Partially sintered copper‒ceria as excellent catalyst for the high-temperature reverse water gas shift reaction

For high-temperature catalytic reaction, it is of significant importance and challenge to construct stable active sites in catalysts. Herein, we report the construction of sufficient and stable copper clusters in the copper‒ceria catalyst with high Cu loading (15 wt.%) for the high-temperature reverse water gas shift (RWGS) reaction. Under very harsh working conditions, the ceria nanorods suffered a partial sintering, on which the 2D and 3D copper clusters were formed. This partially sintered catalyst exhibits unmatched activity and excellent durability at high temperature. The interaction between the copper and ceria ensures the copper clusters stably anchored on the surface of ceria. Abundant in situ generated and consumed surface oxygen vacancies form synergistic effect with adjacent copper clusters to promote the reaction process. This work investigates the structure-function relation of the catalyst with sintered and inhomogeneous structure and explores the potential application of the sintered catalyst in C1 chemistry.

CeO<sub>2</sub>? What is the state of the signal for the H<sub>2</sub>-treated catalyst before the reaction? 4-It is not clear why there are less oxygen vacancies created when testing at 500 °C as opposed to 300 °C (Figure 4a). 5-In the last paragraph on page 14, the authors indicate that the lower apparent reaction order of CO<sub>2</sub> and H<sub>2</sub> on 15CuCe compared to 5CuCe shows the higher ability of this catalyst in adsorption and activation of CO<sub>2</sub> and H<sub>2</sub>. This conclusion is not clear to me. What we can understand from the lower activation of H<sub>2</sub> and CO<sub>2</sub> on 15CuCe is that the reaction rate on this catalyst is less dependent on the concentration of CO<sub>2</sub> and H<sub>2</sub>. This does not lead to the conclusion that CO<sub>2</sub> and H<sub>2</sub> can better be adsorbed and activated for the RWGS reaction. Stronger adsorption of CO<sub>2</sub> or H<sub>2</sub> does not necessarily lead to their higher activity or higher level of participation in the reaction. The authors need to clarify their statements and/or add supporting arguments.
<i>Suggested Improvements:</i> Most suggested improvements are already listed above. However, some additional suggestions are shown below: 1-another test with a catalyst with Cu loading between 15 and 25 might be needed to see if 15% Cu is the optimum loading. Figure 4a and Figure S14, the insets should be explained.

2-In
3-The text need full revision. There are grammatical errors ("strong interacted… structure", "the reductive support with oxygen vacancies was easier to activate CO<sub>2</sub>", first three lines of page 12, "the D1 band was even obvious than F2g peak during…", "have no influence to the intrinsic defects", etc.) as well as typos (Figure 3 caption, etc.) which need to be fixed. <i>Clarity and Context:</i> The manuscript is written clearly and it is understandable for the scientific community. Other than some grammatical errors and typos which were referred to in the previous comments, the text, in general, is well written. For above existing scientific problems, we designed relevant experiments and carried out related researches. We think that it can meet the scope of Nature Communications. We believe our findings will interest a broad readership in Catalysis and Materials Science according to the following aspects: 1. We precisely determined the structure of the copper-ceria catalyst for the

RWGS reaction under harsh reaction conditions.
The structure of the copper-ceria catalyst treated at high temperature (600 ºC) and with reductive atmosphere was clearly revealed. Under such a harsh reaction condition, the CeO2 supports were only partially sintered, but not completely sintered. With comprehensive characterizations, we found that the interaction between copper and ceria maintained well after long-term catalytic tests (for 240 hours), which ensured the copper species with high loading (~15 wt. %) were still very stable on the partially sintered support in the forms of 2D and 3D clusters.

We directly explored the role of the oxygen vacancies in the RWGS reaction.
In situ Raman experiments were designed and carried out to investigate the role of oxygen vacancies. Based on the test results, we found that the surface oxygen vacancies could be created by H2 and consumed by CO2. The in situ generated and consumed surface oxygen vacancies in the copper-ceria catalyst were directly confirmed to be involved in the RWGS reaction.

