Oxophilic Ce single atoms-triggered active sites reverse for superior alkaline hydrogen evolution

The state-of-the-art alkaline hydrogen evolution catalyst of united ruthenium single atoms and small ruthenium nanoparticles has sparked considerable research interest. However, it remains a serious problem that hydrogen evolution primarily proceeds on the less active ruthenium single atoms instead of the more efficient small ruthenium nanoparticles in the catalyst, hence largely falling short of its full activity potential. Here, we report that by combining highly oxophilic cerium single atoms and fully-exposed ruthenium nanoclusters on a nitrogen functionalized carbon support, the alkaline hydrogen evolution centers are facilely reversed to the more active ruthenium nanoclusters driven by the strong oxophilicity of cerium, which significantly improves the hydrogen evolution activity of the catalyst with its mass activity up to −10.1 A mg−1 at −0.05 V. This finding is expected to shed new light on developing more efficient alkaline hydrogen evolution catalyst by rational regulation of the active centers for hydrogen evolution.

Shen et al. report on the synthesis of Ce single atoms and Ru nanoclusters on a N functionalized carbon and apply it for the alkaline HER.The claim of a record mass activity by the authors is not correct.Furthermore, the EXAFS and XANES data do not fit the assumed model of the catalyst.The DFT input model and DFT overpotentials do not align with the experimental analytic and catalytic data.Therefore, the manuscript in its current form is not suitable for publication in any journal.
Here are some of the main issues: 1.The authors cite the following manuscript in their main text: Li et al.J. Mater.Chem.A, 2021,9, 12196-12202.They also include a table to show the superior mass activity of their work compared to previous literature.Li et al.'s work is not shown in this Table (Table S1  and S2) and also not in Figure 4e of the main text.The authors show only works with a worse mass activity.However, Li et al. report a substantially better mass activity (17 A/g at 0.025 V overpotential).This publication was the only one I checked.There might be many other works that reported better mass activities and that have been missed by the authors.
2. Is most of the Ru really in the form of metallic clusters?The XANES edge is way too high for metallic Ru.I cannot see how the XANES data would agree with the following statement: "It was demonstrated in Figure S8 that electrons could be facilely transferred from Ce to Ru with a net electron transfer number of 0.05, which agreed well with the XANES data." 3. The authors claim that they have Ru clusters in their main compounds.In the DFT model, these clusters comprise more than 12 Ru atoms.This contradicts the EXAFS fitting, where Ru has, on average, only two neighbours.In a cluster of 12 or more, this number is tremendously larger.Thus, the majority of Ru is not present in the form of the assumed clusters.
4. The theoretical calculations do not align with the experimental observations.1.The clusters are too large compared to the EXAFS coordination number.Figure 5e implies an overpotential of almost 400 mV, which does not fit to the experimentally observed one at all.Therefore, the DFT analysis is without any meaning for the real system.
5. The plot of the long-term stability measurement (Figure 4f) is missing ticks on the right yaxis to make it easy to read how much the sample deactivates.Using a ruler, I could find out that there is a substantial deactivation in the range of 50-100 mV.If the authors want to know if their material would be stable under industrial conditions, they must measure it at 80 °C and a current density of 400 mA/cm².6.The general statement that the industry would prefer alkaline electrolysis is not true and is an oversimplification.It should be changed.Alkaline and PEM both have their advantages for industrial applications.
7. Replacing Platinum by Ruthenium is not a real advantage as they are similarly rare in the earth's crust.8. Tafel slopes cannot be reliably determined from potentiodynamic methods.They must be measured under steady-state conditions with various CA measurements (see 10.1021/acsenergylett.1c00608 or /10.1016/j.mtener.2022.101123).

Reviewer 2 Overall comments:
The authors presented that combining oxophilic Ce single atoms and fully exposed Ru nanoclusters leads to the outstanding alkaline HER activities by facilely reverse alkaline hydrogen evolution centers.They investigated the effects of oxophilic Ce and exposed Ru nanoclusters for facilitating water dissociation and alkaline hydrogen evolution.The authors provided the in-depth theoretical calculations and experimental results in this work.However, I think that this work lacks some important characterization analysis, and needs a major revision.The comments below should be addressed to be published on high-level Nature Communications.Response: We thank the Reviewer very much for reading through our manuscript and providing such valuable comments on our manuscript, which greatly helps us to improve our manuscript.In the revised manuscript and supporting information, we have updated the important characterizations as suggested by the Reviewer including the higher magnification EDS mapping images of the Ce1-Run/NC catalyst, the survey XPS spectra of the Ce1-Run/NC catalyst, the detailed characterizations for the spent Ce1-Run/NC catalyst after long-term HER measurement, and the CV stability test.Please see below our point-by-point responses to the Reviewer's concerns.

Comment 1:
In introduction part, the authors illustrate that the Ru1-Run catalyst can facilitate the water dissociation with a quite low Gibbs free energy barrier, in which the Ru single atoms (SAs) adsorbed the H while the small Ru nanoparticles (NPs) bonded with the OH, as illustrated in Scheme 1.However, they conflictingly illustrate that the Ru NPs of the Ru1-Run catalyst merely took part in water dissociation and just acted as spectators of hydrogen evolution, besides the Ru SAs were less active to produce hydrogen than the small Ru NPs.Although these conflict arguments have been unveiled in the cited references , the role of Ru SAs and Ru NPs can be tuned in various matrix for optimizing alkaline HER activity.The Ru NPs dominates the H2 evolution compared to the Ru SAs on the metal oxide-based substrates such as sodium cobalt oxide that accelerates the water dissociation.[Adv. Mater.2023, 2301133] In contrast, on the carbon substrates such as defective carbon, the Ru SAs optimize the H2 evolution while the Ru NPs facilitates the water dissociation.[Adv.Sci.2021,8, 20045] Given that the authors assert the reversed active centers to Ru nanoclusters by the oxophilic Ce SA as a main point of this study, the authors should make further detailed and clear explanations on these conflicting arguments.Response: Thanks a lot for the Reviewer's insightful comment.To better contextualize the work and highlight our idea of this study, we have included more detailed discussions in the revised manuscript.Firstly, there are some conflicting arguments regarding the hydrogen evolution mode over the Ru1-Run catalyst in previous reports, which was primarily derived from the influence of varied supports as was also pointed by the Reviewer.A general trend from a pure perspective of reaction energy was that Ru single atom itself was insufficient for dissociating water molecules (rate-determining step of alkaline HER) in terms of its huge energy barrier for water dissociation, due to lacking dual metal binding sites for simultaneously adsorbing H and OH group of water.By contrast, Ru nanoclusters with continuous Ru sites were more efficient for water dissociation with moderate energy barriers and meanwhile the Ru nanoclusters possessed more favorable H adsorption energy relative to that of the Ru single atoms.To better depict this point, we have provided more detailed DFT calculations.
As illustrated in Figure R1a, the Ru single atom (Ru1) exhibited a particularly high Gibbs free energy barrier for water dissociation (up to 1.03 eV) while the value of the Run nanocluster (n=13 in our case, corresponding to a Run nanocluster of about 1 nm) was only 0.2 eV.In addition, water dissociation over the Run nanocluster was the thermodynamically favorable exothermic process with an exothermic energy of 0.49 eV.Whereas, water dissociation was endothermic by 0.67 eV on the Ru1 site.These results suggested that the Run nanoclusters were much more efficient for disassociating water than the Ru single atoms.On the other hand, the H adsorption energy on the Run nanocluster was -0.5 eV (Figure R1b), while the H adsorption energy on the Ru1 site was -0.7 eV.These results revealed that the Run nanoclusters were more active for hydrogen evolution than the Ru single atoms both in terms of its favorable reaction energies for water dissociation and for H adsorption.
However, as far as the dual Ru1-Run catalyst was concerned, the scenario became quite different.To begin with, water dissociation theoretically had two possible patterns on the dual Ru1-Run site: i) the OH group bonded with Ru1 while the H atom bonded with Run (Ru1OH-RunH route); and ii) the H atom bonded with Ru1 while the OH group bonded with Run (Ru1H-RunOH route).To examine which route was the more energy-favored reaction pathway, we have further carried out the DFT calculations for both of the two routes.As displayed in Figure R2, the Gibbs free energy barrier for water dissociation via the Ru1H-RunOH route was 0.49 eV, which was markedly lower than that via the Ru1OH-RunH route (up to 0.7 eV), unveiling that the Ru1H-RunOH route was the more kinetically favored reaction pathway for water dissociation.On the other hand, it was found that water dissociation via the Ru1H-RunOH route was exothermic by 1.42 eV, while water dissociation via the Ru1OH-RunH route was exothermic by only 0.03 eV, showing a 47-fold difference between them.This result further uncovered the more favorable thermodynamics for water dissociation via the Ru1H-RunOH route relative to that via the Ru1OH-RunH route.Therefore, driven by the more beneficial reaction kinetics and thermodynamics of the Ru1H-RunOH route, H would be bonded with the Ru1 and OH would be bonded with the Run after water dissociation, which made further hydrogen evolution occur on the Ru1 side of the dual Ru1-Run site.Whereas, this kind of hydrogen evolution mode via the Ru1H-RunOH route had a major disadvantage that hydrogen evolution proceeded on the less reactive Ru1 sites instead of the more reactive Run sites as described above in Figure R1a-b, which not only decreased the overall alkaline HER activity but unnecessarily required more usage of the precious Ru metal.To this end, new strategies are highly desired to reverse the hydrogen evolution centers from the Ru single atom side to the more reactive Run nanocluster side of the Ru1-Run catalyst to attain more efficient and cost-effective alkaline hydrogen evolution.Intriguingly, in current work, we found that by uniting the highly oxophilic Ce single atoms and the ultrafine Run nanoclusters on a N functionalized carbon (NC) support, the hydrogen evolution sites were facilely reversed to the Run side due to the strong affinity of Ce to OH as verified both by the DFT calculations and the in situ Raman measurements in the main text, which greatly boosted the alkaline HER activity of the Ce1-Run/NC catalyst.In another respect, it should be noted that the N functionalized XC-72 carbon support used in our work was highly graphitized in view of its low ID/IG value compared with the commercial XC-72 carbon of high graphitization degree as demonstrated in Figure R3, which largely excluded the influence of otherwise defective supports.We have also added these contents to the revised manuscript as read in Line 18, Page 2, "From a perspective of reaction kinetics and thermodynamics, water dissociation over the dual Ru1-Run site was more prone to proceed via a "Ru1H-RunOH route" with OH of water adsorbed on the Run side while the H bonded with the Ru1 side as suggested by the theoretical calculations in Figure S1, making further hydrogen evolution primarily occur on the Ru1 side of the dual Ru1-Run catalyst.However, this kind of hydrogen evolution mode had a major disadvantage that the Ru1 sites were less active for hydrogen evolution than the Run sites both in view of the more favorable H adsorption energy and the lower energy barrier for water dissociation over the Run sites than that over the Ru1 sites as displayed in Figure S2, which led to significant decrease of the overall alkaline HER activity and also unnecessarily required more usage of the precious Ru metal.To this end, new strategies are highly desired to reverse the hydrogen evolution centers from the less reactive Ru single atom side to the more reactive Run nanocluster side of the Ru1-Run catalyst to attain more efficient and cost-effective alkaline hydrogen evolution, which is urgently awaited to be explored."

