 3n in the revised manuscript. It can be clearly seen that, even with the standard deviation from different scan rates, the statistically average phase transition potential of Pd was about 150 mV more positive than that of Pd4Ag in KHCO3, reflecting the different M−H interactions.
Additionally, by switching the electrolyte from KHCO3 to K2HPO4/KH2PO4, the phase transition potentials of both Pd and Pd4Ag markedly shift to 0~0.1 VRHE, reflecting the strong proton-donation and fast H sorption kinetics in K2HPO4/KH2PO4. This summary indicates that the comparison between different materials and electrolytes were scientifically solid when the experimental conditions were kept constant (film thickness, potential scan rates, etc.), and further reveal that the conclusion to distinguish the intrinsic properties of catalysts and electrolytes becomes more convincing using systematic and statistical evaluations. Again, we thank the reviewer for the valuable suggestions.
The new data have been updated as new Figure 3n, S18 and S21 in the revised manuscript, and the corresponding discussions have been updated in the manuscript, as the following: Page 12, line 8: "ETS measurements were further conducted to elucidate the H sorption kinetics and rationalize the distinct CO2RR performance of different Pd-based catalysts in different electrolyte environments. First, we have compared the H sorption kinetics by ETS in Ar-saturated KHCO3 and HClO4 at varying scan rates (Figure 3d &S18). In HClO4, significantly larger ETS hysteresis loops were observed to achieve complete H adsorption/desorption when the scan rate was increased from 10 to 80 mV/s, which confirms that the H sorption process is kinetic-dependent. In addition, while the degree of phase transition (corresponding to the ETS current level) kept unchanged in HClO4, lower degree of phase transition (dashed arrow in Figure 3d) was observed in KHCO3 with the increasing scan rate. These results suggest the slow H sorption kinetics in near neutral electrolyte, which is reasonable due to the low concentration of hydronium ion and slow kinetics of water reduction in neutral/alkaline electrolytes 46-49 .
Page 16, line 17: "As summarized in Figure 3m, while the H sorption kinetics is inhibited by CO2RR-related H consumption and/or CO poisoning, it can be promoted by the local proton-donating species including H2CO3 in equilibrium with CO2. It should also be noted that the H sorption is essentially a kinetic-dependent process, and (2) The authors make several claims that the electronic structure of the Pd4Ag alloy is responsible for both the reduced CO poisoning and the reduced H absorption capacity. Based on their data, however, a simple geometric effect can explain their observations. The stoichiometry of Pd4Ag1 implies that, in a homogeneous alloy, every palladium atom in an fcc lattice is in contact with a Ag atom. Ag is known to both rapidly evolve CO (due to the

low CO binding energy) and does not absorb hydrogen into it's lattice. A silver atom near every Pd may therefore inhibit CO poisoning and H absorption-based only on it's proximity and not on orbital hybridization.
If the authors are going to claim electronic structure is the origin of their observation, they need to supplement their data with additional computational evidence showing such an effect.
Reply: We thank the reviewer for this insightful input and valuable suggestion. To further reveal the origin of the reduced CO poisoning and H absorption capacity on Pd4Ag, we have conducted the density-functional theory (DFT) calculations as suggested by reviewer, which lead to better understanding of the system.
Specifically, DFT calculations were conducted with the implicit solvent model to study the binding energy of *H and *CO. The supercell of (√13×√7) R19° with five atomic layers was chosen to construct the Pd (111) surface. Two Pd atoms were replaced by Ag atoms in each atomic layer to construct the Pd4Ag surface (new Figure S13). To calculate the binding energies, the H atom and vertical CO molecule were placed at a distance of 1.5 Å and 1.8 Å from the surface. Adsorption energies of *H and *CO on Pd4Ag and Pd surfaces at different sites were further obtained.
