Boosting thermo-photocatalytic CO2 conversion activity by using photosynthesis-inspired electron-proton-transfer mediators

Natural photosynthesis proceeded by sequential water splitting and CO2 reduction reactions is an efficient strategy for CO2 conversion. Here, mimicking photosynthesis to boost CO2-to-CO conversion is achieved by using plasmonic Bi as an electron-proton-transfer mediator. Electroreduction of H2O with a Bi electrode simultaneously produces O2 and hydrogen-stored Bi (Bi-Hx). The obtained Bi-Hx is subsequently used to generate electron-proton pairs under light irradiation to reduce CO2 to CO; meanwhile, Bi-Hx recovers to Bi, completing the catalytic cycle. This two-step strategy avoids O2 separation and enables a CO production efficiency of 283.8 μmol g−1 h−1 without sacrificial reagents and cocatalysts, which is 9 times that on pristine Bi in H2 gas. Theoretical/experimental studies confirm that such excellent activity is attributed to the formed Bi-Hx intermediate that improves charge separation and reduces reaction barriers in CO2 reduction.

3-Bi2O3 has a bandgap of about 2.5-2.7 eV. Therefore, it can show a peak at around 450-500 nm. So, it seems that the broad peak between 400-600 (in Fig. 2d) is due to the convolution of both LSPR of Bi-Hx and partial Bi2O3 absorption. If the nature of the peak is LSPR, the authors can provide several samples with different particle sizes (by using different electrodeposition times from 3 to 15 min). Then, if the nature of the absorption is LSPR, they would observe a shift in the UV-visible absorption.
4-SAED patterns should be recorded at the edge and basal plane of both Bi-Hx and Bi particles. It can show if there are any other local secondary phases or not. 5-The higher activity of the Bi-Hx/NF could be due to the formation of nanopores during the hydrogenation process, rather than, or not only just because of the proton-assisted electron transfer effect as claimed by the authors. Therefore, the authors should provide direct experimental evidence that the overall surface areas of both samples (Bi/NF and Bi-Hx/NF) are almost the same.
6-(I) For CO2 reduction, the authors had applied heat, hence, there is additional input energy (ΔE), which should be added to the denominator of the AQE. In other words, the authors report the AQE values only based on photo-, but neglecting the thermo-contribution completely (Fig. 4f). (II) I will also suggest the authors to do the test under AM 1.5 solar light (only photo-catalysis, but not thermo-photocatalysis). How much will the AQE be for photo-catalysis? The result shall be compared to other reported systems.
7-Current data did not provide convincing evidence that the hydrogen (in Bi-Hx lattice) can have bifunctional roles (page 12 line 231). Especially, for the claimed point 2, the authors should do the XPS after GC test for different times to prove that there is a continuous shift that can probably be assigned to the consumption of H (i.e. H element belonging to the Bi-Hx phase).
8-The PEC measurements were not carried out in a stable condition. Degradation of the maximum photocurrent density is about 40% after several minutes (Fig. 5d)! The authors need to explain why? Or repeat the experiment. 9-As shown in XPS spectra, the very surface contains the Bi2O3 phase (quite dominant Bi3+ actually), which can change the calculated free energies (Fig. 6b). A serious concern is raised for the potential impact of this surface oxide layer on the whole CO2 reduction process when it is in the vicinity of the Bi-Hx phase. The authors have claimed the energy profiles are not dramatically affected by the presence of Bi oxide (comparing Supplementary Fig. 2b and Main Fig. 6b), however, the computational results can depend on the content of oxide in the model. I think the authors must provide clear evidence, preferably experimental, showing that the surface oxide layer has not had a big impact on the proposed reaction pathway.
10-What will happen if the authors do the same experiment in CO2 and H2O atmosphere? Can the introduction of H2O keep the concentration of hydrogen constant in the Bi-Hx system and provide better stability? (Li-Chyong Chen) Reviewer #3 (Remarks to the Author): Dear Authors find here my concerns. First and foremost for any photocatalytic reaction needs to analyze the band gap of material which is not provided in manuscript. In Fig 2d UV-Vis analysis doesn't show any significant difference between both catalyst then how both can behave differently prove it by providing additional analysis data like Pl(Photo luminescence analysis) that can show the recombination of excitons. From Fig 2b cyclic voltametric analysis nickel foam is not much sensitive toward electrocatalytic reaction thus the use of Nickel foam must be explained. From fig 2b It can also clearly observed the absorption/desorption peak of hydrogen that also questioned the stability of Bi-H x /NFi catalyst. Fig2c XPS analysis shows the significant shift of binding energy though other analysis though there is no extra oxidation state of Bi is reported thus the change must be explained. GC analysis presented the production of CO and hydrogen molecule in the system while photocatalytic reduction does not selective that clearly give the information about formation of other products or hydrocarbons, it require some additional analysis.
Comparative analysis data of 13CO and CO is not provided in manuscript to prove the reliability of reaction mechanism needs to be rechecked. Dissolution of gas in the water requires some extra outer forces to maintain the an equilibrium that is not explained in the manuscript need to be recheck and optimize the reactor conditions. In DFT analysis favors the mechanism of reaction but the same time free energy value with respect to change in the transition, state shows reaction happened under the artificial condition, it was not natural that need to be explained.
1. Li et al present a manuscript on solar-assisted CO 2 reduction. Despite long-standing efforts for developing photocatalysts for efficient conversion of CO 2 into feedstock chemicals, the currently known materials generally have a low performance and have some additional challenges such as co-generation of O 2 and CO 2 reduction products. I really like the concept of spatially and temporally decoupling the oxidation and reduction half reactions of the process, but this is by no means a new idea. It has been previously applied to water splitting (e.g. Nature Energy 2019, 4, 786-795;Science 2014Science , 345, 1326Science -1330 which should be acknowledged here.

Response:
The authors thank the reviewer for the thorough review. We thank the reviewer for liking "the concept of spatially and temporally decoupling the oxidation and reduction half reactions of the process". We also appreciate the reviewers' comments and suggestions that helped us to significantly improve the manuscript. We have addressed the reviewers' comments point-wise in our response below.
The novelty of our manuscript lies in the development of a novel plasmonic Bi catalyst that spatially and temporally reduce CO 2 into CO efficiently by storing H species in Bi. Since CO 2 is a thermodynamically stable molecule, its reduction via the electrocatalytic or photocatalytic approach is significantly more challenging than the splitting of water, and confronted with many fundamental technical hurdles (Adv. Sci. 2017, 4, 1700194). Although the two steps process has been used in water splitting, the application of this concept in CO 2 conversion are still a challenge because the CO 2 reduction process with H 2 O is much more complicated that contains the redox reactions between two different molecules of H 2 O and CO 2 . Furthermore, the material system, operation process and solved problems in the present study are quite different from the above references.

Action:
In the revised version of manuscript, we have discussed this point in more detail on Page 5 of the revised manuscript as shown in the following, and cited the relevant references in their proper places.
Similar with the natural photosynthesis, the decoupled approach for water splitting has been explored recently, in which water oxidation and proton reduction reactions were spatially separated. 16,17 Compared with water splitting, CO 2 reduction is much more difficult because CO 2 is a thermodynamically stable molecule with linear structure 1 . Furthermore, finding a single material to mimic natural photosynthesis for CO 2 conversion is more challenging because it should work as a redox shuttle to bridge the separated water splitting and CO 2 reduction reactions 12,18 , which is more complicated than the only water splitting process.
2. The authors claim that direct CO 2 reduction at semiconductors is not possible, but then include a list of examples from the literature in the SI that show exactly this reactivity. The paragraph describing this is misleading and should be rewritten. It is a common misconception that CO 2 reduction always has to go via the CO 2radical anion which has a redox potential of -1.9 V (not 1.9 V 5. Nevertheless, I like the idea of applying the decoupling concept to the half-reactions of CO 2 reduction, but I struggle to see the usefulness of the particular approach taken here. The authors are applying -1.2 V vs Ag/AgCl at pH 14 to generate a Bi-H intermediate which then reacts at >150°C with CO 2 when irradiated to give CO (all potentials should be given vs. RHE rather than vs Ag/AgCl in line with IUPAC recommendation to make the overpotentials more apparent).