Comment 2:
To address the mentioned conflict arguments, please present a comparison between Ce1-Run/NC and Ru1-Run/NC by providing the experimental results and theoretical calculations including the HER performance and Gibbs free energy diagrams.Response: Thanks for the Reviewer's valuable comments and suggestions.As suggested by the Reviewer, we have prepared a Ru1-Run/NC catalyst by reducing the Ru 3+ impregnated NC support at 180 o C with hydrogen (synthetic details were provided in methods part of the main text).As displayed by the HAADF-STEM image of the synthesized Ru1-Run/NC catalyst in Figure R4a-b, ultrafine Run nanoclusters were uniformly dispersed on the NC support with an average particle size of 0.9±0.2nm, which was similar to that of the Run nanoclusters (1.0±0.2 nm) in the Ce1-Run/NC catalyst.Moreover, the aberration-corrected HAADF-STEM image of the Ru1-Run/NC catalyst in Figure R4c-d clearly showed the copresence of Ru single atoms and Run nanoclusters in the Ru1-Run/NC catalyst.The loading amount of Ru in the Ru1-Run/NC catalyst was measured to be 1.1wt.%by the ICP-OES technique, which was also in line with that of the Ce1-Run/NC catalyst (1.1wt.%).Following the kind advice of the Reviewer, we have further evaluated the alkaline HER activity of the Ru1-Run/NC catalyst.As displayed in Figure R5a and b, the alkaline HER activity of the Ru1-Run/NC catalyst showed an obvious decrease relative to that of the Ce1-Run/NC catalyst both reflected by their linear sweep voltammetry (LSV) curves and corresponding turnover frequency (TOF) value curves of them.To gain theoretical insight into the catalytic activity difference between the Ce1-Run/NC catalyst and the Ru1-Run/NC catalyst, we have then conducted the first-principle DFT calculations.As presented in Figure R6a, the water dissociation over the dual Ce1-Ru13 site (structural model of the Ce1-Run/NC catalyst) was highly exothermic by 4.5 eV, which was 3.2 times that of the dual Ru1-Ru13 site (structural model of the Ru1-Run/NC catalyst), suggesting the Ce1-Run/NC catalyst was much more thermodynamically favorable to dissociate water molecules than the Ru1-Run/NC catalyst.Moreover, the water dissociation energy barrier of the dual Ce1-Ru13 site (0.1 eV) was also markedly lower than that of the dual Ru1-Ru13 site (0.49 eV).These results unveiled both the beneficial reaction thermodynamics and kinetics of the Ce1-Run/NC catalyst for water dissociation than the Ru1-Run/NC catalyst.On the other hand, the H adsorption energy of the dual Ce1-Ru13 site was closer to the optimum value (0 eV) than the dual Ru1-Ru13 site as shown in Figure R6b.In the meanwhile, the electron localization function (ELF) analysis (inset of Figure R6b) disclosed the weaker degree of electron localization in the Ru-H bonding region for the dual Ce1-Ru13 site than that of the dual Ru1-Ru13 site.These results synergistically benefited the alkaline hydrogen evolution over the Ce1-Run/NC catalyst.It was also identified in Figure R6a that the OH desorption energy barriers of the dual Ru1-Ru13 site and the dual Ce1-Ru13 site were quite similar (0.9 eV vs. 1.0 eV), which showed a weak contribution to their alkaline HER activity difference.We have also updated these contents to the revised manuscript as read in Line 4, Page 10, "To compare the alkaline HER activity of the Ce1-Run/NC catalyst with the state-of-the-art Ru1-Run/NC catalyst, we have also prepared a Ru1-Run/NC catalyst by reducing the Ru 3+ impregnated NC support at 180 o C with hydrogen (please see methods part for more details).The HAADF-STEM images and aberration-corrected HAADF-STEM images of the Ru1-Run/NC catalyst in Figure S27 unveiled the copresence of the uniformly dispersed Ru single atoms and Run nanoclusters (0.9±0.2 nm) in the catalyst.It was found that the alkaline HER activity of the Ru1-Run/NC catalyst was markedly lower than that of the Ce1-Run/NC catalyst as suggested both by the LSV curves and the mass activity curves of the two catalysts in Figure S28."Line 1, Page 13, "The DFT calculations suggested that the exothermic energy of the dual Ce1-Ru13 site (4.5 eV) was 3.2 times that of the dual Ru1-Ru13 site (1.42 eV) in water activation process, which unveiled that the dual Ce1-Ru13 sites were more thermodynamically favorable to dissociate water molecules than the dual Ru1-Ru13 sites.In addition, the water dissociation energy barrier of the dual Ce1-Ru13 site (0.1 eV) was also markedly lower than that of the dual Ru1-Ru13 site (0.49 eV).Therefore, the dual Ce1-Ru13 sites were both more thermodynamically and kinetically beneficial to promote water dissociation relative to the dual Ru1-Ru13 sites."Line 22, Page 13, "As shown in Figure 5e, the ∆GH* of the dual Ce1-Ru13 site was obviously closer to the optimum value (0 eV) than the dual Ru1-Ru13 site, which made the Ce1-Run/NC catalyst more efficient for hydrogen evolution."Line 39, Page 13, "Furthermore, the electron localization function (ELF) analysis (inset of Figure 5e) disclosed that a weaker degree of electron localization in the Ru-H bonding region was identified for the dual Ce1-Ru13 site than the dual Ru1-Ru13 site, thus weakening the hydrogen binding strength thereon for more favorable hydrogen production.It was also identified in Figure 5b that the OH desorption energy barriers of the dual Ru1-Ru13 site and the dual Ce1-Ru13 site were quite similar (0.9 eV vs. 1.0 eV), which showed a weak influence on their alkaline HER activity difference."Comment 3: Some papers regarding to the alkaline HER are suggested into the introduction part after the revision process, including Nat.Mater. 2012, 11, 550−557, Mater. Today 2020, 6, 125-138, J. Am. Chem. Soc. 2016, 138, 16174-16181, J. Am. Chem. Soc. 2021, 143, 1399-1408, Adv. Sci. 2018, 5, 1700464, J. Mater. Chem. A, 2019,7, 14971-15005.Response: Thanks for the Reviewer's kind reminding.We have cited these important references in the revised manuscript as was also listed below: Response: Thanks for the Reviewer's kind suggestion.In the revised manuscript, we have included the descriptions for the synthetic procedures in the main text as read in Line 17, Page 3, "In preparation, a N modification procedure was firstly conducted to functionalize the XC-72 carbon (denoted as NC support, please see methods part for details).Ce precursors were then introduced to the NC support by impregnation and allowed for further hydrogen reduction at 700 o C.After acid etching, the NC supported Ce single atoms were acquired (Ce1/NC).To synthesize the united catalyst of oxophilic Ce single atoms and fully-exposed small Ru nanoclusters (Ce1-Run/NC), Ru precursors were then impregnated onto the Ce1/NC and reduced by hydrogen at 250 o C. As a control, the Run nanoclusters were also prepared on the NC support (denoted as Run/NC) using a similar method to the synthesis of the Ce1-Run/NC catalyst except for the absence of Ce.The detailed synthetic process was further displayed in Figure S3 and in the methods part, respectively."