We clearly revealed the synergistic effect between copper cluster and oxygen vacancy.
Based on the experimental results and DFT calculations, we found that the surface oxygen vacancy had synergistic effect with the adjacent copper cluster, improving the activation of CO2 and the formation of active intermediates. The stable copper cluster-oxygen vacancy synergistic sites ensured the copper-ceria catalyst had unmatched activity to catalyze the RWGS reaction, which surpassing almost all other reported metal catalysts. Besides, this copper-ceria catalyst also maintained excellent stability in the 240 h activity evaluation under very harsh reaction conditions (600 ºC, space velocity of 400,000 mL·gcat -1 ·h -1 ).
After all, in this work, the efficient copper-ceria catalyst with both high activity and stability under very harsh reaction conditions has been constructed. And the struction-function relation of the partially sintered copper-ceria catalyst was clearly revealed. We think these findings can provide a valuable reference for the exploration and application of sintered catalyst in the high temperature reaction.
Where is the equilibrium constant of the RWGS reaction at 600 ºC, and α is the CO2 equilibrium conversion rate at 600 ºC. The relevant supplements have been added in the revised manuscript on page 20, line 2627 (highlighted in yellow).
Thanks for the reviewer's comment again. According to the reviewer's comments, the catalytic performances of the 15CuCe catalyst under other reaction atmospheres with H2:CO2 ratios of 2:1 and 4:1 were also evaluated. As shown in Figure R1, the CO2 conversion increased with the increasing of the H2:CO2 inlet ratio at various reaction temperatures. And at all these three H2:CO2 ratios, the 15CuCe catalyst shown excellent catalytic performances, suggesting this catalyst had high catalytic efficiency over a wide range of H2:CO2 ratios. And it was noteworthy that even when the H2:CO2 ratio reached 4:1, no CH4 was detected in the product, which indicated the catalyst catalyzed the RWGS reaction much rather than the methanation reaction. The corresponding description has been added in the revised manuscript on page 5, line 18-20 and page 6, line 1-6. And the data was shown in supporting information as Figure S2 on page S7 (highlighted in yellow).  To explore the durability of the 15CuCe catalyst further, a stability test was measured for 240 hours. As for the test result under the harsh condition (GHSV = 400,000 mLgcat 1 h 1 , 600 ºC) in Figure R2, it maintained more than 85% of its initial activity after 240 h. During the first 20 hours of the stability test, the catalyst experienced slight deactivation, but the activity remained stable in the subsequent test. The result strongly confirmed the excellent stability of the 15CuCe catalyst in the high-temperature RWGS reaction. The corresponding data has been added as Figure 1d in the revised manuscript on page 5 (highlighted in yellow).
Besides, the HAADF-STEM images of the used 15CuCe catalyst after the 240 h stability test ( Figure R3) exhibited that CeO2 nanorods were still partially sintered and the 2D and 3D copper clusters were still anchored on the partially sintered CeO2 nanorods, which undoubtedly prevented the catalyst from deactivation. The corresponding data has been added as Figure 3 in the revised manuscript on page 9 and the corresponding discussion has been added in the revised manuscript on page 9, line 8-13 and page 10, line 1-11 (highlighted in yellow).
Thanks for the reviewer's comment again.

Comment 4:
What about the recyclability of this catalyst. This sample suffer a remarkable ceria agglomeration/collapse.

Response:
Thanks for the reviewer's valuable comments. The ceria nanorod support of the copper-ceria catalyst could suffer partial sintering during the high-temperature RWGS reaction. In order to explore the cyclic stability, the activity evaluation of six start-up cool-down cycles was carried out. And as illustrated in Figure R4, the reaction rate of the 15CuCe catalyst maintained stable in 6 rounds of activity tests, indicating the 15CuCe catalyst had very good stability under the start-up cool-down reaction conditions. It should be noted that the 15CuCe catalyst was treated by the RWGS reaction atmosphere at 600 ºC for one hour before the kinetic tests. The Figure R4 has been added as Figure S6 in the revised supporting information on page S11. And the corresponding description has been shown in page 7, line 10-12 in the revised manuscript (highlighted in yellow). Figure R4. Reaction rate over the 15CuCe catalyst for six start-up cool-down cycles.