Response:
We thank the reviewer for liking the two-step concept for CO 2 reduction in our present work. Our response on this question is divided into two separated parts on the high reaction temperature and electrochemical potential. As discussed below, both the Bi sample that can work under at high temperature, and the electrochemical process for H storage have their own unique advantages in CO 2 reduction.

Response (I):
About the reaction temperature The present Bi nanosheets are a special metal catalyst that has the surface plasmon resonance (SPR) property. We acknowledge the high temperature used in our study, although we do not utilize precious materials, sacrificial organics or cocatalysts. To the best our knowledge, all the reported plasmonic photocatalysts works only when the reaction temperature is relatively high (>150 °C). For example, the reaction temperatures for photocatalytic CO 2 conversion are 200-350 °C for plasmonic Au, and Rh (Nat. Commun. 2017, 8, 14542), 400 °C for plasmonic Pt-Au (ACS Appl. Mater. Interfaces 2018, 10, 408). Based on our understanding, plasmonic photocatalysis driven by photoinduced hot-electrons is a type of catalysis that between thermalcatalysis and photocatalysis. Therefore, it is reasonable that heat input is needed for activing the adsorbed molecules. Furthermore, most of the plasmonic photocatalysts are noble metal nanoparticles, such as Ag, Au, and Rh. While the present Bi is the first example of hot carrier driven highly selective CO 2 -to-CO conversion using a plasmonic metal without combining semiconductors that entirely relies on nonprecious, earth-abundant materials. The discussion on this point was added in our revised manuscript, which could bring novelty of our manuscript with clarity.
The field of heterogeneous photocatalysis has almost exclusively focused on semiconductor photocatalysts (Nat. Mater., 2015, 14, 567). Compared with the semiconductor-based photocatalysis that has dominated the research area since 1972, plasmonic photocatalysis just started from the year 2011(Nat. Chem., 2011. Before this report, plasmonic metals are mainly used as thermal catalyst or as light absorber by combining with semiconductor. Therefore, plasmonic metallic nanostructures represent a new family of photocatalysts, and only a few reports focus on CO 2 reduction (Adv. Mater. 2020, 2000086). In the thermochemical CO 2 -to-CO conversion process on metal catalysts, it is usually required a very high temperature of 550−750 °C (ACS Energy Lett. 2018, 3, 1938. Furthermore, the applications of photocatalytic systems based on plasmonic metals loaded semiconductors are significantly hindered by the low hot charge transfer efficiency from metal to catalytically active sites on the semiconductor, which is attributed to the built-in Schottky barrier at the semiconductor-metal interface (J. Am. Chem. Soc. 2017, 139, 17964).
Although conventional semiconductor photocatalysts offer a promising route to room temperature reactions, in general they exhibit lower reaction rates at higher temperatures due to the relatively low Debye temperatures of the semiconductors (Nat. Mater. 2012, 11, 1044Nat. Commun. 2017, 8, 14542). Therefore, in practical applications, if we want to increase the yield of products, the efficiencies for general photocatalysts can be significantly reduced by increasing the light intensity, which might be a potential factor that can hinder use of semiconductor photocatalysts because high-intensity light, especially IR light, will inevitably increase the temperature of the gas-solid reaction system (Nat. Commun. 2017, 8, 14542).
The plasmonic Bi nanoparticles in the present manuscript characterized by excellent absorption of visible light through the creation of resonant surface plasmons, can utilize concurrently thermal energy and a photon flux to drive catalytic reactions at moderately high temperature of 180 °C, which is much lower than those associated with conventional thermal processes. In fact, the heat input can be provided by the infrared light of sunlight (accounting for ~50% of the solar energy), which implies that the plasmonic Bi photocatalysts can potentially use the entire solar spectrum for driving catalytic reactions by employing a high-temperature solar reactor driven by concentrated solar radiation. In this respect, the ability that plasmonic photocatalysts can work at high temperature have their own unique advantages in CO 2 reduction.

Action (I):
To address this question from the reviewer, the following sentences were added into the revised manuscript on Pages 9-10 and Page 11-12 as:  30,35 . Therefore, in practical applications, the heat effect from concentrated sunlight might be one of the hindrances for efficient solar energy conversion in gas-solid reaction system by using semiconductor photocatalysts 30 . In this respect, the ability of the plasmonic Bi photocatalyst that can work at 180 °C might have its own advantage for CO 2 reduction in gas-solid reactions.

Response (II): About the electrochemical problems
According to the reviewer's suggestion, all the potentials in the revised manuscript were given by RHE. The corresponding formula was included in Methods part in the revised manuscript. The cyclic voltammograms of the as-prepared Bi/NF was modified accordingly as shown in the Fig. R1 below.
Electrochemical hydrogen storage in Bi was performed at a reduction potential of -0.18 V versus RHE (vs. RHE), which is not so high.
The unique advantage of the electrochemical process used in the present work is to convert H + in water to active H bounded on the surface of the Bi catalyst that can perform CO2 reduction with a much higher activity than H 2 . Traditionally, H 2 , which can be produced by electrocatalytic water splitting, steam reforming of natural gas or et al., was used for CO 2 reduction on metals. In these CO 2 reduction processes, the molecular hydrogen should be decomposed into atomic hydrogen first, which always requires an energy input. Compared with H 2 gas, atomic H atoms preformed on the surface not only improve photoinduced charge separation, but also lower the CO 2 reduction barriers as discussed in Fig. 5 and Fig. 6 in the revised manuscript. Therefore, the electrochemical hydrogen storage process provided an effective approach to produce reactive H for CO 2 reduction by transforming protons in water, which is much more active than traditional H 2 (Fig. 4e in the revised manuscript). This is one advantage for introducing electrochemical process in the catalytic circle. Furthermore, the present two-step approach provide an effective model for reducing CO 2 by using cheap and abundant H 2 O. Given the wide diversity of known H-intercalating materials with a wide range of H-affinities, the concept established here provides the basis of segregating bond activation and charge separation processes across a wide range of key energy conversion reactions.