Comment 5: The figure 1a shows the schematic illustration of the structural models of the Ce1-Run/NC, but the author cited the figure 1a with explanations on the ICP-OES results. It should be corrected. The author should provide the results of ICP-OES as an additional table in SI.
Response: Thanks a lot for the Reviewer for pointing out the misleading description.In the revised manuscript, we have corrected the sentence to "The schematic models of the Ce1/NC, Run/NC and Ce1-Run/NC were shown in Figure 1a, Figure 1h and Figure 1o, respectively."In addition, as suggested by the Reviewer, we have also provided the metal loadings of the Ce1/NC, Run/NC, Ru1-Run/NC and Ce1-Run/NC catalysts as acquired by the ICP-OES measurement in Table R1 (also in Table S1 of the revised supporting information).

Table R1.
The metal loading amount of the main catalysts in current work as measured by the ICP-OES technique.

Catalyst
Ru loading (wt.%)Ce loading (wt.%) 1.1 0.03 The metal loading was determined by the inductively coupled plasma optical emission spectrometer (ICP-OES) measurement.Response: Thanks for the Reviewer's helpful comment.In the revised manuscript, we have included the discussion about the Ce-Ce coordination both for the Ce1-Run/NC catalyst and the reference CeO2.As demonstrated by the Fourier transforms of the Ce L-edge EXAFS oscillations in Figure R8a (also in Figure 2g of the revised manuscript), a coordination peak in the second coordination shell at 3.8 Å was identified for the reference CeO2.Corresponding data fitting of the reference CeO2 EXAFS oscillation curve unveiled that this coordination peak was assigned to the Ce-Ce coordination.On the other hand, it was found that the Ce1-Run/NC catalyst presented a coordination peak in the second coordination shell at 3.2 Å, which was shorter than that of the Ce-Ce coordination as displayed in Figure R8a.Careful data fitting showed that this peak was attributed to the Ce-C coordination of the Ce1-Run/NC catalyst in the second coordination shell.To further confirm this, we have carried out the high-resolution EXAFS wavelet transform (WT) characterizations both for the Ce1-Run/NC catalyst and the reference CeO2.As shown in Figure R8b, the Ce1-Run/NC catalyst presented an intensity maximum at a K value of 2.3 Å −1 that was derived from the Ce-C contribution in the second coordination shell.In sharp contrast, the Ce-Ce contribution in the reference CeO2 induced an intensity maximum at a much higher K value of 7.7 Å −1 .This result again revealed the absence of Ce-Ce bond in the Ce1-Run/NC catalyst.
We have also added these contents to the revised manuscript as read in Line 32, Page 6, "We have also examined the Ce-Ce coordination in the Ce1-Run/NC catalyst and the reference CeO2, respectively.As indicated by the Fourier transforms of the Ce L-edge EXAFS spectra in Figure 2g, a coordination peak at 3.8 Å was identified for the reference CeO2, which was assigned to the Ce-Ce coordination by data fitting.Meanwhile, the Ce1-Run/NC catalyst presented a coordination peak at 3.2 Å that was markedly shorter than that of the Ce-Ce coordination.Careful data fitting suggested that this coordination peak of the Ce1-Run/NC catalyst was derived from the Ce-C contribution in the second coordination shell.To further confirm this, we have conducted the high-resolution EXAFS wavelet transform characterizations.As demonstrated in Figure S16, the reference CeO2 displayed an intensity maximum at 7.7 Å −1 (derived from the Ce-Ce contribution) while the Ce1-Run/NC catalyst showed an intensity maximum at 2. Response: We thank the Reviewer for this helpful comment.As displayed in Figure R9a, we have provided the full XPS spectrum of the Ce1-Run/NC catalyst.It was found that the N 1s and C 1s XPS peaks of the Ce1-Run/NC catalyst were clearly identified both from the survey XPS spectrum and from the high-resolution N 1s and C 1s XPS spectra in Figure R9b and c, respectively.However, due to the low mass loadings of Ce in the Ce1-Run/NC catalyst, no Ce XPS peaks were found.Fortunately, the Ce LIII-edge XANES spectra in Figure R9d (also in Figure 2f) confirmed the presence of the Ce element in the Ce1-Run/NC catalyst thanks to the low detection limit of the synchrotron radiation technique.
We have also included these contents in the revised manuscript as read in Line 5, Page 7, "The full XPS spectrum of the Ce1-Run/NC catalyst was also provided in Figure S17, wherein the N 1s and C 1s XPS peaks were evidently identified.However, due to the low mass loadings of Ce in the Ce1-Run/NC catalyst, no Ce XPS peaks were found in the survey XPS spectrum.Thanks to the low detection limit of the aforementioned XAFS measurement, the presence of Ce in the Ce1-Run/NC catalyst was well verified as demonstrated in Figure 2f."Response: Thanks for the Reviewer's valuable comment.As suggested by the Reviewer, we have provided the metal loadings of the tested catalysts in Table R2 (also in Table S3 of the revised supporting information).During the calculations, the mass activities of these catalysts were normalized to the mass of noble metals (Ru or Pt in current case) as commonly employed by previous reports.The calculation details were also provided below and in the revised supporting information.
Comment 11: The Ce1-Run/NC catalyst displayed excellent catalytic stability during chronopotentiometry at 150 mA cm -2 for 100 hr.The author should provide the important characterization results of the catalyst after long-term HER process to reveal its stability.In addition.please provide the additional stability test of continuous cyclic voltammograms (> 1000 cycles).Response: As suggested by the Reviewer, we have employed a variety of techniques to characterize the Ce1-Run/NC catalyst after the stability test.Firstly, the high-angle annular darkfield scanning transmission electron microscopy (HAADF-STEM) images of the spent Ce1-Run/NC catalyst in Figure R10a-b showed the uniformly dispersed Run nanoclusters with an average particle size of 1.0±0.2nm in the catalyst, which was in good agreement with its initial particle size (1.0±0.2 nm).Other than that, the aberration-corrected HAADF-STEM observation of the post-reaction Ce1-Run/NC catalyst in Figure R10c-d further revealed the copresence of the homogeneously distributed Ce single atoms and the Run nanoclusters on the NC support, which was also confirmed by corresponding EDS elementary mapping images in Figure R10e-g as the Ru element signal was primarily located on the small nanoclusters and the Ce element signal was overlapped with the dispersion of single atoms.These characterization results unveiled the robust stability of the Ce1-Run/NC catalyst.
In addition, we have further carried out the continuous cyclic voltammograms measurements for the Ce1-Run/NC catalyst following the Reviewer's kind suggestion.As shown in Figure R11, after 6000 cyclic voltammogram cycles, the obtained LSV curve of the Ce1-Run/NC catalyst remained to be nearly overlapped with its initial one, again indicating the excellent durability of the Ce1-Run/NC catalyst.
We have also included these contents in the revised manuscript as read in Line 24, Page 10, "We have also characterized the Ce1-Run/NC catalyst after the stability test using a variety of techniques.To begin with, as displayed by the HAADF-STEM images of the spent Ce1-Run/NC catalyst in Figure S29a-b, uniformly dispersed Run nanoclusters with an average particle size of 1.0±0.2nm was identified, agreeing well with its initial particle size (1.0±0.2 nm).Moreover, the aberration-corrected HAADF-STEM images of the post-reaction Ce1-Run/NC catalyst in Figure S29c-d unveiled the coexistence of the homogeneously distributed Ce single atoms and the Run nanoclusters on the NC support, which was also verified by corresponding EDS elementary mapping images in Figure S29e-g.These characterization results demonstrated the robust stability of the Ce1-Run/NC catalyst.In addition, we have also performed the continuous cyclic voltammogram measurement to examine the stability of the Ce1-Run/NC catalyst.As shown in Figure S30, after 6000 cyclic voltammogram cycles, the LSV curve of the Ce1-Run/NC catalyst remained to be nearly overlapped with its initial curve, which again disclosed the excellent durability of the Ce1-Run/NC catalyst."Response: Thanks for the Reviewer's valuable comments and suggestions.In Figure 5b, the *H+*OH indicated the adsorbed H and the adsorbed OH on catalyst, respectively.The *H-OH represented for that the H was still adsorbed on the catalyst while the OH was undergoing the process of desorption.To make it clear, we have also added these descriptions to the revised manuscript, as read in Line 10, Page 11, "The *H2O indicated the adsorbed water molecules on the catalyst.The *H+*OH indicated the adsorbed H and the adsorbed OH on the catalyst, respectively.The *H-OH represented for that the H was still adsorbed on the catalyst while the OH was undergoing the process of desorption." To better show the reaction details for each step, we have provided the schematic illustrations for alkaline hydrogen evolution over the dual Ce1-Ru13 site and over the dual Ru1-Ru13 site, respectively, in Figure R12f below and also in the Figure 5f of the revised manuscript.

