Reverse oxygen spillover triggered by CO adsorption on Sn-doped Pt/TiO2 for low-temperature CO oxidation

The spillover of oxygen species is fundamentally important in redox reactions, but the spillover mechanism has been less understood compared to that of hydrogen spillover. Herein Sn is doped into TiO2 to activate low-temperature (<100 °C) reverse oxygen spillover in Pt/TiO2 catalyst, leading to CO oxidation activity much higher than that of most oxide-supported Pt catalysts. A combination of near-ambient-pressure X-ray photoelectron spectroscopy, in situ Raman/Infrared spectroscopies, and ab initio molecular dynamics simulations reveal that the reverse oxygen spillover is triggered by CO adsorption at Pt2+ sites, followed by bond cleavage of Ti-O-Sn moieties nearby and the appearance of Pt4+ species. The O in the catalytically indispensable Pt-O species is energetically more favourable to be originated from Ti-O-Sn. This work clearly depicts the interfacial chemistry of reverse oxygen spillover that is triggered by CO adsorption, and the understanding is helpful for the design of platinum/titania catalysts suitable for reactions of various reactants.

This manuscript by Li and coworkers address the mechanism of CO-induced oxygen spillover from the Sn-doped TiO2 to supported Pt NPs using several in-situ techniques and DFT-based theoretical analyses. I think this is a comprehensive work and publishable. However, this manuscript does not satisfy the quality criteria for publication in Nature Communications. The bottom line is that the oxygen chemistry at the Pt-TiO2 interfaces varies as Sn dopants are introduced. However, some vague experimental points make the authors' statements controversial.
Spillover of oxygen species from the supporting oxides to the supported metal NPs has always been considered in studying the catalytic functionality of the metal-oxide interfaces. Even in the cases of several early studies of CO oxidation catalyzed by oxide-supported metal NPs, reverse oxygen spillover has been considered a possible oxygen-supplying pathway. Moreover, the mechanism reported here by the authors is conceptually equivalent to oxygen taking over, which usually happens during the Marsvan Krevelen-type CO oxidation processes. I understand that the conceptual difference between the reverse oxygen spillover claimed by the authors with a conventional MvK type mechanism is whether the spillover oxygen can lay stable on the surface of Pt clusters or nanoparticles.
I am skeptical whether the presence of Pt4+ directly confirms that oxygen was moved from the Sndoped TiO2 to Pt clusters. It is well-known that Pt clusters supported on a stoichiometric TiO2 donate electrons to the support and that the Pt clusters draw electrons from the reduced TiO2. Moreover, although the oxygen vacancy appeared to be associated with oxygen spillover, bulk oxygen can heal this vacancy thoroughly. If the local coordinative ensemble at the Pt-TiO2 interfaces varies dynamically, I believe that the concentration of the Pt4+-like species may accordingly vary.
I am also not sure whether the surface of sub-nanometer-or nanometer (<2nm)-sized Pt clusters (experimentally observed ones) competitively binds oxygen rather than CO. This is presumably the key to whether the authors' claim can be rationalized or not. I am sure that the small Pt clusters strongly bind CO more than oxygen. In the cases of the larger Pt nanoparticles (than 2nm) in which micro-facets can develop at their surfaces, oxygen may share some portion of the surface with CO. However, it is hard to agree with the authors' claim that their small Pt clusters grep and stabilize oxygen at their surface. I agree that such oxygen can limitedly appear in the early stage of the reaction to which the catalyst is exposed to oxygen without CO. However, such oxygen species can be easily removed by CO (as presented in Fig. 5), and CO will preferentially occupy the surface of the Pt clusters. I am unsure whether an oxygen molecule heals the open position at stage v/VI (Fig. 5).
Moreover, I think assigning the PtO and Pt-O-Ti separately in XAS, or XPS spectra is risky because the local atomic ensemble at the metal-oxide is complicated, and dynamic electronic variation occurs during the reaction (as reported by Daelman et al. Nat. Mater. 2019,18, 1215. Fig. 5 shows the overall CO oxidation pathway following the Langmuir-Hinshellwood mechanism. This is, unfortunately, equivalent to the case of CO oxidation by Pt-oxide clusters, as an oxygen molecule is required to restore the initial Pt-oxide cluster. On page 16, "The energy barrier of CO oxidation by ROS over Pt/Sn0.2Ti0.8O2 was 0.69 eV, which was obviously lower than that of CO oxidation by Olatt (0.90 eV), suggesting that CO oxidation by the O atom transferred through ROS is more preferred over Pt/Sn0.2Ti0.8O2." I think this kind of statement made from the DFT calculations on flat TiO2 surfaces cannot be generalized. The barrier for the direct MvK mechanism will decrease and become sensitive to the surface morphology of TiO2, which is the case of the experimentally synthesized catalyst.