Action (II):
We have provided additional discussion based on the reviewer's comments. The advantages of the electrochemical hydrogen storage process were added in the revised manuscript as shown in the response on the following comment (5), and the discussion in the DFT calculations in Fig. 6 of the revised manuscript. The following sentences were added into the revised manuscript on Pages 14-15, and 23. Furthermore, all the potentials were given by RHE in the revised manuscript.
For the CO 2 reduction process on Bi/NF with H 2 gas, the photocatalytic interface must facilitate two sequence steps (1) the adsorption and decomposition of the H 2 to generate atomic H bound on the surface, and (2) the transfer of these adsorbed H atoms into electron-proton pairs to reduce CO 2 (Fig.   4g). However, the two coupled steps (1 and 2) in both time and space preventing independent optimization of each, and the competitive adsorption of H 2 may impede CO 2 activation at the active sites 36 . Moreover, the decomposition of the H 2 in step (1) will inevitably require a certain energy input. By separating the above two steps (1 and 2), the present approach for CO 2 conversion addresses the incompatibilities of functions (1) and (2), and bypasses the H 2 decomposition process in traditional CO 2 reduction process with H 2 (Fig. 4h). As a result, the improved CO 2 reduction activity on Bi-H x /NF can be attributed to the atomic H atoms preformed on the surface, which not only improve photoinduced charge separation, but also lower the CO 2 reduction barriers. These two positive effects will be closely discussed in the following Fig. 5 and Fig. 6, respectively.
Considering the existence of large variety of H-intercalating materials, the concept developed here may have broad applications for developing efficient and multifunctional catalysts for CO 2 conversion by mimicking photosynthesis.
5. If I were to apply the same voltage (corresponding to 0.87 V overpotential for HER) to a cheap HER catalyst, e.g. Ni 2 P and use the generated H 2 to hydrogenate CO 2 , I would likely get much more CO and probably even more valuable chemicals such as MeOH.
Response: We agree that electrocatalytic H 2 O splitting is an appropriate approach for producing H 2 by using some nonprecious metal based electrocatalysts. However, the electrocatalysts, such as Ni 2 P, cannot catalyse the further hydrogenation reaction of CO 2 . The fascinating point of the Bi catalyst is the bifunctional properties that can not only induce electrocatalytic H 2 O splitting and H storage, but also drive the CO 2 reduction reactions under light irradiation. More importantly, the surface-bound H atoms is a more efficient proton source than H 2 for CO 2 reduction process in the same photochemical reaction conditions, which is another main point of this manuscript.
It is well known that H 2 is traditionally used as the proton source for photochemical CO 2 conversion via the proton-assisted electron transfer approach. In this process, photochemical CO 2 conversion relies on the transfer efficient of electron−proton pairs from the H 2 molecules under light irradiation, and the photochemical interface must facilitate two sequence steps (1) the decomposition of the H 2 to generate surface-bound H atoms, and (2) the transfer of these adsorbed H atoms into electron-proton pairs to reduce CO 2 . However, the two coupled steps (1 and 2) in both time and space preventing independent optimization of each, and competitive adsorption of H 2 may impede CO 2 activation at the active sites. Moreover, the decomposition of the H 2 in step (1) will inevitably require a certain energy input.
In the present manuscript, we establish a novel strategy for CO 2 reduction by exploiting an electron-proton-transfer mediator (Bi-H x ) with recyclability, which was generated from electrochemical H 2 O splitting of Bi electrode. By segregating the above two steps (1 and 2), the strategy in our manuscript addresses the incompatibilities of functions (1) and (2), and avoids the H 2 activation process in traditional CO 2 reduction process using H 2 as proton source. In order to show that the current approach using surface-bound H atoms as proton source has advantages over the traditional H 2 , the controlled experiments for CO 2 reduction were carried out on the same Bi material under the similar experimental conditions. As shown in Fig. 4e, the photochemical CO production activity using surface-bound H atoms on Bi-H x as proton source is 9 times that on pristine Bi using H 2 molecules as the proton source (  For the CO 2 reduction process on Bi/NF with H 2 gas, the photocatalytic interface must facilitate two sequence steps (1) the adsorption and decomposition of the H 2 to generate atomic H bound on the surface, and (2) the transfer of these adsorbed H atoms into electron-proton pairs to reduce CO 2 (Fig.   4g). However, the two coupled steps (1 and 2) in both time and space preventing independent optimization of each, and the competitive adsorption of H 2 may impede CO 2 activation at the active sites 36 . Moreover, the decomposition of the H 2 in step (1) will inevitably require a certain energy input.

By separating the above two steps (1 and 2), the present approach for CO 2 conversion addresses the incompatibilities of functions (1) and (2), and bypasses the H 2 decomposition process in traditional
CO 2 reduction process with H 2 (Fig. 4h). As a result, the improved CO 2 reduction activity on Bi-H x /NF can be attributed to the atomic H atoms preformed on the surface, which not only improve photoinduced charge separation, but also lower the CO 2 reduction barriers. These two positive effects will be closely discussed in the following Fig. 5 and Fig. 6, respectively.
6. Do the authors observe any other CO 2 reduction products?
Response: Other products such as CH 4 , and CH 3 OH were not observed by mass spectrum from GC-MS and GC analysis.

Action:
The corresponding description on this point has been added into the revised manuscript on Pages 10, 12, and 25 as: Other possible products (such as CH 4 , and CH 3 OH) were not detected, confirming that high selectivity for CO 2 reduction was achieved on Bi-H x /NF. Fig. 4c, except for the peaks belonging to CO 2 and CO, no other peaks were observed in 7. It would also be possible to apply the same potential to a Bi-based electrode to directly reduce CO 2 with fairly decent performance and at room temperature (well-documented electrocatalytic activity e.g. J. Mater. Chem. A 2018, 6, 4714-4720 and many others), and with an OER anode separated by a membrane, CO 2 reduction and O 2 generation could also be spatially separated.

Response:
Although the electrocatalytic approach can be operated at room temperature, some drawbacks also exist in this method. As we know that formate is mainly produced on the Bi-based electrodes based on the references (J. Mater. Chem. A 2018, 6, 4714;Nat. Commun. 2018, 9, 1320Nat. Commun. 2019, 10, 2807. However, CO 2 reduction to liquid fuels by electrochemical approach is currently challenged by low product concentrations, as well as their mixture with traditional liquid electrolytes, such as KHCO 3 solution (Nat. Commun. 2020, 11, 3633). Therefore, the product cannot be directly used without further purification. The low concentration nature of the liquid fuel makes the purification process more difficult.
Based on the above references on Bi electrocatalyst, the major product in CO 2 reduction is formate (HCOO − ). Different from acidic form, formate has no obvious usage (Angew. Chem. Int. Ed. 2018, 57, 2943. In sharp contrast, CO gas is a critical feedstock for directly synthesizing a variety of chemicals in industry (Angew. Chem. Int. Ed. 2018, 57, 2943. Therefore, the CO 2 -to-CO conversion with high selectivity achieved on the present Bi catalyst has great potential in practical applications.
Furthermore, the electrocatalytic approach displays poor solubility for gaseous substrates and limits the rate of the substrate bond activation. Therefore, it is difficult to perform direct CO 2 reduction in electrocatalytic approach (J. Am. Chem. Soc. 2019, 141, 28, 11115). Finally, the membrane used in electrocatalysis is expensive and the membrane degradation tends to be occurred at low current densities (Science 2014(Science , 345, 1326(Science -1330. It is lucky that the above three drawbacks do not exist in the present Bi system for CO 2 reduction. However, it should be keep in mind that there remains a noticeable absence of a consummate solution that combines all the ideal properties together in one material.