Reviewer 3 Overall comments: Shen et al. report on the synthesis of Ce single atoms and Ru nanoclusters on a N functionalized carbon and apply it for the alkaline HER. The claim of a record mass
activity by the authors is not correct.Furthermore, the EXAFS and XANES data do not fit the assumed model of the catalyst.The DFT input model and DFT overpotentials do not align with the experimental analytic and catalytic data.Therefore, the manuscript in its current form is not suitable for publication in any journal.Response: We greatly appreciate the Reviewer's valuable comments, which helped us to improve the quality of our work.Also, we sincerely apologize for missing out the papers with higher mass activity than our work as mentioned by the Reviewer.In the revised manuscript, we have deleted the description of a record mass activity and cited more than 100 papers published in recent years to make an exact comparison of the mass activity between our work and previous reports.In addition, we have reconstructed the DFT models in the revised manuscript to better fit the EXAFS and XANES data.Other than that, we have provided detailed explanation on the difference between the overpotentials acquired by the theoretical calculations and the experiment measurements.Furthermore, we also tried to calculate the overpotential using a new "CANDEL implicit solvation model" method by means of which the calculated overpotential value was closer to that acquired by the experiment test.Please see below our point-by-point responses to the Reviewer's concerns.

Comment 1:
The authors cite the following manuscript in their main text: Li et al. J. Mater. Chem. A, 2021,9, 12196-12202.They also include a table to show the superior mass activity of their work compared to previous literature.Li et al.'s work is not shown in this Table (Table  S1 and S2) and also not in Figure 4e of the main text.The authors show only works with a worse mass activity.However, Li et al. report a substantially better mass activity (17 A/g at 0.025 V overpotential).This publication was the only one I checked.There might be many other works that reported better mass activities and that have been missed by the authors.Response: Thanks for the Reviewer's kind reminding and we sincerely apologize for missing out the paper (J.Mater.Chem.A, 2021,9, 12196-12202) as pointed by the Reviewer.In the revised manuscript, we have deleted the claim of a record mass activity and cited more than 100 papers published in recent years hopefully to make an exact comparison of the mass activity between our Ce1-Run/NC catalyst and previous reports on Ru-based alkaline HER catalysts.As displayed in Table R3 and Table R4, there were indeed some better results in previous reports than our work in terms of mass activity.On the other hand, it was noteworthy that the mass activities of our Ce1-Run/NC catalyst measured at -0.1 V and -0.05 V (RHE) were superior to most of the reported Ru-based alkaline HER catalysts.We have also updated these contents in the revised manuscript as read in Line 13, Page 10, "To make an exact comparison of mass activity between the Ce1-Run/NC catalyst and previously reported Ru-based alkaline HER catalysts, we have cited more than 100 papers published in recent years as summarized in Table S4-5 and Figure 4e.It could be identified that the mass activities of the Ce1-Run/NC catalyst measured at -0.1 V and -0.05 V were superior to most of the reported results while there were some better results than the Ce1-Run/NC catalyst."Response: We thank the Reviewer for these very helpful comments.To answer these questions, we have firstly provided the best-fitted EXAFS results of the Ru catalysts in Table R5 (also in Table S2 of the supporting information).It could be found that the mean Ru-Ru coordination number of the Run/NC catalyst was 2. By contrast, the Ru-Ru coordination number of a 1 nm pure Run nanocluster in our case (corresponding to a Ru13 nanocluster as shown in Figure R13) was calculated to be 6.5.Calculation details: the corner Ru atom numbers of the Ru13 nanocluster were 12 and each corner Ru atom possessed a Ru-Ru coordination number of 6.Thus, the total Ru-Ru coordination numbers of the corner Ru atoms were 12*6=72.By contrast, the center Ru atom of the Ru13 nanocluster possessed a Ru-Ru coordination number of 12, which made the total Ru-Ru coordination numbers of the Ru13 nanocluster as 72+12=84.As such, the average Ru-Ru coordination numbers of the Ru13 nanocluster would be 84/13=6.5.Since only Ru-Ru and Ru-N coordination were identified in the Run/NC catalyst, the reduction of its Ru-Ru coordination number was most likely derived from the presence of Ru single atoms, because the Ru single atoms were solely coordinated with N and thus decreased the mean Ru-Ru coordination number of the Run/NC catalyst.Careful aberration-corrected HAADF-STEM observation of the Run/NC catalyst in Figure R14 also verified the presence of Ru single atoms in the Run/NC catalyst.On the other hand, the existence of the partially positively charged Ru single atoms in the Run/NC catalyst inevitably increased the average oxidation state of Ru, because of which the edge absorption energy of the Run/NC catalyst showed a significant increase relative to that of the Ru foil as pointed by the Reviewer.CN is the coordination number, R is the average distance for different coordination pairs, σ 2 is the Debye-Waller factor, and ∆E0 is the inner potential correction.The accuracies of the above parameters are estimated as CN, ±20%; R, ±1%; σ 2 , ± 20%; ∆E0 , ±20%.The data range used for data fitting in k-space (∆k) and R-space (∆R) are 3.0-13.0Å −1 and 1.0-2.9Å for Ru, 3.0-8.0Å −1 and 1.0-2.0Å for Ce, respectively.We have also added these contents to the revised manuscript, as read in Line 11, Page 5: "On the other hand, as suggested by the best fitted Fourier transforms of the Ru K-edge extended XAFS (EXAFS) spectrum in Table S2 (corresponding fitting curves were shown in Figure S9-10), the mean Ru-Ru coordination number of the Run/NC catalyst was about 2, which was markedly lower than that of a pure 1 nm Run nanocluster (6.5, n=13) as displayed in Figure S11.This result manifested the copresence of Ru single atoms and Run nanoclusters in the Run/NC catalyst because the Ru single atoms were solely coordinated with N and thus decreased the mean Ru-Ru coordination number of the Run/NC catalyst.Careful aberrationcorrected HAADF-STEM observation of the Run/NC catalyst in Figure S12 also revealed the presence of Ru single atoms in the Run/NC catalyst.The partially positively charged Ru single atoms in the Run/NC catalyst inevitably increased the average oxidation state of Ru, because of which the edge absorption energy of the Run/NC catalyst showed a significant increase relative to that of the Ru foil." For the second question, we would like to give more detailed description concerning the sentence "It was demonstrated in Figure S8 that electrons could be facilely transferred from Ce to Ru with a net electron transfer number of 0.05, which agreed well with the XANES data."As displayed in Figure 2a, the edge energy of the Ce1-Run/NC catalyst presented a negative shift compared with that of the Run/NC catalyst, which suggested that Ru component in the Ce1-Run/NC catalyst was more electron-rich than that in the Run/NC catalyst.Considering the only difference of Ru between the Ce1-Run/NC catalyst and the Run/NC catalyst was that the Ce1-Run/NC catalyst contained Ce single atoms, we thus concluded that the mean electron density enhancement of Ru in the Ce1-Run/NC catalyst was originated from the Ce electron donation in view of the larger electronegativity of Ru (2.2) than Ce (1.1).To check this assumption, we have performed the Bader charge analysis as displayed in Figure S13 in the revised manuscript.It was found that the Ce electrons could be facilely transferred to Ru with a net electron transfer number of 0.05, which contributed to the mean electron density enhancement of Ru in the Ce1-Run/NC catalyst relative to that in the Run/NC catalyst.The result of the mean electron density enhancement of Ru in the Ce1-Run/NC catalyst as calculated by the Bader charge analysis was in good agreement with corresponding experimental data acquired by the XANES measurement and that was what we want to describe.To avoid the misleading descriptions as pointed by the Reviewer, we have rewritten these contents in the revised manuscript as read in Line 2, Page 6: "It was further found in Figure 2a that the edge energy of the Ce1-Run/NC catalyst showed a negative shift relative to the Run/NC catalyst and thus an enhancement of its mean Ru electron density, 26,27 which possibly derived from Ce electron donation as the only difference of Ru between the Ce1-Run/NC catalyst and the Run/NC catalyst was the presence of Ce single atoms in the Ce1-Run/NC catalyst.To check this assumption, the Bader charge analysis was further conducted.As suggested in Figure S13, the Ce electrons could be facilely transferred to Ru with a net electron transfer number of 0.05, therefore enhancing the mean electron density of Ru in the Ce1-Run/NC catalyst relative to that of the Run/NC catalyst.The mean electron density enhancement of Ru in the Ce1-Run/NC catalyst as calculated by the Bader charge analysis was in good agreement with corresponding experimental data acquired by the XANES measurement."

