Thermally stable single atom Pt/m-Al2O3 for selective hydrogenation and CO oxidation

Single-atom metal catalysts offer a promising way to utilize precious noble metal elements more effectively, provided that they are catalytically active and sufficiently stable. Herein, we report a synthetic strategy for Pt single-atom catalysts with outstanding stability in several reactions under demanding conditions. The Pt atoms are firmly anchored in the internal surface of mesoporous Al2O3, likely stabilized by coordinatively unsaturated pentahedral Al3+ centres. The catalyst keeps its structural integrity and excellent performance for the selective hydrogenation of 1,3-butadiene after exposure to a reductive atmosphere at 200 °C for 24 h. Compared to commercial Pt nanoparticle catalyst on Al2O3 and control samples, this system exhibits significantly enhanced stability and performance for n-hexane hydro-reforming at 550 °C for 48 h, although agglomeration of Pt single-atoms into clusters is observed after reaction. In CO oxidation, the Pt single-atom identity was fully maintained after 60 cycles between 100 and 400 °C over a one-month period.

1. It is argued that the Al3+ centers are critical binding sites for the Pt atoms. However, no Pt-Al interactions were observed or discussed on the XAS data. Some discussion on this should be added.
2. The IR interpretation should be further substantiated. a. For Pt/Al2O3, CO stretching frequencies below 2100 cm-1 have been rigorously assigned to metallic adsorption sites on nanoparticle surfaces at either well coordinated (~2090 cm-1) or undercoordinated (~2070 cm-1) Pt sites (see ACS Catal. 2016, 6, 5599−5609). It is surprising that CO adsorbed to the cationic Pt single atoms don't exhibit stretching frequencies > 2100 cm-1. It is known that both Pt structure and oxidation state play a role in defining the CO stretching frequency, so this should be discussed and somehow rationalized. See J. OF CATALYSIS 169, 382-388 (1997) for a nice discussion on some of these issues. b. The CO band observed in Figures 2F and 4E are missing some signatures of adsorption at single Pt sites. First, the band positions seem to shift with coverage, suggesting dipole interactions that shouldn't exist at single atom adsorption sites. Second, the bands are quite wide in FWHM and have some obvious asymmetry that suggests there are multiple adsorption sites being observed that have different chemical characteristics. c. In the recent Science papers by Stair's group and Datye's groups CO was observed to stick very strongly for stretches that were assigned as single Pt atom adsorption sites. This does not agree with the results presented here. This should be discussed. 3. It is mentioned that the reactivity of the 0.2 Pt-m/Al2O3 actually increases with time on stream. This is suggesting there may be some agglomeration of Pt atoms to form small metallic clusters that are more active for this reaction. Some suggestions of what structural or chemical changes may be occurring to the catalyst that induce this reactivity change should be added.
where no ordered porosity is seen in Fig. 2 or in S8. Furthermore, drawing the circles around atoms with such low contrast (see fig. S8) is questionable with the amorphous background of the alumina showing enough random bright spots.
6) An example of over interpretation or imprecise characterization is on line 165 of page 7. What is 'fast' desorption of CO? When I study their FTIR spectra, I see the bands being fairly resistant to desorption in flowing N2 at 30 C. The CO desorbs only after heating to 100 C. This is true of all of their spectra, which means the CO is quite strongly bound, unlike what they state in the text. 7) On line 171 page 7, they refer to 'embedded' morphology, another ill-defined term. How do they know the atoms are embedded and not on the surface? And if they were embedded, they might not be catalytically active. For this reason they need to report turnover frequency (TOF). They should base this per atom of Pt, since in this manner their catalyst reactivity can be compared with those of other workers. As it stands, they only show lightoff curves which depend on heat and mass transfer effects and do not represent kinetics. They need to report TOF at low conversions so they are free from mass and heat transfer limitations. 8) On line 197 page 8 they discuss how the XPS spectra indicate that the Pt species are mainly dispersed on the exterior domains of the p-alumina. I fail to see how XPS can show this directly. The XPS samples a certain depth of their samples, depending on the energy of the x-rays used and the specific photoelectron being analyzed. And the severe overlap with the Al peaks makes the interpretation of the XPS (figure S 23 and S 24) questionable.