As shown in new Figure S13a, c and new  It should also be noted that, at the three-atom hollow site that involves both Ag and Pd atoms, the charge transfer between Ag and the adjacent Pd is inevitable, and therefore the impacts of electronic structure change and geometric effects presumably co-exist, which is also consistent with our previous statements (Adv. Mater. 2021, 33, 2005821). Based on the quantitative comparison of binding energies in different configuration, we can further conclude that at least at the hollow site adsorption configuration, the impact of the proximity effect is more dominant than the change of electronic structure. Overall, we thank the reviewer for making such valuable suggestion that promotes our understanding on this issue. In this study, our ETS results have shown a reduced in situ phase transition degree and CO poisoning effect in Pd4Ag during electrochemical processes as compared to pure Pd, experimentally confirming the weakened M−H and M-CO interactions in Pd4Ag. The corresponding calculation results were added as new Figure S13 and new Table S1 in the revised SI, and the corresponding discussions were added in the revised manuscript, as the following: Page 9, line 9: "As for Pd4Ag alloying catalyst, the onset potential for H absorption is at 0.074 VRHE, which is close to that of Pd. However, the ISD drop in response to hydride formation is considerably lower (Figure 2c), and the △ R(Pd 4 Ag)H x is calculated to be only 0.14%, which is about 4.12% of △RPdH x , clearly indicating a significantly different M−H interaction compared to pure Pd 44 . As shown in Figure 2d, the △R(Pd 4 Ag)H x is lower than that of pure Pd at each film thickness, with a theoretical intercept of 13%. The M−H interactions in Pd4Ag and pure Pd were further revealed by DFT calculations, as shown in Figure S13 and Table S1. that reveal the H sorption kinetics during CO2RR. The onset potential for phase transition shifted negatively from −0.145 VRHE (i in Figure 3h) to −0.324 VRHE (ii in Figure 3h) after the introduction of CO2 in the electrolyte, and both are considerably lower than that of pure Pd. The more negative phase transition potentials of Pd4Ag further prove its weakened M−H interaction after Ag alloying. Additionally, the CO poisoning effect during CO2RR is not observed on ETS, and H sorption kinetics is largely hindered due to its consumption for intense formate production (dashed arrow in Figure 3h). These in situ observations provide convincing evidence for the strong CO poisoning resistance in Pd4Ag alloying catalysts, which results in the considerably improved conversion rate and stability for CO2RR ( Figure S15&S16). As shown in Figure S13&Table S1, our DFT calculation results also confirm that Ag can reduce the binding energy of poisonous *CO at its surrounding sites. Finally, for similar reason to the Pd case, the inhibition by CO2RR leads to a smaller ETS hysteresis loop for H sorption in Pd4Ag (indicated by dashed arrow in Figure 3f)." Page 17, line 24: "Interestingly, similar trend was not observed in Pd4Ag, demonstrating the unique H sorption thermodynamics in response to the H + concentration in this alloying structure (blue bars in Figure 4c). First, due to the alternation in the electronic structure of Pd and the proximity effect after Ag doping, the H binding energy is reduced on Pd4Ag surface, leading to the lower H/M ratio in Pd4Ag as compared to pure Pd (Figure 4c)." In addition to the Ag doping effect, the H atom in sub-surface hydride layer can serve as a doping element that significantly increase the electron density of Pd and reduces the CO adsorption energy (J. Am. Chem. Soc. 2018, 140, 2880−2889Adv. Energy Mater. 2019, 9, 1802840). In this study, Pd4Ag was found to undergo obvious phase transition (H/M: 0.36~0.43) during CO2RR, contributing to its high current density and stability for formate production. Besides, for Pd4Ag that intrinsically has weak M−H interaction, H sorption kinetics and phase transition can be further promoted by utilizing electrolytes with high proton-donating capacity, and finally promote CO2RR.
The corelated discussions have been updated in the revised manuscript, as the following: Page 21, line 22: "Connection between CO2RR performances and H sorption processes. The dynamic, equilibrium and transient quantification of subsurface H/M level (and therefore the H sorption kinetics) can be correlated to the CO2RR performance under each specific condition. We first performed the density functional theory (DFT) calculations ( Figure S14) to rationalize the reduced CO poisoning and high formate production activity on Pd4Ag surface (as compared to pure Pd), demonstrating the role of charge transfer and geometric effects on the reduced M−H interaction and CO poisoning 20 . In addition, we found that both Pd and Pd4Ag exhibit different CO2RR performances in varying H-donating electrolytes (Figure 3a-c). K2HPO4/KH2PO4 can accelerate the H sorption kinetics, which is reflected by the increased phase transition degree (H/M ratio) in Pd4Ag (from 0.38 in KHCO3 to 0.48 in KH2PO4/K2HPO4) during CO2RR (see Figure 4c and Table S2). Correspondingly, an increase in formate FE of about 8% on Pd4Ag was obtained at −0.4~−0.2 VRHE (Figure 3a), proving that strong proton supply in KH2PO4/K2HPO4 electrolyte and high degree in operando phase transition (H doping) would be beneficial for reducing CO poisoning and promoting formate production 18, 24, 28 (Figure 3b). As a result, Pd4Ag demonstrated a significantly enhanced current density and stability for formate production compared with pure Pd (Figure 3a, b).