Comment 3:
The authors claim that they have Ru clusters in their main compounds.In the DFT model, these clusters comprise more than 12 Ru atoms.This contradicts the EXAFS fitting, where Ru has, on average, only two neighbours.In a cluster of 12 or more, this number is tremendously larger.Thus, the majority of Ru is not present in the form of the assumed clusters.Response: Thanks for the Reviewer's insightful comment.By virtue of careful microscopic observation, uniform Run nanoclusters with an average particle size of about 1 nm were identified in the Run/NC catalysts.To this end, we have built a Ru13 nanocluster with a diameter of about 1 nm for simulating the ultrafine Run nanoclusters based on their similar particle size.As pointed by the Reviewer, the Ru-Ru coordination number of the Ru13 nanocluster model (calculated to be 6.5, Figure R13) was obviously larger than that obtained by the EXAFS fitted result (with a mean Ru-Ru coordination number of 2).Regarding this problem, we have made detailed discussions in response to Comment 2 and demonstrated that the coexistence of Ru single atoms in the Run/NC catalyst was the reason for its low Ru-Ru coordination numbers.Therefore, to better simulate the Run/NC catalyst, we have reconstructed the structure model by considering both the particle size of the Run nanocluster (1 nm) and the presence of Ru single atoms as will be discussed in the following part.
We agree with the Reviewer that due to the low Ru-Ru coordination numbers, the relative content of Ru single atoms would be higher than that of the small Run nanoclusters.To obtain the relative ratio of the Ru single atoms to the Run nanoclusters, we have performed the following calculations.Firstly, we defined the number of the Ru single atoms and the Run nanoclusters (n=13 in our case) of the Run/NC catalyst as x and y, respectively.Therefore, the total Ru atom numbers of the Run/NC catalyst would be (x + 13y).As mentioned above, the Ru-Ru coordination number of the Ru13 nanocluster was 6.5 while the value was 0 for Ru single atoms (with only Ru-N coordination), which meant that the total Ru-Ru coordination numbers of the Run/NC catalyst were (x*0 + 13y*6.5).Therefore, we could acquire the average Ru-Ru coordination number of the Run/NC catalyst through dividing its total Ru-Ru coordination numbers by its total Ru atom numbers as illustrated in equation 1 below.As unveiled by the aforementioned EXAFS fitting results, the average Ru-Ru coordination number (NA) of the Run/NC catalyst was 2. For simplification, we have chosen one Run nanocluster in the Run/NC catalyst as the studying object, corresponding to a y value of 1.As such, the relative ratio of the Ru single atoms to the Run nanoclusters in the Run/NC catalyst was calculated to be 29:1 according to the equation 1 (based on NA=2 and y=1).However, in spite of the large number of Ru single atoms in the Run/NC catalyst, these isolated Ru atoms were primarily insufficient for dissociating water molecule (the ratedetermining step of alkaline HER).As demonstrated in Figure R15, the Gibbs free energy barrier for water dissociation over the Ru single atom was as huge as 1.03 eV.In addition, water dissociation over the Ru single atom was a thermodynamically unfavorable endothermic process with an endothermic energy of 0.67 eV.These results suggested the low reactivity of the Ru single atoms for the alkaline HER.To further confirm this, we have synthesized a pure Ru single atom catalyst (Ru1/NC, the synthesis details were provided in the methods part of the main text) and the formation of Ru single atoms on the NC support was verified by careful aberration-corrected HAADF-STEM observation (Figure R16).Corresponding alkaline HER evaluation results (LSV curve) in Figure R17 showed that the catalytic activity of the Ru1/NC catalyst was much lower than that of the Run/NC catalyst, which further revealed the low alkaline HER activity of the Ru single atoms.Actually, a common practice by previous reports (Adv. Funct. Mater., 2023, 33, 2213058;Appl. Catal., B, 2022, 312, 121378;Small, 2021, 17, 2101163) for simulating the catalyst with coexisted Ru single atoms and Ru nanoclusters/nanoparticles was to build a dual Ru1-Run structure model.Therefore, we have further constructed a dual Ru1-Ru13 structure model as shown in Figure R18 to simulate our Run/NC catalyst by considering both the particle size of the Run nanocluster (about 1 nm, corresponding to n=13) and its neighboring Ru single atoms.The Ru-N coordination numbers for the Ru single atom and for the Ru13 nanocluster were both four according to the EXAFS fitting results.As indicated in Figure R19, the dual Ru1-Ru13 site showed a quite low Gibbs free energy barrier (0.49 eV) for water dissociation and meanwhile the process was exothermic by 1.42 eV, showing a favorable reaction thermodynamics.These results revealed the good alkaline hydrogen evolution activity of the dual Ru1-Ru13 site and made it a reasonable structural model for simulating the Run/NC catalyst.
We have also added these contents to the revised manuscript as read in Line 21, Page 11: "In view of the copresence of Ru single atoms and Run nanoclusters in the Run/NC catalyst, we have built a dual Ru1-Ru13 model for simulating the Run/NC catalyst by considering both the particle size of the Run nanocluster (1 nm, Figure S11) and its neighboring Ru single atoms as was also commonly employed by previous reports for simulating the united catalyst with coexisted Ru single atoms and Ru nanoclusters/nanoparticles. 23,36,37All the structure details of the constructed models were provided in supporting information (Figure S31-34).To begin with, it should be noted that pure Ru single atoms on the NC support were insufficient for dissociating water molecule that was regarded as the rate-determining step of alkaline HER.As demonstrated in Figure S35, the Gibbs free energy barrier for water dissociation over the Ru single atom was as huge as 1.03 eV, indicating its low alkaline HER activity.To further examine the alkaline HER activity of Ru single atoms, we have synthesized a Ru1/NC control (the synthetic details were demonstrated in methods part).The formation of Ru single atoms in the Ru1/NC catalyst was confirmed by the aberration-corrected HAADF-STEM images of it in Figure S36.As further revealed by the alkaline HER evaluation results (LSV curve) in Figure S37, the catalytic activity of the Ru1/NC catalyst was much lower than that of the Run/NC catalyst.By contrast, the dual Ru1-Ru13 sites presented a quite low Gibbs free energy barrier for dissociating water molecules compared with that of the pure Ru single atoms (Figure S1 and Figure S35), making the dual Ru1-Ru13 site a reasonable structural model for simulating the Run/NC catalyst."