Comment 5:
What about the reproducibility of this catalyst. Both from the point of view of the synthesis and the reaction.

Response:
The reviewer's valuable comment was appreciated. To ensure that the synthesis and catalytic performance of the 15CuCe catalyst were repeatable, another two batches of samples (sample 2 and sample 3) were synthesized by using the same method. As shown in Figure R5, the newly synthesized samples exhibited CO2 conversion rates which were the same to that of the previously synthesized catalyst (sample 1), indicating the synthesis and catalytic performance of the 15CuCe catalyst were very repeatable. Figure R5. Catalytic activities of the 15CuCe catalysts prepared in different batches.  Response: In order to know exactly the structure of the 15CuCe catalyst after H2 pretreatment, the HAADF-STEM and XPS measurements were conducted. As shown in Figure R6, ceria supports could maintain the rod-like morphology after pretreatment under H2 flow, which indicated ceria nanorods have not undergone obvious sintering prior to the RWGS reaction. In addition, 2D and 3D copper clusters could be clearly observed on the surface of CeO2 nanorods. The width of the 2D layered clusters ranged from 1 nm to 3.5 nm, which was slightly smaller than the copper clusters after long-term RWGS reaction. And the average width and thickness of the 3D hemisphere shaped clusters were similar to that of the copper clusters after stability test, indicating the excellent stability of the copper clusters. The As shown in Figure R7, the Cu 2p XPS spectra of the 15CuCe catalyst after H2 pretreatment suggested the Cu 2+ species was reduced to Cu + /Cu 0 species. Thus, the statement about that Cu 2+ was reduced to Cu + /Cu 0 species during the RWGS reaction was not appropriated. The Figure R7 has been added in the revised supporting information as Figure S16 on page S21, and the corresponding description has been shown in the revised manuscript on page 11, line 2-9 (highlighted in yellow).

Comment 6:
Thanks for the reviewer's comments again.  Response: Thanks for the reviewer's valuable comments. In order to explore the effect of ceria calcination on the metal-support interaction, we prepared the CuCe catalyst with ceria support which was calcined at 600 ºC for the Cu deposition. The obtained catalyst was denoted 15CuCe-600. As shown in Figure R8a, the TPR profile of the 15CuCe-600 catalyst could be deconvoluted into four peaks. The peaks denoted as α and β could be assigned to the progressive reduction of the dispersed CuOx species to Cu + /Cu 0 species. The γ peak was related to the Cu-[Ox]-Ce structures. And the θ peak could be attributed to the reduction of bulk CuO phase. Compared with the 15CuCe catalyst with ceria support which was not calcined previous to Cu deposition, the appearance of θ peak suggested that during the preparation of the catalyst, the change of the interaction between metal and support caused the increase of the size of 13 copper species. However, compared with the CeO2 support and CuO, the significant shift of reduction peaks to low temperature meant that even though the CeO2 support was calcined prior to loading copper, the interaction between copper and ceria in the prepared catalyst still maintained. Compared with the 15CuAl catalyst, the 15CuCe-600 catalyst still had much stronger metal-support interaction. Besides, the activity test result exhibited that the 15CuCe-600 catalyst had inferior activity than the 15CuCe catalyst ( Figure R8b). Therefore, the catalyst with better catalytic activity could be prepared by depositing copper on uncalcined ceria nanorods. Thanks for the reviewer's valuable comments again.   For above existing scientific problems, we designed relevant experiments and carried out related researches. We think that it can meet the scope of your journal, and is justified for the Nature Communications. We believe our findings will interest a broad readership in Catalysis and Materials Science according to the following aspects: 1. We precisely determined the structure of the copper-ceria catalyst for the

RWGS reaction under harsh reaction conditions.
The structure of the copper-ceria catalyst treated at high temperature (600 ºC) and with reductive atmosphere was clearly revealed. Under such a harsh reaction condition, the CeO2 supports were only partially sintered, but not completely sintered. With comprehensive characterizations, we found that the interaction between copper and ceria maintained well after long-term catalytic tests, which ensured the copper species with high loading (~15 wt. %) were still very stable on the partially sintered support in the forms of 2D and 3D clusters.