Action:
We have provided additional discussion based on the reviewer's comments. The following sentences were added into the revised manuscript on Page 22 as: Although electrocatalytic reduction of CO 2 into formate (HCOO − ) on Bi electrode can reach a much higher FE efficiency than the electrochemical H 2 O splitting on the present Bi/NF, much higher overpotentials between -0. 67 and -0.87 V (vs. RHE) were used in previous reports [43][44][45] . Furthermore, the direct reduction of CO 2 by the electrochemical approach is not a perfect technology because it still suffers from low product concentrations in the mixture of traditional liquid electrolytes 46 . As a result, the product cannot be directly used without further purification. However, the purification of the low concentration liquid fuel in electrolytes not only compromises energy efficiency, but also is a technical challenge 46 .
It is well known that different from acidic form, formate has no obvious usage 47 . In contrast, CO gas is a critical feedstock for directly synthesizing a variety of chemicals in industry 47 . Therefore, compared with electrochemical CO 2 reduction on Bi, the CO 2 -to-CO conversion achieved on the present Bi catalyst has great potential in practical applications. Furthermore, the electrocatalytic approach displays poor solubility for gaseous substrates and limits the rate of the substrate bond activation. Therefore, it is difficult to perform direct CO 2 reduction in electrocatalytic approach 36 . In addition, although the gas product and O 2 generation in electrocatalytic CO 2 conversion could also be spatially separated by a membrane, the used membrane is expensive and tends to be degraded at low current densities 17 . However, the above three drawbacks do not exist in the present catalytic system for CO 2 reduction. Based on our understanding, there is a noticeable absence of a consummate material that can solve all of these challenges in CO 2 reduction together.
8. The authors mention water splitting and overall CO 2 splitting several times throughout the manuscript and claim O 2 generation in the abstract, but in fact there is no water oxidation performed anywhere in the manuscript or SI, but only the reductive half-reaction. This must be rephrased in the manuscript.

Response:
Thanks for your suggestion. The produced H 2 and O 2 products in electrochemical water splitting were identified by gas chromatography (Agilent 7890B) and Neofox-GT oxygen probe (Ocean opticals), respectively. 9. From the data provided in the manuscript, it is difficult to work out exactly how efficient the presented system is. There is no chronoamperometry given, so it is not known how much charge is  2018, 6, 4714;Nat. Commun. 2018, 9, 1320Nat. Commun. 2019, 10, 2807, much higher overpotentials between -0. 67 and -0.87 V (vs. RHE) were used. Furthermore, the present manuscript does not aim to develop high efficient electrocatalyst, but rather demonstrate a novel two-step reaction model mediated by plasmonic Bi-H x for CO 2 conversion by mimicking natural photosynthesis, which not only improves the CO 2 conversion activity but also avoids O 2 separation from CO product.
However, improving the electrochemical performance of Bi is very important for reducing the overall energy input in the present two-step CO 2 reduction process, which is the subject of ongoing investigations in our laboratory.  [43][44][45] . Furthermore, the direct reduction of CO 2 by the electrochemical approach is not a perfect technology because it still suffers from low product concentrations in the mixture of traditional liquid electrolytes 46 . As a result, the product cannot be directly used without further purification. However, the purification of the low concentration liquid fuel in electrolytes not only compromises energy efficiency, but also is a technical challenge 46 .
10. The authors give an amount of Bi-H formed in the material and that only 25% of that is used to generate CO, but it is not clear how this was calculated. Can the presence of Bi-H be confirmed and quantified by Raman spectroscopy, ideally as a function of the applied bias/electrolysis time?

Response:
The 25% H used to generate CO is a misunderstanding, which was caused by our unclear description. The fact is that we compared the CO 2 activity of Bi-H x with pristine Bi (using H 2 as the proton source) in the same experimental conditions. We want to express that the total amount of protons stored in Bi-H x is 25% of that added in Bi reaction system (calculated from the added amount of H 2 molecules).

Action:
We have revised the corresponding sentences in the revised manuscript on Pages 13 to make this point more clearly: It is well known that H 2 is traditionally used as the proton source for photochemical CO 2 conversion via the proton-assisted electron transfer approach 3 . In order to show that the current approach using surface-bound H atoms as proton source has advantages over the traditional H 2 , the controlled experiment for CO 2 reduction was carried out on Bi/NF material in dissociative H 2 gas (0.5 mL). As shown in Fig. 4e, although the number of hydrogen atoms stored in the Bi-H x /NF systems (13.52 μmol) is only about 25% of that from H 2 in Bi/NF reaction system, the CO 2 reduction activity of Bi-H x /NF reaches 9 times that of Bi/NF. 11. Can the authors rule out that the increased activity of Bi-H is not only due to the increased presence of Bi(0) in the material? Is the ratio of Bi(0) and Bi(3) changed after photocatalysis?

Response (I):
We thank the reviewer for this constructive comment. In order to exclude that the increased activity of Bi-H x is not caused by the increased amount of Bi (0)

Action:
The following discussions were added into the revised manuscript on Page 13-14 as: As shown in the XPS spectra in Fig. 2d, (Supplementary Fig. 6). In the following, the controlled experiment for CO 2 reduction was carried out on the sample after heat treatment. Fig. 4d, no performance degradation on CO production was found on Bi-H x /NF when the surface Bi 0 species was replaced by Bi 3+ (Fig. 4f)  12. The authors claim that the performance of their photocatalyst is much better than what is known from the literature. Is this really true? Table S1 is incomplete doesn't list the conditions under which the CO 2 reduction was performed, but I would assume that most of the examples were done at room temperature, whereas the present work was done at >150°C, and I would expect the performance to be drastically lower if the present system was operated at room temperature, so it is like comparing apples and oranges. In addition, the given examples actually use water as electron donor, whereas the present work uses Bi-H as sacrificial electron donor. The additional potential and efficiency losses of oxidising water would need to be factored in for a fair comparison. The observed quantum yield is also quite low, other (room temperature) processes achieve much better performances (e.g. ACS Appl.

Compared with the activity of Bi-H x /NF in
Energy Mater. 2020, 3, 5, 4509-4522). Temperatures and donors should be added to the table and it might be fairer to compare the present work with other examples of thermal CO 2 hydrogenation, which usually operates at >200°C, which isn't that far off and has much higher activities.

Response:
We agree that it is very difficult to compare our results with the literatures. This is not because of our experimental conditions, but rather the significant diversity of utilized materials and reported experimental conditions for photocatalytic CO 2 conversion. Improving quantum yield is the Holy Grail in all photocatalytic processes, and in particular for CO 2 conversion. However, the present manuscript should not emphasize CO 2 conversion efficiency after carefully considering the reviewer' comments, but rather report the following two important findings: First, our study demonstrates a novel two-step reaction model mediated by plasmonic Bi-H x for CO 2 conversion by mimicking natural photosynthesis, which not only improves the CO 2 conversion activity but also avoids O 2 separation from CO product. Second, we also demonstrates the first example of hot carrier driven highly selective CO 2 -to-CO conversion using bare plasmonic metal nanoparticles that entirely relies on nonprecious, earth-abundant materials. To improve the quality of this manuscript, we have provided an additional discussion about this advantage of our study in the revised manuscript.
We believe it is important to compare the apparent quantum efficiency of our system with other plasmonic systems that have reported photocatalytic CO 2 conversion. Even though there are dozens of reports at room temperature, our measured CO formation rate of 283.8 μmol·g −1 ·h −1 on Bi-H x is even higher than many previous reports of photocatalytic CO 2 conversion (ACS Catal. 2016, 6, 7485-7527). Even though there are lots of reports on plasmonic metals for photocatalytic CO 2 reduction, these plasmonic metals were mainly coupled with semiconductor photocatalysts. To date, only several reports for photocatalytic CO 2 conversion were focused on plasmonic systems without coupling semiconductor, and all of them were based on noble-metals such as Au, Au-Ag, and Rh (Nat. . 2017, 8, 14542;ACS Appl. Mater. Interfaces 2018, 10, 408). To objectively understand the quantum efficiency of the plasmonic Bi catalysts, activities of thermal CO 2 hydrogenation on metals, photocatalytic CO 2 reduction based on plasmonic metal (Au, Ag, and Rh), and plasmonic metal coupled with semiconductor systems were discussed systematically based on reaction temperature, and the used light intensity in the revised manuscript.