Reviewer #3 (Remarks to the Author): In this work, the authors synthesized Pt-based catalysts supported on SnxTi1-xO2, TiO2(anatase) and TiO2(rutile) supports. Catalyst structures were thoroughly characterized by electron microscopy, XRD, AP-XPS, DRIFTS and in-situ RAMAN. The catalytic activity was probed by CO oxidation, and it was found that Pt/SnxTi1-xO2 exhibited improved low temperature CO oxidation than Pt/TiO2(anatase and Pt/TiO2(rutile). Therefore, the authors devoted extended efforts to understand the nature of such improved low temperature activity promotion when Sn was used to dope TiO2. Overall, Pt particle size was similar across all samples and CO-TPR experiments revealed improved lattice oxygen availability on Pt/SnxTi1-xO2. AP-XPS and in-situ RAMAN were critical to identify that upon the exposure of Pt/SnxTi1-xO2 to CO+O2, increased Pt4+ species and weaker Ti-O-Sn bonds were detected, which suggests lattice oxygen from Ti-O-Sn could be mobilized towards Pt centers (process called Reverse Oxygen Spillover by the authors (ROS)). Computational chemistry calculations (AIMD) illustrated that ROS occurs on model Pt4O4/Ti-O-Sn and that lowest energy barrier CO oxidation pathway occurs via ROS rather than other routes. I recommend this article for publications only if the authors can satisfactorily address the following concerns, clarifications, and improvements to the manuscript. 1. Novelty Q1.1. Reverse spill over of oxygen has recently been reported in JACS (J. Am. Chem. Soc. 2023, 145, 2523−2531. Such publica on is crucially relevant for the present manuscript and should be included in the introduction. It also raises the question on the novelty of this manuscript. Authors should clearly clarify the novelty of this manuscript or what additional knowledge or novelty this manuscript brings over what already been reported in the JACS report. Q1.2. The authors should provide clear rationales on why Sn was selected to dope TiO2 instead of any other promoters. What are the reasons/hypothesis/preliminary work behind such selection? What is the generality of selecting a promoter such as Sn to achieve such performances. 2. Clarifications Q2.1. Line 86. The author should use the term USDRIVE, not USDRIVER. (Please see the following reference: https://www.energy.gov/eere/vehicles/us-drive) Q2.2. How were TOFs estimated? I could not find how the fraction of active Pt species was estimated for the TOF calculations. Please clarify and provide a detailed explanation of methods. Q3.3. Include detailed CO-TPR (shown in Figure 2e) experiment description in methods section. To assess the validity of author's claim about correlation between active species in the CO-TPR and O2-TPD experiments (shown in figure 2e and 2f) it is crucial for reviewers to understand exactly how the CO-TPR was carried out, including any pre-treatments with O2 or purge of weakly adsorbed O2 species. Since the main finding of this paper (Reverse Oxygen Spillover on Ti/SnxTi1-xO2) is based CO-TPR and O2-TPD experiment analysis (lines 222-226), I highly suggest that further clarification of experiments is provided before this work can be considered for publication. Q3.4. It would be highly appreciated if authors could indicate which Sn-O-Ti bonds are broken in CO oxidation cycle on the ROS route to facilitate understanding of reaction cycle schemes. 3. Improvements Q3.1. The discussion on why several reduction treatments were done to finally select 300 °C in H2 is not scientifically interesting/relevant to the main goal of the study (understanding the ROS effect) and should be placed in the SI section. Q3.
2. Indicate error bars in figures S5B and S5C (data set in Arrhenius plots) as well as in all figures containing TOF measurements. Q3.3. Lines 126. I recommend including a more elaborated explanation on why the observed CO and O2 partial orders are ~zero. The following work can help with that Angew.Chem. Int.Ed.2021,60,26054-26062. Q3.4. Lines 137-145. Is the observed sulfur tolerance expected for Pt-and TiO2-based materials? If SO2tolerance is not novel due to Sn doping (I don't think so), discussion of relevant literature is appropriate. (example: NATURE CATALYSIS | VOL 2 | JULY 2019 | 614-622)

Response to referees
Reviewer #1 (Remarks to the Author): In the manuscript by J. Chen et al., the experimental and theoretical results on the enhanced catalytic activity on Sn-doped Pt/TiO2 catalysts that were associated with the low temperature reverse oxygen spillover. A combination of near-ambient-pressure XPS, in situ Raman/Infrared spectroscopies, and ab initio molecular dynamics (AIMD) simulations revealed that the ROS was triggered by CO adsorption at Pt 2+ sites, followed by bond cleavage of Ti-O-Sn moieties nearby and the appearance of Pt 4+ species. I find the result addresses the role of the interfacial chemistry of ROS that is triggered by CO adsorption, which is an important subject in heterogeneous catalysis. However, there are many points that need to be clarified, also additional experimental results should be presented to support the key claim of the paper. Below are my detailed comments that need to be addressed.