We directly explored the role of the oxygen vacancies in the RWGS reaction.
In situ Raman experiments were designed and carried out to investigate the role of oxygen vacancies. Based on the test results, we found that surface oxygen vacancies could be created by H2 and consumed by CO2. The in situ generated and consumed surface oxygen vacancies in the copper-ceria catalyst were directly confirmed to be involved in the RWGS reaction.

We clearly revealed the synergistic effect between copper cluster and oxygen vacancy.
Based on the experimental results and DFT calculations, we found that After all, in this work, the efficient copper-ceria with both high activity and stability under very harsh reaction conditions was constructed. And the struction-function relation of the partially sintered copper-ceria catalyst was clearly revealed. We think these findings can provide a reference for the exploration and application of sintered catalyst in the high temperature reaction. We are looking forward to your next comments. Even though the CeO2 support suffered severe sintering, the interaction between copper and ceria could maintained well after the long time treatment under high temperature and reductive atmosphere, which ensured the copper clusters stabilized on the partially sintered CeO2 nanorods.

Validity:
The data and results are valid and the data seem to be collected and interpreted in an acceptable way based on the description of the experiments.
Response: Thanks for the reviewer's valuable comments.

Significance:
The Response: The reviewer's valuable comment was highly appreciated by us. The significance of this work is as follows: 1. In this work, the partially sintered copper-ceria catalyst with sufficient and stable copper sites was constructed, which exhibited excellent catalytic performance under the very harsh conditions. The abundant and stable copper clusters with sufficient surface vacancies accounted for the extraordinary activity and stability. We found that the interaction between copper and ceria maintained well after long-term RWGS reaction, which ensured the copper species with high loading (~15 wt. %) stabilized on the partially sintered CeO2 nanorods in the forms of 2D and 3D clusters. According to the reviewer's suggestions, standard copper-ceria catalysts with the same composition were synthesized by using ceria nanocube and ceria nanoparticle as supports. Based on the characterization results and activity evaluation, the copper-ceria catalyst synthesized by ceria nanorod showed unique structural advantage and better catalytic performance ( Figure R9). On CeO2 nanorods, copper species were more dispersed and oxygen vacancies were easier to form, which made the copper-ceria catalyst with CeO2 nanorods as support have the best initial activity. Besides, the CeO2 nanorods had better structural stability (Table R2), which prevented the inactivation caused by the excessive sintering of the catalyst. In addition, the standard copper-ceria catalyst prepared by conventional impregnation (IMP) method had much inferior activity due to the poor dispersion of copper species ( Figure R11). Copper species were easier to be anchored on the ceria support by DP method rather than IMP method ( Figure   R11).   Figure S7c  (2) The specific BET surface areas (Table R2) suggested that the used 15CuCe-NR catalyst had a much larger specific surface area than that of the used 15CuCe-NP catalyst, which confirmed that the CeO2 nanorod had better structural stability than the CeO2 nanoparticle. In our previous report, it was proved that the CeO2 nanorod was more resistant to sintering than the CeO2 nanoparticle during the air-calcination at high temperature (C. J. Jia et al, J. Am.

The active sites in supported catalysts
Chem. Soc. 2019, 141, 17548-17557). Excessive sintering of the catalyst will lead to the deactivation. As shown in Figure R10, the stability evaluation of the 15CuCe-NP catalyst indicated that it lost 30% of its origin activity within 20 h. The Figure R10 has been shown in the revised supporting information as Figure S5 on page S10, and the responding description has been added in the revised manuscript on page 7, line 7-9 (highlighted in yellow). To sum up, ceria nanorod was a better choice to prepare copper-ceria catalyst with more efficient catalytic activity in the high-temperature RWGS reaction, when compared with the ceria with other morphologies (nanoparticle and nanocube). Thanks for the reviewer again.