Action:
In the revised version of manuscript, we have discussed and compared our efficiencies with literature reports on photocatalytic CO 2 conversion as well as similar plasmonic systems for CO 2 conversion. The Refs. (Nat. Commun. 2017, 8, 14542;ACS Appl. Mater. Interfaces 2018, 10, 408;ACS Appl. Energy Mater. 2020, 3, 5, 4509-4522 andNat. Commun. 2017, 8, 27;Adv. Mater. 2020, 2000086 and et al.) were added in the revised manuscript. The following sentence was added into the revised manuscript on Pages 10 and 15-16 as: More details on the photocatalytic CO 2 reduction efficiencies on semiconductor systems can be found in the recent review articles 1,4

. Rather than the catalytic efficiency, the main point of the present manuscript is to demonstrate a novel two-step reaction model for CO 2 conversion by using a plasmonic Bi metal that entirely relies on nonprecious materials.
13. I am not 100% convinced by the mechanism. What exactly is being oxidized in the process? The fact that the non-hydrogenated catalyst still shows some activity is confusing. Where are the reducing equivalents coming from? What is being oxidized in this case? Do the authors have some experimental evidence to support the proposed mechanism or is this based entirely on calculations?
What is being oxidised in the photoelectrochemical experiments?
Response: This is a very reasonable concern. Our response on this question is divided into two separated parts as shown in the following.

Response (I):
The Bi catalyst without hydrogen storage showed CO 2 reduction activity because H 2 molecules were used as reducing agent and as the proton source. In this process, H 2 was oxidized under light illumination. We are very sorry for the unclear description in our original manuscript, which indeed cause misleading to the reader as the reviewer indicated. The authors appreciate the reviewer for catching this mistake.

Action (I):
We have replaced that statement with a clarified version in the main text. Please review Page 13 of the revised manuscript for our response as shown in the following. For clarity, the descriptions in Fig. 4e were modified correspondingly as shown in the Fig. R7 above.

It is well known that H 2 is traditionally used as the proton source for photochemical CO 2 conversion
via the proton-assisted electron transfer approach 3 . In order to show that the current approach using surface-bound H atoms as proton source has advantages over the traditional H 2 , the controlled experiment for CO 2 reduction was carried out on Bi/NF material in dissociative H 2 gas (0.5 mL). As shown in Fig. 4e,

although the number of hydrogen atoms stored in the Bi-H x /NF systems (13.52 μmol)
is only about 25% of that from H 2 in Bi/NF reaction system, the CO 2 reduction activity of Bi-H x /NF reaches 9 times that of Bi/NF.

Response (II):
In order to experimentally prove that the thermal-photocatalytic CO 2 reduction on Bi-H x was performed by the proton-assisted electron transfer approach, the following experiments were designed. First, CO 2 reduction on Bi-H x /NF was performed at 180 °C without light illumination by using H 2 O (200 μL) as proton source. The negligible CO production activity as shown in Fig. R8 below demonstrates that photoinduced charges are the primary factor for CO 2 reduction on Bi-H x /NF. Fig. R9 below, the formation of no CO on the bare Bi without storing H justifies the hypothesis that the hot-carrier transfer into unoccupied orbitals in CO 2 for C-O bond dissociation can be hardly occurred. Based on the above two experiments, we can concluded that CO production performance originates from the synergetic effect between H + and hot e -.

Second, as shown in
To further understand the reaction mechanism, the effect of irradiation intensity on the thermal-photocatalytic performance of Bi-H x was examined (Fig. R10). The strong dependence of the evolution rate of CO on the light intensity suggests that the photogenerated charges were responsible for the CO 2 reduction. This result also mean that synchronous production of reactive H + and e − pairs in Bi-H x was achieved by the consumption of photogenerated holes by the stored H.
Moreover, the XPS curves of Bi-H x /NF after different thermo-photocatalytic reaction times were measured (Fig. R11). It can be seen that the XPS band ascribed to the oxidation state of Bi progressively blue-shifts with increasing reaction time, which is consistent with the fact that the consumption of H species proceeds during the CO 2 reduction reaction on Bi-H x /NF. The above experiments, taken together, provide a solid evidence that the proton-assisted electron transfer approach indeed took place in the CO 2 reduction process of Bi-H x /NF as shown in Fig. 5a-c, which is consistent with the DFT calculations in Fig. 6.

Action (II): Based on the reviewer's comments, the proton-assisted electron transfer mechanism on
Bi-H x has been investigated step by step, which provided directly experimental evidence on this reaction mechanism. A detailed discussion was added in the revised manuscript on Page 18-19 as shown in the following.

In order to experimentally prove that the thermal-photocatalytic CO 2 reduction on Bi-H x was
performed by the proton-assisted electron transfer approach, the following experiments were designed.

First, CO 2 reduction on Bi-H x /NF was performed at 180 °C without light illumination by using H 2 O
(200 μL) as proton source. The negligible CO production activity (Supplementary Fig. 7) demonstrates that photoinduced charges are the primary factor for CO 2 reduction on Bi-H x /NF.

Second, under the thermal-photocatalytic conditions, the formation of negligible CO on the bare
Bi/NF without using proton sources (Supplementary Fig. 8

) confirms that the CO 2 -to-CO conversion
can hardly occur only with hot-electron transfer into unoccupied orbitals of adsorbed CO 2 . Based on the above two experiments, we can conclude that the thermal-photocatalytic CO production performance on Bi-H x /NF should originate from the synergetic effect between H + and hot e − , which is accordant with previous reports 1−3 . The above experiments, taken together, provide a solid evidence that the proton-assisted electron transfer approach indeed took place in the CO 2 reduction process of Bi-H x /NF as shown in Fig. 5a-c, which is consistent with the DFT calculations in Fig. 6.
To further understand the reaction mechanism, the effect of irradiation intensity on the thermal-photocatalytic CO 2 reduction activity of Bi-H x /NF was examined (Supplementary Fig. 9). The strong dependence of the evolution rate of CO on the light intensity agrees well with the proposed mechanism that the synchronous production of reactive H + and e − pairs in Bi-H x Fig. 10). The reaction time dependent XPS spectra also agree well with the proton-assisted electron transfer mechanism for CO 2 reduction on Bi-H x /NF. In addition, based on the XPS analysis of Bi-H x /NF, no obvious change on the ratio of Bi 0 and Bi 3+ is observed after 27 h reaction, indicating the good stability of the catalyst under thermo-photocatalytic reaction conditions. 14. Overall, I am not convinced this brings the quality and novelty needed for Nature

Communications.
Response: In this revised manuscript, we considered all the comments of reviewers, and have tried our best to revise the manuscript accordingly. In addition to the novel two-step approach for CO 2 conversion as involved in our original manuscript, we have addressed the ambiguities related to proton-coupled electron transfer process by designing new experiments and have strengthened the cooperative contribution of proton and electron transfer in the two-step catalysis, which enhance the mechanistic aspect of our paper and provide theoretical and experimental insights on this two-step approach. Furthermore, we also take a first step by overcoming the challenges related to the cost, abundance, and hence the feasibility of implementing plasmonic systems for light-driven photocatalysis of important chemical transformations, which could bring novelty of our manuscript with clarity. To improve and strengthen the manuscript, we performed additional discussion about this important point in the revised manuscript. We truly appreciate the reviewer to keep in mind these important points while reevaluating this work for Nature Communications.
1. In the present work, Y. Li et al. study a hydrogen-stored Bi (Bi-H x ) system as an electron proton-transfer mediator for CO 2 -to-CO conversion. They claimed that the stored hydrogen not only can react with CO 2 molecule to produce H 2 O and CO but also it plays a role as a hole scavenger to promote the charge separation. The overall reaction is carried out under light at a certain temperature (180 °C), resulted in a high CO production yield of 283.3 μmol g -1 h -1 . Using DFT calculations, they show that the energy barrier via a proton-assisted electron transfer pathway is much smaller on the Bi-H x than the pristine Bi. However, they miss a deeper discussion on the material characterizations that are reflected in my comments below. This manuscript reports original results and a novel idea.
Hence, I think this work can be published in Nature Communications after major revisions.