Response: Thank you very much for your valuable comments, which allow us to further improve the manuscript. Hereinafter the one-by-one response to each specific comment as well as the revision accordingly. However, I think the Pt nanoparticles with the size of 1-2 nm should exhibit some metallic peaks, which were absent in the current investigation. I think the authors need to mention the earlier studies on the size dependence of Pt oxidation states and discuss why metallic Pt peaks are not visible.
Response: Thank you very much for your valuable suggestions. We have carefully read the literature you mentioned and conducted research on some related studies.
Previous studies have suggested a correlation between the oxidation state of surface Pt and the nanoparticle size of the catalyst. In general, the smaller the nanoparticle size of Pt clusters, the higher the oxidation state. For instance, research by Somorjai (Angew. Chem. Int. Edit. 47, 9212-9228 (2008)) has demonstrated that in Pt20 clusters (~0.8 nm), Pt is primarily present as Pt 2+ , while in Pt40 clusters (~1.5 nm), Pt is predominantly present as Pt 0 . However, the valence state of Pt clusters can also vary depending on the chemical environment, even when the particle size is the same.
(Langmuir 20, 2915-2920 (2004).) observed that the Pt 4f7/2 peak presented at 73.0 eV for Pt30 nanoparticles, suggesting that in this case, Pt is mainly present in the form of Pt 2+ .
In our study, the SAC-STEM HAADF images (Fig. 2d) revealed an average particle size of Pt nanoclusters on the surface of the Pt/Sn0.2Ti0.8O2 catalyst of approximately 1.2 nm, which is close to the particle size of Pt30 nanoclusters.
Considering the findings of the aforementioned literature, it is acceptable that the dominant valence state of Pt on the surface of the Pt/Sn0.2Ti0.8O2 catalyst is 2+. To further confirm the stability of Pt 2+ on the surface of the Pt/Sn0.2Ti0.8O2 catalyst, we recorded the XPS spectra of Pt 4f after H2 reduction at 300-500°C (Fig. S2b). The results showed that Pt 2+ can still exist stably on the surface of the Pt/Sn0.2Ti0.8O2 catalyst even after H2 reduction at 500 °C. The aforementioned literature review and discussion of the valence state of Pt in the catalyst of this study have been added to Supplementary Note 2, with details presented as follows: "Previous studies have suggested a correlation between the valence state of surface Pt and the nanoparticle size of the catalyst. In general, the smaller the nanoparticle size of Pt clusters, the higher the oxidation state. For instance, research by Somorjai S1 has demonstrated that in Pt20 clusters (~0.8 nm), Pt is primarily present as Pt 2+ , while in Pt40 clusters (~1.5 nm), Pt is predominantly present as Pt 0 . However, the valence state of Pt clusters can also vary depending on the chemical environment, even when the particle size is the same. Ozturk et al. S2 found that Pt 4f7/2 peaks were located at 74.6 eV (representing primarily Pt 4+ ) and 73.3 eV (representing primarily Pt 2+ ) for Pt40 nanoparticles deposited as a thick and a thin layer, respectively. Ye et al. S3 observed that the Pt 4f7/2 peak presented at 73.0 eV for Pt30 nanoparticles, suggesting that in this case, Pt is mainly present in the form of Pt 2+ . Therefore, in this study, XPS spectroscopy was utilized to investigate the oxidation state of surface Pt under different H2 reduction temperatures and before/after H2 reduction at 300 °C. As shown in Supplementary Fig. S2b, the Pt/Sn0.2Ti0.8O2 catalyst surface contained a small amount of Pt 4+ and a large amount of Pt 2+ before H2 reduction. After H2 reduction at 300 °C, Pt 4+ on the surface of Pt/Sn0.2Ti0.8O2 was reduced to Pt 2+ , and no further reduction from Pt 2+ to Pt 0 was observed even at an increased reduction temperature of 500 °C, indicating that Pt 2+ can remain stable on the surface of Pt/Sn0.2Ti0.8O2. Furthermore, after H2 reduction at 300 °C, Pt species on the surface of Pt/Sn0.2Ti0.8O2, Pt/TiO2-R, and Pt/TiO2-A catalysts were all present mainly in the form of Pt 2+ ( Supplementary Fig. S3)."