The influence of preparation method:
To explore the importance of the preparation method, we synthesized the reference catalyst by typical impregnation (IMP) method. The reference catalyst was denoted 15CuCe-IMP. As shown in Figure R11a, the 15CuCe-IMP exhibited much lower RWGS conversion than that the 15CuCe prepared by DP method. The distinct diffraction peaks of CuO in the XRD result ( Figure R11b) indicated the worse dispersion of copper species for the fresh 15CuCe-IMP sample. And the TEM picture of the fresh 15CuCe-IMP sample ( Figure R11c) illustrated that the ceria nanorod suffered more severe sintering during the preparation of catalyst, causing the decrease of the specific surface area (60.3 m 2 g  ) ( Table R2). The interaction between copper and ceria was relatively weak, which could make neither the active metal, nor the ceria support stable. The copper species with poor dispersion also accelerated the sintering of ceria. Above data demonstrated that for the copper-ceria catalyst, the dispersion degree of copper species had an important effect on the activity and structural stability. From the perspective of catalyst synthesis method, copper was easier to be anchored on the ceria support by DP method rather than IMP method. The data has been added in the revised supporting information as Figure   S14 on page S19, and the corresponding description has been shown in the revised manuscript on page 10, line 14-16 (highlighted in yellow). Thanks for the reviewer again.

The size of the CeO2 crystals:
The reviewer mentioned that the size of CeO2 crystal reduced after reaction for 70 h.
Actually, the size of the CeO2 crystals increased after stability test. The weak diffraction peaks might be caused by using too little sample (˂10 mg) in the XRD measurement of the used 15CuCe catalyst. In order to get better data, more sample after 70 h of RWGS reaction was used to repeat the XRD test. As shown in figure R12,

Response:
We thank for the reviewer's comment. In order to detect other products that might be existed in addition to CH4 and CO, the products and reactants in the RWGS reaction catalyzed by the 15CuCe catalyst were further detected online by using two tandem gas chromatography equipped with a flame ionization detector (FID) and a thermal conductivity detector (TCD). In the catalytic performance evaluation, no other products were detected besides CO, which again confirmed the CO selectivity was 100%. Besides, the carbon balance was found to be greater than 97% in all the tests. Carbon balance below 100% might be due to the carbon deposition.
From the Raman results in Figure R13a, there were no obvious peaks of carbon deposition over the used 15CuCe catalyst after 20 h RWGS reaction at 600 ºC. And as shown in Figure R13b, CO2 desorption peaks appeared in the O2-TPO and Ar-TPD results, which suggesting the CO2 signal in the O2-TPO result might come from the CO2 desorption rather than carbon deposition. Thus, the carbon deposition on the catalyst surface was not severe during the RWGS reaction. Thanks for the reviewer again.   Figure R14. In Figure R14a    Response: The reviewer's comment is highly appreciated by us. As shown in Figure   R14a, when the sample was exposed to the CO2 gas flow, even with the formation of carbonate species, no gaseous signal of CO was detected, which meant carbonate species might not be converted directly to CO. According to the reviewer's suggestion, the relevant DFT study was performed to explore the reaction mechanism further. As shown in Figure R15, CO2 was more easily adsorbed at VO-A.
In the presence of H2, Cu atoms captured H2 molecule and broken H-H bond, then transferred H atom to CO2, shown as Figure Table R3 has been added in the revised supporting information as Table S2 on page S33 (highlighted in yellow). Thanks for the reviewer again. Figure R16. RWGS reaction mechanism occurred in the VO-A. The red, blue and black lines indicated different reaction paths, and the structural diagrams with the red, blue or black stroke corresponded to the reaction paths, respectively.

Comment 2:
There are more than two peaks in the H2-TPR profiles of the catalysts.
These peaks need to be identified and described.   From the XRD results of the fresh and used 15CuCe catalysts in Figure R19, the XRD peaks of CeO2 did not shift to higher angle, indicating the Cu species were supported on CeO2 support rather than incorporated into the CeO2 lattice. In addition, the HAADF images also suggested copper species were supported on the surface of ceria.    (highlighted in yellow). Thanks for the reviewer again.

Suggested Improvements:
Most suggested improvements are already listed above. However, some additional suggestions are shown below:

Comment 1: Another test with a catalyst with Cu loading between 15 and 25 might be needed to see if 15% Cu is the optimum loading.
Response: In order to explore whether 15% was the optimal loading, the activity tests of the 18CuCe and 22CuCe catalysts were measured. As shown in Figure R22, the 18CuCe and 22CuCe catalysts showed lower activity than the 15CuCe catalyst.
Therefore, based on existing experimental results, 15% Cu was the optimum loading.
Thanks for the reviewer again. Figure R22. Activities of the 15CuCe, 18CuCe and 22CuCe catalysts.