Response:
We appreciate positive comments on our manuscript. Based on the comments below, we have improved the manuscript and it has further raised the quality of the manuscript.
2. The CO 2 reduction process is carried out under light at temperatures much higher than room temperature, which is used in the conventional photocatalytic CO 2 reduction. Therefore, I suggest the authors use the term "thermo-photocatalytic" (or "photo-thermocatalytic", depending on the primary importance of a thermo-or photo-effect) in the title and other related parts of the manuscript.

Response:
We thank the review for this suggestion.

Action:
The term "thermo-photocatalytic" was used in the title and other related parts of the manuscript. Response: (I) The XPS measurements on the Bi-H x and Bi were repeated. As shown in Fig. R12 below, the peaks belong to the Bi 3+ and Bi 0 can be clearly separated. The XPS data in Fig. R12 were similar with the result reported in the literature (ACS Catal. 2020, 10, 743−750), ensuring the correctness of this test. Furthermore, from the repeated XPS spectra in Fig. R11 above, it can be confirmed that similar spectra were obtained in XPS measurements of these samples, also confirming the correctness of the XPS tests in this time. Response: (II) This problem was not existed in the new XPS data as shown in Fig. R12 below. Response: (III) According to the suggestion of the reviewer, the deconvoluted oxygen spectra of the two samples are shown in Fig. R13 below. O 1s XPS spectra were deconvoluted into three peaks that are ascribed to Bi-O bands, surface hydroxyl oxygen, and adsorbed O 2 , which further confirms the generation of Bi−O in Bi 2 O 3 on the surface of the two samples.
In addition, the C 1s XPS spectra of the two samples were also examined as shown in Fig. R14 below. The two samples show similar C 1s XPS spectra which is consistent with the XPS results of Bi and O. The carbon species on the two sample can deconvoluted into three peaks corresponding to C−C, C−O, and C=O bonds, respectively (Chem. Eng. J. 2019, 374, 231), which is mainly caused by the adsorbed CO 2 . Based on this fact, the C 1s XPS spectra might be not helpful in understanding the oxidation state of Bi. Therefore, the C 1s XPS spectra were not included in the revised manuscript. Response: (IV) To investigate the thickness of Bi 2 O 3 on Bi, the transmission electron microscope (TEM) analysis was performed. As shown in Fig. R15 below, amorphous Bi 2 O 3 layers with a thickness of ~3 nm were clearly seen on the out surface of Bi of Bi-H x and Bi, respectively.   Fig. R12, an updated figure on XPS measurements has been embedded in the revised manuscript. The high-resolution spectrum of oxygen spectra of Bi-H x and Bi were added in the supporting information (Supplementary Fig. 1), which indicate that only Bi 2 O 3 was formed on the surface of Bi-H x and Bi. The HRETM image at the edge of Bi-H x and Bi were examined, and Bi 2 O 3 layers with a thickness of ~3 nm were clearly seen as shown in Fig. R15. Moreover, the corresponding discussions were added in the revised manuscript on Pages 7 and 8.

Action: As shown in
High-resolution XPS spectra of O 1s for the two samples were examined to further confirm the surface chemistry of Bi (Supplementary Fig. 1) As shown by the XPS spectra in Fig. 2d, the surface of Bi can be easily oxidized into Bi 2 O 3 in the air.
To investigate the formed Bi 2 O 3 layer, the TEM analysis was performed on the edge of the Bi-H x and Bi sheets. As shown in Fig. 3h and i, amorphous Bi 2 O 3 layers with a thickness of ~3 nm were clearly formed on both surfaces of Bi-H x and Bi.
4. Bi 2 O 3 has a bandgap of about 2.5-2.7 eV. Therefore, it can show a peak at around 450-500 nm. So, it seems that the broad peak between 400-600 (in Fig. 2d) is due to the convolution of both LSPR of Bi-H x and partial Bi 2 O 3 absorption. If the nature of the peak is LSPR, the authors can provide several samples with different particle sizes (by using different electrodeposition times from 3 to 15 min).
Then, if the nature of the absorption is LSPR, they would observe a shift in the UV-visible absorption.
Response: According to the suggestion of the reviewer, the morphology of Bi-H x nanostructures were tuned by varying the deposition time (1, 4, and 15 min). SEM images in Fig. R16 below show that the size of the Bi sample was great affected by the deposition time. XRD measurements show that elemental Bi was formed at these situations as shown in Fig. R17a below. The optical absorptions of the obtained samples in Fig. R17b below shows that the localized surface plasmon resonance (LSPR) effect of a nanostructure is strongly correlated to the size of the sample. For example, no LSPR absorption was shown on the nanoparticulate of Bi with a diameter of about 100 nm, and the LSPR peak increases in intensity and red-shifts with increasing size of Bi. This phenomenon confirms that the absorption peak in Fig. 3h in revised manuscript is indeed caused by the LSPR absorption of Bi. To prove that this peak is caused by the localized surface plasmon resonance (LSPR) absorptions, the Bi nanostructures with sizes of ~100 and ~ 400 nm were further synthetized at the deposition time of 1 and 4 min, respectively ( Supplementary Fig. 2). XRD measurements show that elemental Bi was formed at these situations ( Supplementary Fig. 3a). Then, the corresponding absorption spectra of the obtained Bi/NF at different deposition time were tested (Supplementary Fig. 3b). The Bi sample with a diameter of ~100 nm shows a bandgap absorption onset of ~600 nm, whereas the absorption peak at approximately 480 nm in Fig. 3h disappears. This optical phenomenon implies that the absorption peak between 400 and 600 nm in Fig. 3h was induced by LSPR, rather than from the coupling effect of Bi and Bi 2 O 3 layer. Moreover, by comparing the spectra of Bi samples with sizes of ~100 nm and 400 nm, we can find that the position of the absorption peak shows a red-shift with the size increasing ( Supplementary Fig. 3b), which is consistent with the characteristic of LSPR absorption that is strongly correlated to the shape and size of the materials 25 . The above optical properties of the samples confirm that the absorption peak at 480 nm in Fig. 3h can be attributed to the LSPR of Bi, which is consistent with the previous report 26 .

SAED patterns should be recorded at the edge and basal plane of both Bi-H x and Bi particles. It can
show if there are any other local secondary phases or not.
Response: As suggested, the structures of Bi-H x and Bi particles were characterized by selected-area electron diffraction as shown in Fig. R18 below, which confirmed the single crystalline nature of the samples Based on the results in SAED and HRTEM, the as-synthesized products consist of a single Bi phase except the amorphous Bi 2 O 3 layers.

Action:
We have now included the corresponding SAED patterns in Fig. 3 in the manuscript. The corresponding discussions were added in the revised manuscript on Page 8. Fig. 3h) and Bi-H x (insert of Fig. 3i) show that only the spots ascribing to the rhombohedral Bi are observed, suggesting the single crystalline nature of the samples.