(c)
Binding energy (eV) Intensity (a.u.) Pt/TiO 2 -A Original caption indicated that XPS data of Pt4f spectra after the sample was exposed to O2, CO+O2, and CO at 100 °C. The way it is presented is so confusing.
The partial pressure and the types of gas during the near-ambient-pressure XPS, in situ Raman/Infrared spectroscopies, should be more clearly indicated.
Response: We sincerely apologize for any confusion caused by our previous description. We have revised the titles of Pt/TiO2-A was exposed to 1 mbar of O2, 0.5 mbar of CO with 0.5 mbar of O2, and 1 mbar of CO in turn. d-f, In situ Raman spectra recorded at 100 °C after (d) Pt/Sn0.2Ti0.8O2, (e) Pt/TiO2-R and (f) Pt/TiO2-A was exposed to 1% O2, 1% CO + 1% O2, and 1% CO in turn under ambient pressure. and (c) Pt/TiO2-A was exposed to 1% CO + 1% O2 at a designated temperature under ambient pressure. d-f, In situ NAP-XPS Pt 4f spectra recorded at 30-100 °C after (d) Pt/Sn0.2Ti0.8O2, (e) Pt/TiO2-R and (f) Pt/TiO2-A was exposed to 0.5 mbar of CO with 0.5 mbar of O2 at a designated temperature."

I think the temperature dependence of XPS measurement under CO and CO+O2
should be addressed in order to claim the low temperature reverse oxygen spillover. I am wondering if the authors can measure the XPS of Pt/Sn0.2Ti0.8O2 catalysts under gas conditions by changing the temperature to verify the evolution of Pt 2+ and Pt 4+ peaks.
Response: CO2 is generated when CO is solely introduced over Pt/Sn0.2Ti0.8O2 due to the consumption of active oxygen species on the surface (see Fig. 2e). Consequently, the chemical state of Pt on the surface of Pt/Sn0.2Ti0.8O2 catalyst is significantly affected by the prolonged introduction of CO alone, which has been demonstrated by in situ DRIFTS analysis, as depicted in Fig. 1 (for review only). The characteristic peak representing Pt δ+ -CO in the in situ DRIFTS spectra of Pt/Sn0.2Ti0.8O2 shifts to lower wavenumbers as the temperature rises during the introduction of CO alone. In contrast, the characteristic peak representing Pt δ+ -CO in Pt/Sn0.2Ti0.8O2 shifts to higher wavenumbers as the temperature rises during the introduction of CO+O2 (Fig. 4a).
The characteristic peak of Pt δ+ -CO is sensitive to various factors, such as Pt dispersion, Pt microstructure (i.e., steps, crystal planes, etc.), and charge transfer in the vicinity of the Pt sites (Nat. Catal. 2, 873-881 (2019)). Therefore, the prolonged introduction of CO alone significantly affects the chemical state of Pt on the surface of Pt/Sn0.2Ti0.8O2 catalyst. Consequently, we were unable to perform in situ NAP XPS analysis on Pt/Sn0.2Ti0.8O2 catalyst at different temperatures during the prolonged introduction of CO alone, as the active oxygen species of the catalyst continuously consumed.
However, simultaneous introduction of CO+O2 ensures dynamic stability of the Pt/Sn0.2Ti0.8O2 surface, thus we conducted in situ NAP-XPS analysis on Pt/Sn0.2Ti0.8O2, Pt/TiO2-R, and Pt/TiO2-A at different temperatures during the introduction of CO+O2, as shown in Fig. 4d-f. The results indicate that only Pt/Sn0.2Ti0.8O2 possessed the Pt 4+ species after the introduction of CO+O2, and the Pt 4+ content increased with the increase of reaction temperature, suggesting that ROS was more significant with the increase of CO oxidation rate over Pt/Sn0.2Ti0.8O2.

Fig. 4 | In situ DRIFTS and in situ NAP-XPS spectra of Pt/Sn0.2Ti0.8O2, Pt/TiO2-R and
Pt/TiO2-A recorded at 30-100 °C after exposure to CO+O2. a-c, In situ DRIFTS spectra recorded at 30-100 °C after (a) Pt/Sn0.2Ti0.8O2, (b) Pt/TiO2-R and (c) Pt/TiO2-A was exposed to 1% CO + 1% O2 at a designated temperature under ambient pressure. d-f, In situ NAP-XPS Pt 4f spectra recorded at 30-100 °C after (d) Pt/Sn0.2Ti0.8O2, (e) Pt/TiO2-R and (f) Pt/TiO2-A was exposed to 0.5 mbar of CO with 0.5 mbar of O2 at a designated temperature. 4. It is hard to follow the way of XPS deconvolution. It was mentioned that the peaks at ~72.2 eV and ~74.8 eV could be assigned to 4f7/2 and 4f5/2 signals of Pt 2+ species, whereas the peaks at ~74.4 eV and ~77.8 eV to 4f7/2 and 4f5/2 signals of Pt 4+ species, respectively. These two peaks of 4f7/2 and 4f5/2 signals should have a similar number of FWHM (full width of half maximum), but they are randomly changing (see Fig. S3, etc.). This point needs to be clarified.