It should also be mentioned that the enhanced proton reduction kinetics in KH2PO4/K2HPO4, as revealed by ETS investigation, simultaneously promotes the competing HER (Figure 3c) Rev. B 1996, 54, 11169). The projector-augmented wave (PAW) method (J. Phys. Condens. Matter 1994, 6, 8245;Phys. Rev. B 1994, 49, 16223) was applied to describe the electron-ion interactions. A kinetic energy cutoff for the plane wave expansions was set to be 520 eV. The method of Methfessel-Paxton (MP) was applied and the width of the smearing was chosen as 0.2 eV. The supercell of (√13×√7) R19° with five atomic layers was chosen to construct the Pd (111) surface, and two Pd atoms were replaced by Ag atoms in each atomic layer to construct the Pd4Ag surface. More than 10 Å of vacuum space was used to avoid the interaction of the adjacent images. For sampling the reciprocal space, k-points of Γ-centered 4×3×1 were used for surface calculations. All structures were fully relaxed until the force components were less than 0.03 eV·Å -1 . Implicit solvent model was used in our calculations by VASPsol (J. Chem. Phys. 2014, 140, 084106;J. Chem. Phys. 2019, 151, 234101). A Debye screening length of 9.61 Å was chosen, which corresponds to a bulk ion concentration of 0.1 M. The non-electrostatic parameter, TAU, was set to zero for the purpose of convergence.
The adsorption energy of CO is defined as:

Eb,CO=E*CO−E*−ECO
where E*CO, E* and ECO are the total energy of the surface with adsorbed CO, pristine surface and CO molecule in the gas phase, respectively.
The adsorption energy of H is defined as: where E*H, E* and EH 2 are the total energy of the surface with adsorbed H, pristine surface and H2 molecule in the gas phase, respectively." Minor Points: (1) Figure 2, panel a,b,c. Please clearly indicate which y-axis corresponds to the plotted data. As it currently stands, it is very difficult for the reader to interpret these plots. The authors have adressed the majority of my concerns. I believe the paper is now fit for publication. However I would request that the authors address one final point.

Reply
In the discussion of Figure 3, the authors discuss whether CO poisoning may be occuring on each catalyst (Pd and Pd4Ag) during CO2 reduction; this discussion is unclear. My understanding is that with Pd, the authors observe either a shift in the potential of or a decrease in the amount of Habsorption in Pd, and argue this could be due to CO poisoning (preventing Hads and therefore Habs

The authors have addressed the majority of my concerns. I believe the paper is now fit for publication. However
I would request that the authors address one final point.
In the discussion of Figure 3, the authors discuss whether CO poisoning may be occurring on each catalyst (Pd

Reply:
We thank the reviewer for pointing out this issue of clarity in the interpretation of our ETS results. In Figure   3, the effect of switching from Ar to CO2 indeed have "similar effects on the Habs hysteresis loops for both Pd and Pd4Ag", emphasized by the shift of phase transition potential and decrease of ETS current (ISD), leading to the alternated ETS loop characteristics. Hence, it is concluded that CO2RR inhibits H sorption process for both Pd and Pd4Ag. On the other hand, however, the analysis that leads to the discussion of CO poisoning actually comes from the more detailed comparison of fine ETS characteristics within the potential range of -0.2 to -0.4 VRHE region.
First, as shown in the Figure 3g and 3k, an expected smooth linear decrease of ISD (black curves, starting from "i" in Figure 3g&3k) can be observed on Pd in Ar-electrolyte due to the gradually increased level of Habs, which serves as a normal baseline of the H sorption kinetics in Pd. In sharp contrast, we found that this smooth behavior changes to a distinctive nonlinear behavior during CO2RR, with turning points emphasized by "ii, iii and iv" on red curves in Figure 3g&3k. Based on the product analysis in different potential range and existing reaction mechanism of Pd-catalyzed CO2RR, we further assign the different fine regions as the following. For pure Pd: (i) is the onset of phase transition in Ar; (ii) is the onset of phase transition in CO2, where the (i→ii, iii) shift is resulted from H consumption along with CO2RR, and after (iii) the increased H kinetics overweigh the H consumption from CO2RR to give another smooth declining region (iii→iv); and (iv) H sorption during CO2RR accompanied by the presumable surface CO poisoning at high overpotential, where the emerging COads reduces the H kinetics and hence the declining slope to form a turning point iv. In short, we propose here the appearance of the turning point iv and the nonlinear ETS behavior during CO2RR in this fine region is indicative of CO poisoning.