Comment 4:
The theoretical calculations do not align with the experimental observations.1.The clusters are too large compared to the EXAFS coordination number.Figure 5e implies an overpotential of almost 400 mV, which does not fit to the experimentally observed one at all.Therefore, the DFT analysis is without any meaning for the real system.Response: We thank the Reviewer for these important comments.As pointed by the Reviewer, the clusters are too large compared to the EXAFS coordination number.For this question, we have provided detailed explanation in response to Comment 2 and Comment 3 and unveiled that the copresence of Ru single atoms was the reason for the low Ru-Ru coordination numbers of the Run/NC catalyst.To better simulate the Run/NC catalyst, we have reconstructed the model for the Run/NC catalyst by taking into account of both the particle size of the Run nanoclusters (about 1 nm) and the coexistence of Ru single atoms.The rationale of the reconstructed Ru1-Ru13 model for simulating the Run/NC catalyst was also detailedly discussed in the response to Comment 3.   4 demonstrated that the alkaline HER activity of the Ce1-Run/NC catalyst was much higher than the Run/NC catalyst, the dual Ce1-Ru13 sites might be much more reactive than the dual Ru1-Ru13 sites because it was the sole difference between the Ce1-Run/NC catalyst and the Run/NC catalyst.
As indicated in Figure R21, water dissociation over the dual Ce1-Ru13 sites was highly exothermic by 4.5 eV, which was 3.2 times that of the dual Ru1-Ru13 sites (1.42 eV).This result unveiled that the dual Ce1-Ru13 sites were much more thermodynamically favorable to dissociate water molecules (rate-determining step of alkaline HER) than the dual Ru1-Ru13 sites.On the other hand, the water dissociation energy barrier over the dual Ce1-Ru13 sites (0.1 eV) was significantly lower than that over the dual Ru1-Ru13 sites (0.49 eV).These results disclosed the excellent catalytic reactivity of the dual Ce1-Ru13 sites for alkaline HER compared with the dual Ru1-Ru13 sites, and also suggested that it was reasonable to use the dual Ce1-Ru13 structural model (Figure R22) to simulate the highly efficient Ce1-Run/NC catalyst.All of these contents are now added to the revised manuscript as read in Line 19, Page 12: "As far as the Ce1-Run/NC catalyst was concerned, both the Ru single atoms and Ce single atoms in it were insufficient for dissociating water in terms of their huge Gibbs free energy barriers for water dissociation as demonstrated in Figure S35 and Figure S38, respectively.Keeping in mind that the experimental test results in Figure 4 demonstrated that the alkaline HER activity of the Ce1-Run/NC catalyst was much higher than the Run/NC catalyst, it was thus reasonable to use the dual Ce1-Ru13 model for simulating the Ce1-Run/NC catalyst because it was the only difference between the Ce1-Run/NC catalyst and the Run/NC catalyst."Line 1, Page 13, "The DFT calculations suggested that the exothermic energy of the dual Ce1-Ru13 sites (4.5 eV) was 3.2 times that of the dual Ru1-Ru13 sites (1.42 eV) in water activation process, which unveiled that the dual Ce1-Ru13 sites were more thermodynamically favorable to dissociate water molecules than the dual Ru1-Ru13 sites.In addition, the water dissociation energy barrier of the dual Ce1-Ru13 sites (0.1 eV) was also markedly lower than that of the dual Ru1-Ru13 sites (0.49 eV).Therefore, the dual Ce1-Ru13 sites catalyst were both more thermodynamically and kinetically beneficial to promote water dissociation relative to the dual Ru1-Ru13 sites."The Reviewer also mentioned the difference between the overpotentials obtained through the theoretical calculations and the experiment measurements.Concerning this problem, we would like to make a detailed explanation.In current work, the H adsorption energy as shown in Figure 5e was calculated by the "computational hydrogen electrode model" that was proposed by J. K. Nørskov, et al. in 2005(J. Electrochem. Soc. 2005, 152 (3), J23).By virtue of its robustness and relatively easy operation, the "computational hydrogen electrode model" was the mostly used method for calculating H adsorption energy of electrocatalyst at present.However, this method was greatly affected by the polarization effect during the calculations.As a result, though the calculated reaction energy reflected the reasonable reaction trend, the value of the calculated reaction energy commonly displayed an evident deviation from the experiment test result.To overcome this disadvantage, in the revised manuscript, we have further employed a new "CANDEL implicit solvation model" method that could mediate the polarization effect to calculate the H adsorption energies.During the calculations, the structures during reaction were fully relaxed until the final force on each atom was less than 0.01 eV Å -1 .The H adsorption energies on the structure models of the dual Ce1-Ru13 site and the dual Ru1-Ru13 site were obtained by adding the vibrational contribution of H to the electronic energy of corresponding reaction systems.As displayed in Figure R23, the calculated H adsorption energies of the dual Ce1-Ru13 site and the dual Ru1-Ru13 site via the "CANDEL implicit solvation model" method were -0.096 eV and -0.142 eV, respectively, which were both closer to the experimentally measured onset potentials of them than that calculated through the "computational hydrogen electrode model" method.We have also added these contents to the revised manuscript as read in Line 24, Page 13: "In another respect, the calculated H adsorption energies of the dual Ce1-Ru13 site and the dual Ru1-Ru13 site via the "computational hydrogen electrode model" method 39  were obviously larger than that of the experimentally measured onset potentials of them.This result was due to the fact that the reaction system was a grand canonical ensemble in the computational hydrogen electrode model, which was significantly affected by the polarization effect.As a consequence, despite that the calculated reaction energy suggested the reasonable reaction trend, the value of the calculated reaction energy commonly displayed a marked deviation from the experiment test result.Therefore, we have further performed the H adsorption energy calculations for the dual Ce1-Ru13 site and the dual Ru1-Ru13 site taking the "CANDEL implicit solvation model" method.As indicated in Figure S39, the calculated H adsorption energies on the dual Ce1-Ru13 site and on the dual Ru1-Ru13 site using this method were -0.096 eV and -0.142 eV, respectively, which were both closer to the experimentally measured onset potentials of them than that calculated through the "computational hydrogen electrode model" method."Comment 5: The plot of the long-term stability measurement (Figure 4f) is missing ticks on the right y-axis to make it easy to read how much the sample deactivates.Using a ruler, I could find out that there is a substantial deactivation in the range of 50-100 mV.If the authors want to know if their material would be stable under industrial conditions, they must measure it at 80 °C and a current density of 400 mA/cm².Response: Motivated by the Reviewer's kind suggestion, we have conducted the stability measurements at 80 °C and a current density of 400 mA/cm 2 on an alkaline anion-exchangemembrane water electrolysis (AEMWE) device using the Ce1-Run/NC as the cathodic catalyst and the nickel foam as the anodic catalyst.The obtained chronopotentiometry curve was presented in Figure R24 and we have added ticks on the right y-axis to make it easy to read how much the sample deactivates following the Reviewer's kind suggestion.It was found that the Ce1-Run/NC catalyst was quite stable during the stability test under the AEMWE conditions for 100 hours without obvious deactivation.We have also included these contents in the revised manuscript as read in Line 18, Page 10: "To examine the practical application potential of the Ce1-Run/NC catalyst, we have further measured the durability of the Ce1-Run/NC catalyst on an alkaline anion-exchange-membrane water electrolysis (AEMWE) device using the Ce1-Run/NC as the cathodic catalyst and the nickel foam as the anodic catalyst.The chronopotentiometry curve acquired by setting the reaction temperature at 80 °C and the current density of 400 mA/cm 2 in Figure 4f suggested that the Ce1-Run/NC catalyst displayed quite good stability for 100 hours testing."

Comment 6:
The general statement that the industry would prefer alkaline electrolysis is not true and is an oversimplification.It should be changed.Alkaline and PEM both have their advantages for industrial applications.Response: We thank the Reviewer for this very helpful comment and apologize for the misleading descriptions.To more clearly describe the contents as pointed by the Reviewer, we have rewritten the sentence in the revised manuscript as read in Line 26, Page 1: "In practical applications, alkaline water electrolysis and proton exchange membrane (PEM)-based water electrolysis in acid electrolyte both have their advantages for producing hydrogen as was well reviewed by previous reports. 3In a typical PEM system, a proton exchange membrane was used as solid electrolyte by means of which proton could be facilely transferred to cathode, 4 enabling fast hydrogen evolution kinetics.However, unlike the direct proton supply in acid HER, the proton was provided by the dissociation of H2O during the alkaline HER (Volmer step, equation 1). 5-7The substantial energy barrier for the cleavage of OH-H bond and sluggish supply of proton significantly impeded the reaction rate of alkaline HER. 8-10Even for the commercial Pt/C electrocatalysts, about two orders of magnitude reduction in HER activity was commonly identified when used in alkaline electrolyte. 11To this end, promoting electrocatalytic water dissociation over catalyst in alkaline electrolyte became of paramount importance to boost corresponding HER activity." Comment 7: Replacing Platinum by Ruthenium is not a real advantage as they are similarly rare in the earth's crust.Response: We do agree with the Reviewer that both platinum and ruthenium are of low contents in the earth's crust.In spite of that, the ruthenium metal has two distinct advantages over the platinum metal toward the alkaline HER: (i) the metal price of ruthenium (ca.15 $ g −1 , Sept 2023) was less than half that of platinum (ca.34 $ g −1 , Sept 2023), making the ruthenium a more attractive candidate for alkaline HER in view of cost.(ii) the ruthenium exhibited a much lower energy barrier for water dissociation relative to that of the platinum as displayed in Figure R25 (adapted from J. Am.Chem. Soc. 2016, 138, 49, 16174-16181).This in turn largely promoted the water dissociation (rate-determining step of alkaline HER) and thus improved the alkaline HER efficiency, which also potentially reduced the usage amount of Ru metal.We have added these contents to the revised manuscript as read in Line 4, Page 2: "In the past decades, Ru has shown great potential to substitute the Pt for alkaline HER by virtue of two-fold: (i) the metal price of Ru (ca.15 $ g −1 , Sept 2023) was less than half that of the Pt (ca.34 $ g −1 , Sept 2023); and (ii) the energy barrier for water dissociation over Ru was much lower than that over Pt, which largely promoted the water dissociation and thus improved the overall alkaline HER efficiency, also potentially reducing the usage amount of Ru. 12-14 " Figure R25.The Gibbs free energy diagram of HER on Ru and Pt, respectively (adapted from the J. Am.Chem.Soc. 2016, 138, 49, 16174-16181).
Comment 8: Tafel slopes cannot be reliably determined from potentiodynamic methods.They must be measured under steady-state conditions with various CA measurements (see 10.1021/acsenergylett.1c00608 or /10.1016/j.mtener.2022.101123).Response: We thank the Reviewer for this very important comment.As suggested by the Reviewer, we have remeasured the Tafel slopes of the catalysts in the revised manuscript using the potentiodynamic methods.It was demonstrated in Figure R26 (corresponding CA curves were also shown in Figure R27) that the remeasured Tafel slope values of the Ce1-Run/NC catalyst, the 20wt.%Pt/C catalyst, the Run/NC catalyst, and the Run-CeO2/NC catalyst were 41.5 mV dec -1 , 60.6 mV dec -1 , 134.4 mV dec -1 , and 180.7 mV dec -1 , respectively.It was evident that the Tafel slope value of the Ce1-Run/NC catalyst was the lowest among these catalysts, which suggested the fast reaction kinetics for hydrogen evolution over the Ce1-Run/NC catalyst.We have also updated these results in the revised manuscript as read in Line 26, Page 8: "We have also measured the Tafel slopes of these catalysts by virtue of the potentiodynamic method as suggested previously. 33,34It was demonstrated in Figure 4c that the obtained Tafel slope value of the Ce1-Run/NC catalyst (41.5 mV dec -1 ) was markedly lower than that of the 20wt.%Pt/C catalyst (60.6 mV dec -1 ), the Run/NC catalyst (134.4 mV dec -1 ), and the Run-CeO2/NC catalyst (180.7 mV dec -1 ), indicating fast reaction kinetics for hydrogen evolution over the Ce1-Run/NC catalyst."