Response: Thank you very much for your suggestion. Following your advice, we have re-deconvoluted all the XPS spectra. During this process, we strictly fixed the peak area ratio of Pt 4f5/2 to 4f7/2 at 3:4, and set the FWHM of all Pt 2+ and Pt 4+ peaks to be identical. The resulting XPS spectra, after the re-deconvolution process, are shown in      This manuscript by Li and coworkers address the mechanism of CO-induced oxygen spillover from the Sn-doped TiO2 to supported Pt NPs using several in-situ techniques and DFT-based theoretical analyses. I think this is a comprehensive work and publishable. However, this manuscript does not satisfy the quality criteria for publication in Nature Communications. The bottom line is that the oxygen chemistry at the Pt-TiO2 interfaces varies as Sn dopants are introduced. However, some vague experimental points make the authors' statements controversial.
Response: Thank you very much for your valuable comments, which allow us to further improve the manuscript. Hereinafter the one-by-one response to each specific comment as well as the revision accordingly.
Spillover of oxygen species from the supporting oxides to the supported metal NPs has always been considered in studying the catalytic functionality of the metal-oxide interfaces. Even in the cases of several early studies of CO oxidation catalyzed by oxide-supported metal NPs, reverse oxygen spillover has been considered a possible oxygen-supplying pathway. Moreover, the mechanism reported here by the authors is conceptually equivalent to oxygen taking over, which usually happens during the Mars-van Krevelen-type CO oxidation processes. I understand that the conceptual difference between the reverse oxygen spillover claimed by the authors with a conventional MvK type mechanism is whether the spillover oxygen can lay stable on the surface of Pt clusters or nanoparticles.
Response: Thank you for your careful review of our research. The CO catalytic oxidation process over Pt/TiO2 catalyst generally follows the traditional MvK mechanism, where CO is adsorbed on Pt sites and then oxidized to CO2 by lattice oxygen in the support. Similarly, CO oxidation on Pt/Sn0.2Ti0.8O2 catalyst follows a MvK-like reaction pathway. The primary difference is that after CO adsorbs on the Pt site, it does not directly react with the lattice oxygen in the support. Instead, the lattice oxygen in the support migrates to the Pt site, where it reacts with the adsorbed CO to form CO2. This migration process is known as reverse O spillover (ROS). In recent years, there has been a growing interest among researchers in studying the migration of oxygen on the catalyst surface. For instance, some related works on ROS are reviewed in Ref. ACS Catal. 2021, 11, 3159. Moreover, recent works by Hensen et al. (Refs. Nat. Catal. 2021, 4, 469 andNat. Catal. 2020, 3, 526) and Vayssilov et al. (Ref. Nat. Mater. 2011, 10, 310) have also mentioned the promoting effect of ROS on reactions over ceria related catalysts.
As for our manuscript, the principal innovation is that we modulated the rutile TiO2 by Sn doping to activate low-temperature (< 100 °C) ROS in Pt/TiO2 catalyst, and illustrated the rich interfacial chemistry of ROS from Sn-doped TiO2 (SnTiO2) to Pt sites in low-temperature CO oxidation with a combination of near-ambient-pressure XPS, in situ Raman/Infrared spectroscopies, and ab initio molecular dynamics (AIMD) simulations. We observed for the first time, to the best of our knowledge, the transformation of low-valent Pt 2+ to high-valent Pt 4+ with the presence of reducing gas CO, which indicates the CO-adsorption induced ROS on the catalyst. I am skeptical whether the presence of Pt 4+ directly confirms that oxygen was moved from the Sn-doped TiO2 to Pt clusters. It is well-known that Pt clusters supported on a stoichiometric TiO2 donate electrons to the support and that the Pt clusters draw electrons from the reduced TiO2. Moreover, although the oxygen vacancy appeared to be associated with oxygen spillover, bulk oxygen can heal this vacancy thoroughly. If the local coordinative ensemble at the Pt-TiO2 interfaces varies dynamically, I believe that the concentration of the Pt 4+ -like species may accordingly vary.