Interestingly for Pd4Ag, the variation trend of fine ETS in the same potential region (Figure 3h and 3l) is relatively simple, and no obvious switch of ETS behavior corresponding to the CO poisoning (emergence of turning point iv and the nonlinear ETS behavior) is observed. Hence, we concluded that there is "a lack of CO poisoning with Pd4Ag" based on the ETS characteristics of Pd/ Pd4Ag under CO2RR conditions, which in consistent with its high working stability and the DFT calculation results, both reported in this work or in literatures.
Overall, ETS results show the competition between H sorption and different surface reactions (CO2RR and CO poisoning). While we are confident with the precision and reproducibility of these fine ETS characteristics, we would also agree with reviewer that the interpretation of the ETS measurements and surface competition processes is not fully "evidencing", and it is uncertain whether there is absolutely no CO poisoning on Pd4Ag only based on ETS results. Meanwhile, this is also not the major point/conclusion in the work. Further rigorous proof of the in situ COads may require the use of other in situ spectroscopies (such as FTIR) that match micro/nano device platform under CO2RR conditions, and the concurrent ETS measurement can then provide strong support from another perspective and help to quantify this process. This is certainly the future direction we will pursue. At this point, to better address this issue and to prevent the over-interpretation of the ETS data, we have now revised the discussion of Figure 3 to make the description more clarified and rigorous. The new Figure 3g,h,k,l were slightly modified for easier visualization, and new discussions have been updated in the manuscript, as the following: Page 14: "Importantly, these H-involving processes during CO2RR can be reflected on the ETS signals regarding H sorptions. On this basis, as the unusual change in ISD (Figure 3g) is in well correspondence with the high formate FE (>70%) in the potential range >−0.2 VRHE, our results therefore confirm that proton consumption and site blocking by intermediate adsorptions during formate production significantly reduce the H diffusion kinetics and level of phase transition under scanning potential (non-equilibrium) condition. When the potential continues to decrease to a more negative potential (−0.3 VRHE, indicated by iv in Figure 3g), the decline of ISD is slowed down (leading to a clear two-stage, non-linear ETS characteristic within the range of −0.2 to −0.4 VRHE) probably due to severe CO poison, which inhibits the production of Hads and subsequent H absorption 32 ." Page 15: "Similarly, Figure 3f depicts the CV and ETS curves of Pd4Ag in 0.1 M Ar-and CO2-saturated KHCO3 that reveal the H sorption kinetics during CO2RR. The onset potential for phase transition shifted negatively from −0.145 VRHE (i in Figure 3h) to −0.324 VRHE (ii in Figure 3h) after the introduction of CO2 in the electrolyte, and both are considerably lower than that of pure Pd. The more negative phase transition potentials of Pd4Ag further confirm its weakened M−H interaction after Ag alloying. Additionally, no change of fine ETS characteristics at high overpotentials (−0.2 to −0.4 VRHE) was observed, which indicates an obvious CO poisoning effect, and H sorption kinetics is expectedly hindered due to the H consumption for intense formate production (dashed arrow in Figure   3h). These in situ observations are also consistent with the strong CO poisoning resistance in Pd4Ag alloying catalysts, which results in the considerably improved conversion rate and stability for CO2RR ( Figure S15&16). As shown in Figure S13 and Table S1, our DFT calculation results also confirm that Ag can reduce the binding energy of poisonous *CO at its surrounding sites [20][21]25 . Finally, for similar reason to the Pd case, the inhibition by CO2RR leads to a smaller ETS hysteresis loop for H sorption in Pd4Ag (indicated by dashed arrow in Figure 3f)." Page 16: "In addition to the proton-donating effect, the influence of formate production and CO poisoning (as a result of the CO2RR process) on H sorption in K2HPO4/KH2PO4 is also presented on ETS curves of Pd (ii, iii and iv in Figure 3k respectively). Again for Pd4Ag, no obvious CO poisoning signal (lack of point iv in Figure 3l) is reflected on ETS in K2HPO4/KH2PO4."