Response to the Reviewers' comments:
Reviewer 2 Overall comments: The manuscript has revised based on the comments, providing the additional experimental and theoretical results.The data and analysis adequately test the hypothesis and support their novelty.But, the language and description aspects of scholarly presentation is still poor.The manuscript structure including the abstract and conclusion require further improvement to be published on high-level Nature Communications.Response: We are very grateful for the Reviewer's highly positive assessment of our revision.As suggested by the Reviewer, in the revised manuscript, we have carefully modified the manuscript structure and rewritten the abstract and conclusion.Please see below our point-bypoint responses to the Reviewer's concerns.

Comment 1:
For example, the HER should be written by the hydrogen evolution reaction, at first in the abstract part.Response: Thanks a lot for the Reviewer's kind suggestion.In the updated abstract part, we have revised the word "HER" to "hydrogen evolution reaction (HER)" to make it clear.
Comment 2: In addition, they illustrate the improved Ru atom efficiency by the strong oxophilicity including the abstract part.However, as shown in the results of ICP-OES, the Ce1-Run/NC shows the similar Ru loading to that of the Ru1-Run/NC, which further require adequate description.Response: Thanks for the Reviewer's valuable comment and kind advice.As suggested by the Reviewer, we have included more clear descriptions on the improvement of Ru atom efficiency of the Ce1-Run/NC catalyst relative to that of the Ru1-Run/NC catalyst.As read in in Line 10, Page 10, "Moreover, in spite of the similar Ru loadings between the Ce1-Run/NC catalyst and the Ru1-Run/NC catalyst, the mass activity normalized to per milligram of Ru of the Ce1-Run/NC catalyst was markedly higher than that of the Ru1-Run/NC catalyst as displayed in Figure S29, which unveiled the higher Ru atom efficiency for alkaline hydrogen evolution over the Ce1-Run/NC catalyst than that over the Ru1-Run/NC catalyst."Comment 3: Figure S28 shows the LSV curves and the TOF of catalysts, but the author illustrates the LSV curves and the mass activity curves at page 10 line 314.Response: Thanks for the Reviewer's kind reminding and we sincerely apologize for the typo.In the updated manuscript and supporting information, we have revised the "mass activity curves" to "TOF values" with reference to Figure S28 accordingly.

Comment 4:
In addition, they assume the strengthening of the Raman vibration peaks (1533 and 1390 cm -2 ) is related to OH effects during the alkaline HER.This assumption needs to be supported with the related references.Response: Thanks for the Reviewer's helpful suggestions.However, as we mentioned in the main text, after extensive paper searching, we found that there was no record of the two Raman vibration peaks of the Ce1-Run/NC catalyst at 1533 cm -1 and 1390 cm -1 .Therefore, we used the Gaussian fitting method to examine the two peaks' assignment as commonly employed previously.It was found that the Raman vibration peaks of the Ce1-Run/NC catalyst at 1533 cm - 1 and 1390 cm -1 were assigned to the Ce-N stretching vibrations in the Ce1-Run/NC catalyst.It was also identified that when introducing OH to the Ce-N sites of the Ce1-Run/NC catalyst, corresponding Raman peak intensities at the 1533 cm -1 and 1390 cm -1 were significantly strengthened, which was in good agreement with the in situ Raman spectra measurement results as displayed in Figure 4c.As such, both the Gaussian fitting results and the experiment data of the in situ Raman measurement revealed the OH strengthening effect on the Raman vibration peaks of the Ce1-Run/NC catalyst at the 1533 cm -1 and 1390 cm -1 .caption of figures were carefully checked and the typo and spacing were corrected accordingly.Moreover, we have also added the references (Ref 2 to Ref 116) previously incorporated in Table S4 and S5 to the supporting reference part to make it more readable.

Reviewer 3 Overall comments:
The authors have satisfactorily addressed most of my previous comments and thus this manuscript is recommended for publication.Response: We are very grateful for the Reviewer's highly positive evaluation of our revision.

Comment 1:
The authors should note that potentiodynamic means that the Tafel slopes are measured from LSV or CV and steady-state means that they are measured using CA or CP measurements, where the current has stabilized.Please check the usage of these terms in the manuscript again.Response: Thanks for the Reviewer's kind reminding and in the revised manuscript, we have carefully checked the use of these terms as mentioned by the Reviewer.To make it clear, we have rewritten the sentence in Line 24, Page 8 to "We have also measured the Tafel slopes of these catalysts under steady-state conditions using the CA measurements as suggested previously. 33,34"

Figure R1 .
Figure R1.(a) The Gibbs free energy diagrams for water dissociation over the Ru single atom (Ru1) and over the Run nanocluster (n=13), respectively.(b) The H adsorption energies on the Ru single atom and on the Run nanocluster, respectively.

Figure R2 .
Figure R2.The Gibbs free energy diagrams for water dissociation via the Ru1OH-RunH route and via the Ru1H-RunOH route, respectively.

Figure R3 .
Figure R3.The Raman spectra of the Ce1-Run/NC catalyst, Run/NC catalyst, NC support, and the XC-72 carbon, respectively.

Figure R4 .
Figure R4.(a) and (b) The HAADF-STEM images of the Ru1-Run/NC catalyst with varied magnifications.(c) and (d) The aberration-corrected HAADF-STEM images of the Ru1-Run/NC catalyst.The inset of (b) is the histogram of Run particle-size distribution of the Ru1-Run/NC.

Figure R5 .
Figure R5.(a) The LSV curves for the Ru1-Run/NC catalyst and the Ce1-Run/NC catalyst during the alkaline HER evaluations.(b) The TOF value curves of the Ru1-Run/NC catalyst and the Ce1-Run/NC catalyst during the alkaline HER evaluations.To gain theoretical insight into the catalytic activity difference between the Ce1-Run/NC catalyst and the Ru1-Run/NC catalyst, we have then conducted the first-principle DFT

Figure R6 .
Figure R6.(a) The Gibbs free energy diagrams for alkaline hydrogen evolution over the dual Ru1-Ru13 site and the dual Ce1-Ru13 site.(b) The H adsorption energies of the dual Ru1-Ru13 site and the dual Ce1-Ru13 site.The insets of (b) are the ELF pictures of *H on the dual Ce1-Ru13 site and on the dual Ru1-Ru13 site, respectively.

Comment 6 :
The authors argued that the EDS mapping images indicated the uniform dispersion of the Ru nanoclusters and the Ce single atoms in Ce1-Run/NC.However, the EDS mapping images of Ce1-Run/NC (Figure 1r-u) were measured at lower magnification compared to those of the Ce1/NC and Run/NC, which shows no clear distribution of Ru nanoclusters and the Ce single atoms.Response: Motivated by the Reviewer's kind suggestion, we have further performed the energy-dispersive X-ray spectrometry (EDS) elementary mapping analysis of the Ce1-Run/NC catalyst under high magnifications.As displayed in Figure R7a and e, the aberration-corrected high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) images of the Ce1-Run/NC catalyst disclosed uniformly dispersed single atoms and nanoclusters on the NC support.Corresponding EDS elementary mapping images of the Ce1-Run/NC catalyst in Figure R7b-d and Figure R7f-h suggested that the Ru element signal was concentrated on the small nanoclusters while the Ce element signal was highly overlapped with the distribution of single atoms.We have also modified the Figure R7a-d to the revised manuscript as displayed in the updated Figure 1r-u.