Response: We sincerely appreciate this remark from the reviewer, and find it especially helpful in improving the discussion of our manuscript. We would like to address our response as follows.
First of all, let's consider the well-known hydrogen spillover, the fundamentals of which have been well documented and widely applied in catalyst design. The hydrogen spillover process was depicted in Figure 1   Then, compared with the hydrogen spillover process that leads to the reduction of the support, the reverse O spillover (ROS) process, in which the lattice oxygen transfers from the support to the nearby active site, results in accordingly the oxidation of the active site. Besides the reverse spillover of oxygen, other possibilities that may increase the valence of Pt species was discussed. In the section of Discussion in revised manuscript, we discussed three possibilities of Pt 4+ generation only when Pt/Sn0.2Ti0.8O2 was exposed to CO or CO+O2. As shown below: "The in situ NAP-XPS spectra suggest that Pt 2+ was the major Pt species of Pt/Sn0.2Ti0.8O2, Pt/TiO2-R and Pt/TiO2-A after the moderate reduction, but only on Pt/Sn0.2Ti0.8O2 catalyst that the generation of Pt 4+ species was observed when it was exposed to CO or CO+O2. Three possibilities may lead to the formation of Pt 4+ on Pt/Sn0.2Ti0.8O2: (i) Pt 2+ captures gaseous O2 thus being oxidized into Pt 4+ ; (ii) Pt 2+ donates electron to the support, therefore resulting in Pt 4+ formation; (iii) lattice O in the support migrates to the vicinity of Pt 2+ (i.e. the ROS process), causing the oxidation of Pt 2+ to Pt 4+ . In situ NAP-XPS results (Fig. 3a) indicate that Pt 4+ was not observed when Pt/Sn0.2Ti0.8O2 was exposed to O2 alone, eliminating the first possibility. Regarding the second possibility, it is well-known that Pt clusters supported on stoichiometric TiO2 donate electrons to the support, and the Pt clusters draw electrons from the reduced TiO2. Therefore, when reducing gas CO is introduced, Pt should receive electrons from the support, resulting in lower Pt valence. However, NAP-XPS (Fig. 3a) shows that when reducing gas CO was introduced, Pt 2+ on the surface of Pt/Sn0.2Ti0.8O2 is oxidized to Pt 4+ , indicating that electron transfer between Pt and the support cannot result in the Pt 4+ formation. Therefore, the ROS process induced by CO adsorption should be the cause of the oxidation of Pt 2+ to Pt 4+ ." Hence, we believe that the Pt/Sn0.2Ti0.8O2 catalyst undergoes the process of O migration from the support to the active site (i.e. Chem. Soc. 2021, 143, 196) proposed that Pt-based catalysts primarily follow the MvK mechanism during the high-temperature oxidation of toluene. Meanwhile, the large size of the toluene molecule necessitates a significant supply of oxygen during the oxidation process, rendering the ROS process's contribution to the process negligible. We evaluated the catalytic performance of Pt/Sn0.2Ti0.8O2 and Pt/TiO2 catalysts in the toluene catalytic oxidation reaction, and the findings revealed that the Pt/Sn0.2Ti0.8O2 catalyst exhibited the lowest catalytic activity for toluene oxidation (see Figure 2 (for review only)), which indicates that the Pt/Sn0.2Ti0.8O2 catalyst has the lowest reactivity of lattice oxygen. Therefore, the results indirectly indicate that the reason for the Pt/Sn0.2Ti0.8O2 catalyst's superior performance in the CO catalytic oxidation process is due to the occurrence of the ROS process.   Pt/TiO2-R and (f) Pt/TiO2-A was exposed to 0.5 mbar of CO with 0.5 mbar of O2 at a designated temperature.

ROS) during the CO catalytic oxidation reaction.
I am also not sure whether the surface of sub-nanometer-or nanometer (<2nm)-sized Pt clusters (experimentally observed ones) competitively binds oxygen rather than CO. This is presumably the key to whether the authors' claim can be rationalized or not. I am sure that the small Pt clusters strongly bind CO more than oxygen. In the cases of the larger Pt nanoparticles (than 2nm) in which micro-facets can develop at their surfaces, oxygen may share some portion of the surface with CO. However, it is hard to agree with the authors' claim that their small Pt clusters grep and stabilize oxygen at their surface. I agree that such oxygen can limitedly appear in the early stage of the reaction to which the catalyst is exposed to oxygen without CO. However, such oxygen species can be easily removed by CO (as presented in Fig. 5), and CO will preferentially occupy the surface of the Pt clusters. I am unsure whether an oxygen molecule heals the open position at stage v/VI (Fig. 5).