The SAED patterns of Bi (insert of
6. The higher activity of the Bi-H x /NF could be due to the formation of nanopores during the hydrogenation process, rather than, or not only just because of the proton-assisted electron transfer effect as claimed by the authors. Therefore, the authors should provide direct experimental evidence that the overall surface areas of both samples (Bi/NF and Bi-H x /NF) are almost the same.

Response:
We have measured the specific surface areas of Bi 2 O 3-x and Bi. The result showed that the specific surface areas of Bi-H x and Bi were 3.3356 m 2 •g −1 and 2.1039 m 2 •g −1 , respectively, demonstrating that the improved CO 2 reduction activity of Bi-H x /NF cannot be mainly ascribed to the effect of surface area.

Action:
The corresponding results are added in the revised manuscript on Page 14 as:

In addition, the specific surface areas of the two samples were studied by the Brunauer-Emmett-Teller (BET) method based on the nitrogen adsorption isotherm, showing that the BET surface areas of
Bi-H x and Bi are 3.3356 m 2 •g −1 and 2.1039 m 2 •g −1 , respectively. This result demonstrates that the surface area should not be the main factor for the enhanced CO 2 reduction activity of Bi-H x /NF. 7. (I) For CO 2 reduction, the authors had applied heat, hence, there is additional input energy (ΔE), which should be added to the denominator of the AQE. In other words, the authors report the AQE values only based on photo-, but neglecting the thermo-contribution completely (Fig. 4f). (II) I will also suggest the authors to do the test under AM 1.5 solar light (only photo-catalysis, but not thermo-photo-catalysis). How much will the AQE be for photo-catalysis? The result shall be compared to other reported systems.

is the CO evolution rate (mol/s), N A is Avogadro constant, W is the total energy input (W), A is irradiation area (m 2 ), t is the time of light illumination (s), λ is the corresponding wavelength (m), h
is 6.62 × 10 −34 J•s −1 , and c is 3.0 × 10 8 m•s −1 .

Response (II):
According to the suggestion, photocatalytic CO 2 reduction on Bi-H x /NF was carried out under AM 1.5 solar light irradiation (100 mW cm −2 ) at ambient temperature. However, negligible amount of CO was produced as shown in Fig. R19. This result is not beyond expectation because Bi is a plasmonic metal photocatalysts. To the best our knowledge, all the reported plasmonic photocatalysts based on metal in different reactions, such as CO 2 reduction, can only work when the reaction temperature is relatively high (>150 °C). For example, the reaction temperatures for photocatalytic CO 2 conversion are 200-350 °C for plasmonic Au, and Rh (Nat. Commun. 2017, 8, 14542), 400 °C for plasmonic Pt-Au (ACS Appl. Mater. Interfaces 2018, 10, 408). (10) 8. Current data did not provide convincing evidence that the hydrogen (in Bi-H x lattice) can have bi-functional roles (page 12 line 231). Especially, for the claimed point 2, the authors should do the XPS after GC test for different times to prove that there is a continuous shift that can probably be assigned to the consumption of H (i.e. H element belonging to the Bi-H x phase).

Response:
We thank the review for this suggestion, which is helpful for understanding the CO 2 reduction mechanism on Bi-H x . According to the suggestion, the XPS curves of of Bi-H x at different reaction times were tested. As shown in Fig. R20 below, the XPS band progressively blue-shifts with increasing reaction time, which is clearly attributable to a losing of H species.  Supplementary Fig. 10 Fig. 10). The reaction time dependent XPS spectra also agree well with the proton-assisted electron transfer mechanism for CO 2 reduction on Bi-H x /NF. In addition, based on the XPS analysis of Bi-H x /NF, no obvious change on the ratio of Bi 0 and Bi 3+ is observed after 27 h reaction, indicating the good stability of the catalyst under thermo-photocatalytic reaction conditions. 9. The PEC measurements were not carried out in a stable condition. Degradation of the maximum photocurrent density is about 40% after several minutes (Fig. 5d)! The authors need to explain why?
Or repeat the experiment.
Response: According to the suggestion, the experiment was repeated accordingly. As shown in Fig.   R22 below, no obvious degradation on photocurrent density was observed in this time. 10. As shown in XPS spectra, the very surface contains the Bi 2 O 3 phase (quite dominant Bi 3+ actually), which can change the calculated free energies (Fig. 6b). A serious concern is raised for the potential impact of this surface oxide layer on the whole CO 2 reduction process when it is in the vicinity of the Bi-H x phase. The authors have claimed the energy profiles are not dramatically affected by the presence of Bi oxide (comparing Supplementary Fig. 2b and Main Fig. 6b), however, the computational results can depend on the content of oxide in the model. I think the authors must provide clear evidence, preferably experimental, showing that the surface oxide layer has not had a big impact on the proposed reaction pathway.

Response:
In order to experimentally prove that the surface oxide layer have a negligible effect on the CO 2 reduction process, the Bi(0) on the surface Bi-H x sample was transferred into Bi 2 O 3 by treating the sample in air at 80 °C for 0.5 h. As proved by the XPS in Fig. R22 below, only the peaks belonging to Bi(3) were observed, indicating the surface of the sample was fully covered by Bi 2 O 3 layer after the heat treatment. Then, the controlled experiment for CO 2 reduction was carried out on the Bi-H x sample after heat treatment. By comparing the activity in Fig. R23 below with that in Fig 4d in the revised manuscript, no catalytic performance degradation was found when Bi(0) was replaced by Bi (3), indicating that the formed Bi 2 O 3 layer in air did not affect the CO 2 reduction activity on Bi-H x /NF, which is consistent with the DFT calculations in Fig. 6. Despite the ~3 nm amorphous Bi 2 O 3 layer surrounding the Bi core, tunneling of the hot carrier to the oxide surface is still viable due to a high-density of defect states in the amorphous Bi 2 O 3 layer and a high energy of the hot charge carriers produced by localized surface plasmon resonance (Adv. Mater. 2020, 2000086). Action: The photocatalytic CO 2 reduction activity of the Bi-H x /NF sample heated in air at 80 °C for 0.5 h was tested and shown in Fig. R23. We have provided corresponding detailed discussion in the revised manuscript on Pages 13-14 as: As shown in the XPS spectra in Fig. 2d, (Supplementary Fig. 6). In the following, the controlled experiment for CO 2 reduction was carried out on the sample after heat treatment. Fig. 4d, no performance degradation on CO production was found on Bi-H x /NF when the surface Bi 0 species was replaced by Bi 3+ (Fig. 4f) Response: According to the reviewer's comments, we performed the CO 2 reduction reaction in CO 2 and H 2 O atmosphere by adding 200 μL H 2 O in the reactor before the thermo-photocatalytic reaction.

Compared with the activity of Bi-H x /NF in
As shown in Fig. R24 below, the CO 2 production activity stability of Bi-H x /NF in CO 2 + H 2 O + Ar atmosphere is similar with that in CO 2 + Ar. This result indicated that the introduction of H 2 O cannot keep the concentration of hydrogen constant in the Bi-H x system. This is easily understand based on the fact that the state of H in H 2 O is different from that in Bi-H x /NF. This is a good idea that need to be systematically studied in our future work. Action: Considering that the introduction of H 2 O induced negligible effect on the CO 2 reduction performance of Bi-H x /NF, this result was not included in the revised manuscript.

Dear Authors
Find here my concerns.
Response: Thanks very much for your feedback. We have addressed fully the comments to improve the quality of our manuscript.
1. First and foremost for any photocatalytic reaction needs to analyze the band gap of material which is not provided in manuscript.