Comment 2:
In Figure 4a and Figure S14, the insets should be explained.

<b>REVIEWER COMMENTS</B>
Reviewer #1 (Remarks to the Author): Authors have made a significant effort to address most of the points indicated by the referees. Hence the paper has been remarkably improved.
From this reviewer´s perspective, I am satisfied with the responses for all my comments except for the first one related to equilibrium data. Authors are deliberately ignoring the competing process: CO2 methanation on the basis of the observed full selectivity to CO in the experiments. This is not convincing and it is in fact a wrong practice to assume that experimental data could give fundamental process equilibrium trends. Experimental conversion and equilibrium conversion must be decoupled and treated as separate part of the story. The fact that experimentally authors do not observe methane does not change at all the underlying equilibrium thermodynamics and hence the represented equilibrium plot is not correct. There are many references in literature including the correct equilibrium plot.
Despite the manuscript improvement, I stand by my initial assessment and I believe this paper is more suitable for a specialized journal in catalysis and not suitable for Nature Comm. It is does not level up the standards of the journal and it is not novel enough and does not have sufficient impact for Nature Comm. Actually, the paper provides incremental advances rather than innovative results.
Reviewer #3 (Remarks to the Author): The authors have carefully revised the manuscript according to the reviewer's comments. The have seized all suggestions and significantly improved the manuscript. Although the study is very comprehensive and explains the behavior and the structure of the catalyst with great detail, the studied catalyst suffers from somewhat limited technical relevance, since ceria nanorods are unlikely to be applied as support material in industrial processes. Nevertheless, the results are of sufficient novelty and detail to justify publication in Nature Communications.
I have the following comments that should be considered by the authors: In Figure R1, it is unclear to which axis the shown bars refer. In Figure R3 and R6, the differentiation between 2D and 3D clusters is not convincing. Due to the cyclindrical shape of the nanorods, 3D clusters distributed on the surface of the nanorods could appear as flat or twodimensional (2D) depending on the viewing angle. Please provide more convincing resuls and arguments for the claimed 2D clusters.

To Reviewer #1
Comment: Authors have made a significant effort to address most of the points indicated by the referees. Hence the paper has been remarkably improved.    12 . As shown in Figure R2, under mild treatment conditions, the structure of CeO2 had no change, and the copper species could be stabilized on the support in the form of dispersed clusters. Therefore, it was expected that copper clusters would remain stable on the ceria under mild treatment conditions. However, in our work, we explored the structural transformation of copper-ceria catalyst after pretreatment and long-term reaction. As shown in Figure   R3, after pretreatment by H2, the ceria nanorods were not sintered and the copper species with high copper loading (~15 wt.%) were dispersed as clusters on the support. When the catalyst underwent long-term RWGS reaction (for 70 and 240 h) under harsh conditions (600 ºC, GHSV=400,000 mLgcat 1 h 1 ), the highly dispersed copper clusters was still anchored on the partially sintered ceria nanorods, which has never been reported in previous research work. This finding clearly reveals the structure of partially sintered copper-ceria catalyst and illustrates the copper-ceria catalyst can be used not only in low-temperature reactions but also have application potential in high-temperature hydrogenation reactions.   13 . But as shown in Figure R4, under the operating temperature of 600 ºC and at a space velocity of only ~60,000 ml gcat −1 h −1 , the CO2 conversion of this catalyst was only 18%. In our work, the optimal 15CuCe catalyst showed the activity of 146.6 molCO2gcat -1 s -1 at 600 ºC, which far exceeded that of other catalysts reported in literatures ( Figure 4 and Table R1). Meanwhile, our copper-ceria catalyst also exhibited solid stability in the long-term test (for 240 h), which indicated that the partially sintered catalyst has indeed achieved a combination of excellent activity and high stability.  Table R1 for more details).