Response: After carefully consideration of your comment, we agree with the notion that small Pt clusters exhibit a stronger affinity towards CO compared to oxygen. Our  Figure S23, where the adsorption energy of O2 on Pt/SnTiO2 and Pt/TiO2 catalyst surfaces is close to zero, indicating that O2 is unlikely to adsorb onto small Pt clusters.

DFT simulations also confirm this observation, as demonstrated in Supplementary
Conversely, CO is more likely to adsorb onto small Pt clusters (as evidenced by the energy changes in steps I→II and i→ii of Figure 5e).

Supplementary Fig. S1
Adsorption energy of O2 on Pt/Sn0.2Ti0.8O2 and Pt/TiO2-R. We concur with the notion that it is difficult to capture and stabilize oxygen on the surface of small Pt clusters. Fig. 3a (Fig. 3a), even without O2 supply. This oxidation process is facilitated by the migration of lattice O in the support to the Pt clusters (i.e. the ROS process). It is noteworthy that the ROS process is triggered by CO, which is subsequently oxidized by the O originating from the ROS process to produce CO2 (the generation of CO2 was evidenced by Fig. 2e). Thus, during CO oxidation, the small Pt clusters do not need to capture and stabilize gaseous oxygen on their surface. Gaseous O2 only need to heal the oxygen vacancy on the support, which is generated by ROS process.   which can adsorb the O2 reactants." Therefore, we believe that the process of "an oxygen molecule healing the open position at stage v/VI (Fig. 5e)" is reasonable.
Moreover, I think assigning the PtO and Pt-O-Ti separately in XAS, or XPS spectra is risky because the local atomic ensemble at the metal-oxide is complicated, and dynamic electronic variation occurs during the reaction (as reported by Daelman et al. Nat. Mater. 2019,18, 1215).
Response: In XPS spectra, we only assigned Pt species to Pt 2+ and Pt 4+ (Fig. 3a-c and Fig. 4d-f). In XAS spectra, the fitting results were listed in Table S4.   Fig. 5 shows the overall CO oxidation pathway following the Langmuir-Hinshelwood mechanism. This is, unfortunately, equivalent to the case of CO oxidation by Pt-oxide clusters, as an oxygen molecule is required to restore the initial Pt-oxide cluster.
Response: As described in Ref. Nat. Commun. 2015, 6, 8675: "The generally accepted by platinum is one of Langmuir-Hinshelwood (LH) mechanism requires the adsorption and reaction of molecular CO with atomic oxygen over metallic platinum surfaces". It suggests that Langmuir-Hinshelwood mechanism of CO catalytic oxidation is the reaction between adsorbed CO and adsorbed O2 molecule. As you mentioned, Fig. 5 shows that CO was oxidized by Pt-oxide clusters. This means CO was oxidized by lattice O, i.e. following the MvK mechanism (Ref. Nat. Commun. 2019, 10, 1358, which said that "It has also been observed that the relevant step during CO oxidation in a MvK reaction mechanism is the reaction between CO adsorbed on Pt and oxygen from the lattice"). Therefore, we believe Fig. 5 mainly shows the overall CO oxidation pathway following the MvK mechanism. On page 16, "The energy barrier of CO oxidation by ROS over Pt/Sn0.2Ti0.8O2 was 0.69 eV, which was obviously lower than that of CO oxidation by Olatt (0.90 eV), suggesting that CO oxidation by the O atom transferred through ROS is more preferred over Pt/Sn0.2Ti0.8O2." I think this kind of statement made from the DFT calculations on flat TiO2 surfaces cannot be generalized. The barrier for the direct MvK mechanism will decrease and become sensitive to the surface morphology of TiO2, which is the case of the experimentally synthesized catalyst.
Response: Thank you very much for your suggestions. Fig. 5e  restoring the catalyst to its initial state. Therefore, in this reaction cycle, we have considered the reaction pathway and energy changes of CO catalytic reaction on the catalyst surface with vacancies. Certainly, we understand that the catalyst model used in DFT simulations often differs from the actual situation, and DFT is also difficult to simulate all actual reaction processes. Therefore, we have revised this description to indicate that only in this DFT simulation, the result of a lower energy barrier for the ROS reaction pathway was obtained. The revised description is as follows: "The results of this DFT simulation show that the energy barrier of CO oxidation by ROS over Pt/Sn0.2Ti0.8O2 was 0.69 eV, which was lower than that of CO oxidation by Olatt (0.90 eV), suggesting that CO oxidation by the O atom transferred through ROS is probably more preferred over Pt/Sn0.2Ti0.8O2." Reviewer #3 (Remarks to the Author): In this work, the authors synthesized Pt-based catalysts supported on SnxTi1-xO2, TiO2(anatase) and TiO2(rutile) supports. Catalyst structures were thoroughly characterized by electron microscopy, XRD, AP-XPS, DRIFTS and in-situ RAMAN.