Response:
The present Bi nanosheets are a special metal catalyst that has the surface plasmon resonance (SPR) property. For metal nanostructures with high free electron mobility, there is an inherent oscillation frequency of valence electrons against the restoring force of the positively charged nuclei. SPR is excited when the frequency of incident photon matches the natural oscillation frequency of valence electrons, leading to the coherent oscillation of electrons in energy and space.
From above discussion, we can conclude that SPR is a unique optical property based on the free valence electrons that is different from traditional semiconductor based on the excitation from band gap. Therefore, the band gap of Bi was not provided in manuscript. Response: Based on reviewer's suggestion, photoluminescence (PL) spectra of Bi-H x /NF and Bi/NF were obtained on a Fluorescence spectrophotometer (F-7000, Hitachi, Japan) at room temperature.

In
Steady-state PL spectra of the as-prepared samples were shown in Fig. R25 below. It can be seen that the PL intensity of Bi-H x /NF was significantly lower than that of Bi/NF, indicating a much lower recombination rate of charge carriers and a faster charge transfer within the sample Bi-H x /NF, relative to sample Bi/NF. This result strongly suggested that the charge separation efficiency of Bi/NF can be improved by the hydrogen storage process. To further study the separation of photoinduced electron-hole pairs under light illumination, steady-state photoluminescence (PL) analysis was performed on the two samples. As shown in Fig. 5e, the PL intensity of Bi-H x /NF is significantly decreased in compared with that of Bi/NF, indicating that the incorporation of H atoms on Bi/NF has effectively suppressed the radiation recombination of charge carriers, which is helpful for CO 2 reduction reaction on Bi-H x /NF. Fig 2b cyclic voltametric analysis nickel foam is not much sensitive toward electrocatalytic reaction thus the use of Nickel foam must be explained.

From
Response: Nickel foam, which has the advantages of high porosity, superior performance of fluid penetration, and good mechanical performance, is a suitable substrate for depositing Bi catalyst.

Action:
To address this question, the following sentences were added into the revised manuscript on Page 5 as: The NF was chosen as substrate for depositing Bi because of its good mechanical performance and high porosity with interconnected framework structure that is favourable for easy contact between electrolyte and electrode, and fast ion transport. Fig 2b, it can also clearly observed the absorption/desorption peak of hydrogen that also questioned the stability of Bi-H x /NF catalyst.

Response:
We thank the reviewer for this comment. The cyclic voltammetry (CV) curves in Fig. 2b were shown to prove the electrochemical hydrogen storage behavior of Bi/NF. The CV curve in Fig.   2b is from Bi/NF and not from Bi-H x /NF. In the present work, the Bi/NF catalyst was used for CO 2 reduction by reversible storage and release of hydrogen. Therefore, the stored H in Bi-H x /NF should be easily released. In this respect, an appropriate stability of Bi-H x /NF is needed. In addition, stability of Bi/NF in the overall CO 2 reduction process has been fully proved in Fig. 4d, in which no obvious decrease in CO 2 reduction activity was observed after 81 h of reaction. 5. Fig. 2c XPS analysis shows the significant shift of binding energy though other analysis though there is no extra oxidation state of Bi is reported thus the change must be explained.
Response: The significant shift of binding energy should be due to the electron donation of hydrogen atoms.
Action: To address this question, the following sentences were added into the revised manuscript on

Page 7 as:
However, for Bi-H x /NF, the peaks belonging to the oxidation states of Bi are obviously red-shifted relative to those for Bi/NF, which is related to incomplete oxidation of Bi due to the electron donation of hydrogen atoms.
6. GC analysis presented the production of CO and hydrogen molecule in the system while photocatalytic reduction does not selective that clearly give the information about formation of other products or hydrocarbons, it require some additional analysis.

Response:
The result on GC-MS also indicated that except for CO, no other products, such as CH 4 and CH 3 OH, were detected as shown in Fig. 4c.

Action:
To address this question, formation of other hydrocarbons was evaluated by GC, and GC-MS.
The following sentences for describing the result and the detailed analysis method used were added into the revised manuscript on Page 10, 12, and 25.
Other possible products (such as CH 4 , and CH 3 OH) were not detected, confirming that high selectivity for CO 2 reduction was achieved on Bi-H x /NF. Fig. 4c, except for the peaks belonging to CO 2 and CO, no other peaks were observed in Fig. 4c, further confirming the high selectivity for CO 2 reduction on Bi-H x /NF. This result is consistent with that of gas chromatography.

As shown in
For detecting CH 3 OH, the gaseous products from the reactor were analysed by an offline GC (Fuli Corp., China) equipped with a flame ionized detector (FID). The equipped column was TDX-01.
7. Comparative analysis data of 13 CO and CO is not provided in manuscript to prove the reliability of reaction mechanism needs to be rechecked.
Response: For comparison, the isotopic 12 CO 2 labelling experiment was carried out on Bi-H x /NF in the revised manuscript. As shown in Fig. R26 below, 12 CO with m/z = 28 is the main product for 12 CO 2 reduction, which further confirms that CO was produced from the thermo-photocatalytic reduction of CO 2 on Bi-H x /NF. Action: To address this question, the following sentences were added into the revised manuscript on

Page 12 as:
For comparison, the isotopic 12 CO 2 labelling experiment was also performed on Bi-H x /NF ( Supplementary Fig. 4), and 12 CO with m/z = 28 is found to be the main product for 12 CO 2 reduction, which further demonstrates that CO was produced from the thermo-photocatalytic reduction of CO 2 on Bi-H x /NF. 8. Dissolution of gas in the water requires some extra outer forces to maintain an equilibrium that is not explained in the manuscript need to be recheck and optimize the reactor conditions.

Response:
We thank the reviewer for this comment. This might be caused by our unclear description on the overall CO 2 reaction process. The overall CO 2 reaction process contains two steps. In the first step, splitting of H 2 O into H atoms and O 2 and storage of the H atoms in Bi nanoparticles (denoted as Bi-H x ) were simultaneously achieved by an electrochemical approach. In the second step, the obtained Bi-H x was used as a reducing equivalent to reduce CO 2 to CO by in situ generating H + /e − pairs under light irradiation. In fact, the two step reactions were proceeded in two separate reactors as described in the Methods part of the manuscript. The first reaction for electrochemical H storage on Bi was performed in an aqueous solution of KOH (1 M) electrolyte. After that, the formed Bi-H x /NF was used for CO 2 reduction in another gas-solid phase reactor. Therefore, no water was present in the CO 2 reduction process after the formation of Bi-H x .
Action: For clarity, the unclear sentences were changed in the revised manuscript on Page 10 and 11 as: The sentence "The CO 2 reduction performance of Bi-H x /NF was evaluated in a 300 mL closed chamber with a quartz window on top, in which both the temperature and light illumination could be controlled." in the original manuscript on Page 7 was changed to: After the electrochemical hydrogen storage process of Bi/NF in 1 M KOH electrolyte, the formed Bi-H x /NF was used for CO 2 reduction in another gas-solid reactor with a quartz window on top, in which both the temperature and light illumination could be controlled.
The sentence "As a control experiment, pristine NF was treated at a reduction potential of -1.2 V for 10 min in 1 M KOH electrolyte and placed in the same reactor for testing the CO 2 reduction activity." in the original manuscript on Page 8 was changed to: As a control experiment, pristine NF was treated at a reduction potential of -0.18 V for 10 min in 1 M KOH electrolyte and then placed in the gas-solid reactor for testing the thermo-photocatalytic CO 2 reduction activity.