How does oxygen vacancy in the catalyst play a role in the reaction process?
Defects in solid catalysts have an important effect on their catalytic activity.
Although many previous reports have pointed out that oxygen vacancy was related to the catalytic activity of CO2 reduction, it is still unclear how the oxygen vacancy is The specific synergistic catalytic effect between copper cluster and oxygen vacancy also reflects that catalytic reactions are more likely to occur at the metal cluster-oxygen vacancy interfaces, which provides guidance for the design and development of efficient supported catalysts.  In situ Raman of the 15CuCe catalyst with H2/CO2 switching under 300 ºC and 500 ºC, respectively. (d) Chemisorption of CO2 on the 10Cu/CeO2{111} surface. Five oxygen vacancies, named after VO-A to VO-E, were made comparisons, and the selected CO2 was located close to VO-A. (e) The adsorption energy that CO2 was bound to VO on the CeO2{111} surface was obviously weaker than those of Cu dropped ceria sites.
In summary, the highlights of this work can be divided into the following points: (1) In a situation where sintered catalysts are often not favoured to catalysis, while we used a simple deposition-precipitation method and in situ sintering strategy to construct partially sintered copper-ceria catalyst with large number of stable copper cluster sites.
(2) For the RWGS reaction, the partially sintered copper-ceria catalyst exhibited a coexistence of excellent activity and high stability, which raised the RWGS reaction activity to a new level.
(3) The combination of experimental results and theoretical calculations revealed the precise pathway for the involvement of oxygen vacancies in CO2 activation and highlighted the importance of the synergistic effect between active metal clusters and oxygen vacancies in the catalytic reaction.
We believe that our work is comprehensive, innovative and will be definitely helpful for researchers in not only catalysis, but also in general chemistry, materials Although the study is very comprehensive and explains the behavior and the structure of the catalyst with great detail, the studied catalyst suffers from somewhat limited technical relevance, since ceria nanorods are unlikely to be applied as support material in industrial processes. Nevertheless, the results are of sufficient novelty and detail to justify publication in Nature Communications.
I have the following comments that should be considered by the authors: In Figure R1, it is unclear to which axis the shown bars refer.
In Figure R3 and R6, the differentiation between 2D and 3D clusters is not convincing.
Due to the cyclindrical shape of the nanorods, 3D clusters distributed on the surface of the nanorods could appear as flat or twodimensional (2D) depending on the viewing angle. Please provide more convincing results and arguments for the claimed 2D clusters.
Response: According to the reviewer's previous comments and suggestions, we added relative experiments and carefully revised the manuscript, which undoubtedly improved our work a lot. And thanks for the suggestions on our work again, we will reply to your comments one by one and look forward to your next comment. Figure R1, it is unclear to which axis the shown bars refer.

Response:
In order to more clearly demonstrate the catalytic performance of the catalyst under different reaction gas compositions, we divided the original graph into two figures to show the CO2 conversion and CO selectivity respectively. The Figure   R8 has been added as Figure S2 in the revised supporting information on page S7. Thanks for the reviewer's valuable comments again.

Comment 2:
In Figure R3 and R6, the differentiation between 2D and 3D clusters is not convincing. Due to the cyclindrical shape of the nanorods, 3D clusters distributed on the surface of the nanorods could appear as flat or twodimensional (2D) depending on the viewing angle. Please provide more convincing results and arguments for the claimed 2D clusters.

Response:
The definition of 2D and 3D clusters is based on the observed shapes of the clusters on CeO2 nanorods. As shown in Figure R9 and Figure R10, the diameter of 3D clusters is around 2 nm, while the thickness and width of 2D clusters are less than 1 and 3.5 nm. As the reviewer mentioned, we indeed observed 3D clusters which appear as round shape on the surface of CeO2 nanorods. Because atomic number of Cu is much smaller than that of Ce, the extra brightness introduced by 3D clusters on the surface of CeO2 is relatively small but still visible, as indicated by red dashed circles. On the other hand, due to thin layer of 2D cluster (less than 1 nm), the introduced brightness on the surface of CeO2 is almost invisible. If the incident electron beam is parallel to the plane of 2D clusters, the 2D cluster would appear as nanorods as indicated by two parallel dashed lines near the edge of CeO2.
The Figure R9 and R10 have been added as Figure S9 and