The catalytic activity was probed by CO oxidation, and it was found that Pt/SnxTi1-xO2 exhibited improved low temperature CO oxidation than Pt/TiO2(anatase and Pt/TiO2(rutile). Therefore, the authors devoted extended efforts to understand the nature of such improved low temperature activity promotion when Sn was used to dope TiO2. Overall, Response: Thank you for providing us with an important recently-published reference.
After careful examination, we found that the core content of this article is the synthesis of single-atom catalysts (SACs) Pt/CeO2 through treatment under O2 and N2 conditions. However, the CO catalytic oxidation activity of SAC Pt/CeO2 catalysts obtained through these two treatment methods differs significantly. Through Raman spectroscopy and computational studies, the authors revealed the distribution of various Pt1On-Ce δ+ species in each specific SACs, and found that the minority species of Pt1O4-Ce 3+ -Ov, accounting for only 14.2%, affords the highest site-specific reactivity for low-temperature CO oxidation among the other abundant counterparts, i.e., Pt1O4-Ce 4+ and Pt1O6-Ce 4+ . This work elucidates the quantitative distribution and dynamic transformation of varied single-atom species in a given SAC, offering a more intrinsic descriptor and quantitative measure to depict the inhomogeneity of SACs.
In this literature, the authors speculate that the formation of Pt1O4-Ce 3+ -Ov derived from the reverse O spillover (ROS) during the synthesis process, but there is little visual experimental evidence. The original text is described as follows: "However, based on our results, the lattice oxygen also promotes the oxidation and dispersion of Pt atoms, likely via reverse oxygen spillover. 12,17,24 This is further confirmed by the higher Ce 3+ /Cetotal value (20.2%) in Pt/CeO2-N600 than those of Pt/CeO2-O600 (15.3%) and Pt/CeO2-O800 (13.4%) (Figure 2c and Table S1). The absence of reduction peaks at low temperatures (269 and 275 °C for Pt/CeO2-N600 and Pt/CeO2-N800, respectively) indicates that the lattice oxygen at the interface are readily consumed during the nonoxidative dispersion by reverse spillover from Ce 4+ to Pt SACs 25-27 (Figure 2d)." As for our manuscript, we modulated the rutile TiO2 by Sn doping to activate low-temperature (< 100 °C) ROS in Pt/TiO2 catalyst, and illustrated the rich interfacial chemistry of ROS from Sn-doped TiO2 (SnTiO2) to Pt sites in low-temperature CO oxidation with a combination of near-ambient-pressure XPS, in situ Raman/Infrared spectroscopies, and ab initio molecular dynamics (AIMD) simulations. We observed for the first time, to the best of our knowledge, the transformation of low-valent Pt 2+ to high-valent Pt 4+ with the presence of reducing gas CO. In conclusion, the innovation of our manuscript is different from that of the JACS report.
Based on your suggestion, we have included this article in our Introduction section.
As shown below: " Response: Thank you for bringing the logical flaw in our Introduction to our attention.
We selected Sn as dopant because SnO2 has similar crystal structure as rutile TiO2.
Therefore, doping Sn will not significantly change the bulk structure but alter the oxygen symmetry in TiO2. It is well-documented that the creation of asymmetric oxygen will increases its mobility and thus benefits to the ROS process. In the revised manuscript, we describe the rationale for selecting Sn-doped TiO2 as the support in the Introduction as follows: "Since SnO2 possess a similar structure as rutile TiO2 (ref "Transient CO oxidation was tested in a fixed-bed quartz micro-reactor. The reduced catalysts were loaded into the reactor and heated to the reaction temperature under N2 without any further pretreatment. The reaction temperature was fixed at 100 or 200 °C with a CO concentration of 1% and N2 as the balance gas, without the supply of O2. The gas flow rate was set to 100 mL·min -1 with a GHSV of 60,000 mL gcat -1 h -1 . The infrared gas analyzer (Gasmet Dx-4000) was utilized to measure the concentrations of CO and CO2 in both the inlet and outlet streams." 3. Improvements Q3.1. The discussion on why several reduction treatments were done to finally select 300 °C in H2 is not scientifically interesting/relevant to the main goal of the study (understanding the ROS effect) and should be placed in the SI section.
Response: Thank you for your suggestion. The discussion regarding to reduction treatments was placed into the Supplementary Note 2 in Supporting Information. As shown below: "The H2 pretreatment process utilized a 5% H2 concentration and a treatment time