Comment 8 :
Figure R8.(a) Fourier transforms of the Ce LIII-edge EXAFS oscillations in the R space of the Ce1-Run/NC catalyst and the reference CeO2.(b) The wavelet transformation for the EXAFS signals of the Ce1-Run/NC catalyst and the reference CeO2.
mass =   ×   I: Measured currents, A; mcat: Mass of catalyst on electrode, g; : Metal loading of catalyst based on the ICP-OES measurements;

Figure R10 .
Figure R10.(a) and (b) The HAADF-STEM images of the post-reaction Ce1-Run/NC catalyst.The inset of (b) is corresponding histogram of particle-size distribution of (b).(c) and (d): The aberration-corrected HAADF-STEM images of the post-reaction Ce1-Run/NC catalyst.(e)-(g) Corresponding energy-dispersive X-ray spectrometry (EDS) elementary mapping images of (d).

Figure R11 .
Figure R11.The stability test of the Ce1-Run/NC catalyst by virtue of the continuous cyclic voltammogram measurements conducted in 1.0 M KOH electrolyte.

Figure R12 .
Figure R12.Reaction energy diagrams of alkaline HER over the dual Ce1-Ru13 site and over the dual Ru1-Ru13 site.(a) Surface electrostatic potential distribution of the dual Ce1-Ru13 site.(b) The Gibbs free energy diagrams for complete hydrogen evolution reaction over the dual Ce1-Ru13 site and over the dual Ru1-Ru13 site.(c,d) Partial density of state (PDOS) of Ru13 nanoclusters in the dual Ce1-Ru13 site and in the dual Ru1-Ru13 site, respectively.(e) The H adsorption energies on the dual Ce1-Ru13 site and on the dual Ru1-Ru13 site, respectively.The insets of (e) are the ELF pictures of *H on the dual Ce1-Ru13 site and on the dual Ru1-Ru13 site, respectively.(f) The schematic illustrations for each step of the alkaline HER process over the dual Ce1-Ru13 site and over the dual Ru1-Ru13 site, respectively.The *H2O indicated the adsorbed water molecules on the catalyst.The *H+*OH indicated the adsorbed H and the adsorbed OH on the catalyst, respectively.The *H-OH represented for that the H was still adsorbed on the catalyst while the OH was undergoing the process of desorption.

Figure R13 .
Figure R13.(a) The simulated Ru13 nanocluster (corresponding to a Run size of 1 nm).(b) The Ru-Ru coordination numbers of the corner and center Ru atoms in the Ru13 nanocluster.

Figure R14 .
Figure R14.(a)-(d) The aberration-corrected HAADF-STEM images of the Run/NC catalyst.It was identified that Ru single atoms and small Run nanoclusters coexisted on the NC support in the Run/NC catalyst.
NA represents for the average Ru-Ru coordination number in the Run/NC catalyst;x denotes the number of the Ru single atoms in the Run/NC catalyst; y indicates the number of the Run nanoclusters in the Run/NC catalyst;

Figure R15 .
Figure R15.The Gibbs free energy diagrams for water dissociation over the Ru single atom catalyst (Ru1/NC).

Figure R17 .
Figure R17.The LSV curves for the Ru1/NC catalyst and the Run/NC catalyst during the alkaline HER evaluations.The Ru loading amounts of the Ru1/NC catalyst and the Run/NC catalyst were 0.2wt.%and 1.2wt.%,respectively.

Figure R18 .
Figure R18.(a) Side view and (b) top view of the dual Ru1-Ru13 structural model.

Figure R19 .
Figure R19.The Gibbs free energy diagrams for water dissociation over the dual Ru1-Ru13 sites.The insets of the picture are corresponding schematic illustrations for each step.

Figure R20 .
Figure R20.The Gibbs free energy diagrams for water dissociation over the Ce1/NC catalyst.The insets of the picture are corresponding schematic illustrations for each step.

Figure
Figure R20 (0.8 eV for Ce1), we have further examined the water dissociation properties of the dual Ru1-Ru13 sites and the dual Ce1-Ru13 sites in the Ce1-Run/NC catalyst by the DFT calculations.Keeping in mind that the experimental test results in Figure4demonstrated that the alkaline HER activity of the Ce1-Run/NC catalyst was much higher than the Run/NC catalyst, the dual Ce1-Ru13 sites might be much more reactive than the dual Ru1-Ru13 sites because it was the sole difference between the Ce1-Run/NC catalyst and the Run/NC catalyst.As indicated in FigureR21, water dissociation over the dual Ce1-Ru13 sites was highly exothermic by 4.5 eV, which was 3.2 times that of the dual Ru1-Ru13 sites(1.42 eV).This result unveiled that the dual Ce1-Ru13 sites were much more thermodynamically favorable to dissociate water molecules (rate-determining step of alkaline HER) than the dual Ru1-Ru13 sites.On the other hand, the water dissociation energy barrier over the dual Ce1-Ru13 sites (0.1 eV) was significantly lower than that over the dual Ru1-Ru13 sites (0.49 eV).These results disclosed the excellent catalytic reactivity of the dual Ce1-Ru13 sites for alkaline HER compared with the dual Ru1-Ru13 sites, and also suggested that it was reasonable to use the dual Ce1-Ru13 structural model (FigureR22) to simulate the highly efficient Ce1-Run/NC catalyst.All of these contents are now added to the revised manuscript as read in Line 19, Page 12: "As far as the Ce1-Run/NC catalyst was concerned, both the Ru single atoms and Ce single atoms in it were insufficient for dissociating water in terms of their huge Gibbs free energy barriers for water dissociation as demonstrated in FigureS35and FigureS38, respectively.Keeping in mind that the experimental test results in Figure4demonstrated that the alkaline HER activity of the Ce1-Run/NC catalyst was much higher than the Run/NC catalyst, it was thus reasonable to use the dual Ce1-Ru13 model for simulating the Ce1-Run/NC catalyst because it was the only difference between the Ce1-Run/NC catalyst and the Run/NC catalyst."Line 1, Page 13, "The DFT calculations suggested that the exothermic energy of the dual Ce1-Ru13 sites (4.5 eV) was 3.2 times that of the dual Ru1-Ru13 sites (1.42 eV) in water activation process, which unveiled that the dual Ce1-Ru13 sites were more thermodynamically favorable to dissociate water molecules than the dual Ru1-Ru13 sites.In addition, the water dissociation energy barrier of the dual Ce1-Ru13 sites (0.1 eV) was also markedly lower than that of the dual Ru1-Ru13 sites (0.49 eV).Therefore, the dual Ce1-Ru13 sites catalyst were both more thermodynamically and kinetically beneficial to promote water dissociation relative to the dual Ru1-Ru13 sites."

Figure R21 .
Figure R21.The Gibbs free energy diagrams for water dissociation over the dual Ce1-Ru13 site and over the dual Ru1-Ru13 site.Corresponding schematic illustrations for each step over the

Figure R22 .
Figure R22.(a) Side view and (b) top view of the dual Ce1-Ru13 structural model.

Figure R23 .
Figure R23.The H adsorption energies of the dual Ce1-Ru13 site and the dual Ru1-Ru13 site calculated by the CANDEL implicit solvation model method.

Figure R24 .
Figure R24.The chronopotentiometry curve of the Ce1-Run/NC catalyst under AEMWE conditions at a reaction temperature of 80 °C and a current density of 400 mA/cm 2 .The inset of Figure R24 was the photograph of the AEMWE device.

Figure R26 .
Figure R26.The Tafel slope values of the Ce1-Run/NC catalyst, the 20wt.%Pt/C catalyst, the Run/NC catalyst, and the Run-CeO2/NC catalyst, measured by the potentiodynamic method.

Figure R27 .
Figure R27.Corresponding CA curves for the Ce1-Run/NC catalyst, the 20wt.%Pt/C catalyst, the Run/NC catalyst, and the Run-CeO2/NC catalyst, during the calculation of the Tafel slopes of them by the potentiodynamic method.

Table R2 .
The loading amount of the Ru or Pt of the tested catalyst obtained by the ICP-OES measurement.

Table R3 .
The mass activity comparison between the Ce1-Run/NC catalyst and previously reported Ru-based alkaline HER catalysts at -0.05 V vs. RHE in 1M KOH electrolyte.
Comment 2: Is most of the Ru really in the form of metallic clusters?The XANES edge is way too high for metallic Ru.I cannot see how the XANES data would agree with the following statement: "It was demonstrated in FigureS8that electrons could be facilely transferred from Ce to Ru with a net electron transfer number of 0.05, which agreed well with the XANES data.".

Table R5 .
The best-fitted EXAFS results of the Run/NC and the Ce1-Run/NC catalysts.