9) The EXAFS and XANES are convincing that their samples show absence of Pt-Pt coordination. But these are air exposed samples, where the Pt is exposed to air. I did not see any in-situ XPS demonstrating the nature of the working catalyst. For example, Figure 4 shows the EXAFS of the 0.2 m-alumina sample after CO oxidation, but I am not sure if the sample was exposed to air during transfer to the EXAFS? 10) Ultimately the evidence for the single atom nature of the 0.2 Pt/m-alumina lies in the CO FTIR which shows that the band position is different from that of 0.2 Pt/p-alumina. But it is a difference in degree, which means the same bands are present but in different proportions. So, what I see is that the authors have a sample that is kinetically trapped into a state where its high dispersion is preserved (due to the low loading). But this is a metastable state, since we see some larger clusters in this sample. This means that if the Pt atoms come into contact with each other, they will grow to form clusters. Which is why their sample retains its characteristics only at low loadings (as they admit in line 132 on page 5. 11) Despite the high dispersion of the Pt/m-alumina, the reactivity is comparable to the other catalysts. This is why they need to show TOFs to establish whether the reactivity is truly superior. For the other probe reaction, Fig. 5g, I noticed that the m-alumina catalyst is actually lower in reactivity than the commercial sample. And the high selectivity ( Fig. 5a) is only seen at near zero conversion. The catalysts need to be compared at similar conversion.
Reference 45: the title of the paper, "eeposition" should be "deposition".
The manuscript by Zhang et al. reports on the preparation and application of a new stable versatile single-atom Pt/γ-Al 2 O 3 catalyst for three representative reactions, under either oxidative or reducing gas mixtures at elevated temperatures up to 550 o C. The advantages of the catalyst were attributed to the clear single-atom feature of the platinum cations that were stabilized by the alumina lattice oxygen in a square-planar structure. Moreover, the embedded platinum atoms in the alumina substrate improved the thermal stability of the overall support texture. The work is a joint effort from multiple groups, and a formidable amount of raw/processed data has been included to show various properties of the new catalyst. Overall, the paper presents a compelling case for the stabilization of single Pt atoms in alumina to high temperatures, if the material is prepared as the authors demonstrate, and this is an important new piece of work in the area of single-atom supported metal catalysts. However, the data interpretation is confusing in several places throughout the paper. To meet the standards of Nature Communications, more work is needed to address in greater depth the (rather phenomenological) presentation of the experimental findings. A major revision with re-evaluation is thus recommended.
My detailed comments are as follows: 1) The materials investigated do not reflect the industrial significance advocated in the introduction. Indeed, the autocatalyst uses a lot of platinum every year as the authors cited, but NOT as Pt/Al 2 O 3 catalysts. The formulation of Pt/Al 2 O 3 has long been phased out from the mainstream three-way catalysts that are being used in conventional gasoline engine emission controls. Platinum is too volatile to survive the high-temperature aging, and palladium (major) and rhodium (minor) are the dominant precious metals being used. For diesel and other leanburn gasoline engine emissions, platinum has been used widely, but again not as Pt/Al 2 O 3 catalysts. The Pt-Pd alloy catalysts are now being widely used worldwide for this purpose. The alloy catalysts generally have a multifold better performance than the Pt-only catalysts for CO and HCs oxidation. Therefore, the catalyst development from the current work does not go into the heart of the technical challenge that we are facing today. The discussion in this section could be changed to reflect the practice more accurately.
2) According to the EXAFS results, it is claimed that the single-atom centric Pt-O4 planar structure will fully survive the H2 reduction at 400 oC, and that this is indeed the versatile catalytic center for various reactions. However, if one compares Fig. 3D with Fig. 4F, a clear decrease of the Pt-O peak intensity in the R-space of EXAFS data can be observed for the same "best" 0.2Pt/m-Al2O3-H2 catalyst. Doesn't this indicate the evolution of the catalytic center? Along the same line, why does the "unreducible" Pt-O4 planar structure in the 0.2Pt/m-Al2O3 catalyst give strong reduction peaks in H2-TPR (Fig. S2)? What are the reducible species? Is adsorbed oxygen an issue even after the 50 oC-pretreatment in H2? These experiments should be presented as 2 nd or 3 rd cycle, without exposure to oxygen between the cycles. If real, how do the reducible species relate to the catalytic activity? Why do the other higher Pt loaded Pt/m-Al2O3 catalyst (comprising clusters plus atoms) prepared by the same method give weaker reduction peaks? If the Pt-O coordination was intact before and after reduction, why do the 0.2Pt/m-Al2O3-H2 and 0.2Pt/m-Al2O3-O2 catalysts behave differently in CO oxidation? In Ref. 42, where the authors cited the DFT results to explain the reaction mechanism, the alumina support was found not the part of the CO oxidation catalysis. There is a critical inconsistency here, which the authors must address. The authors should also analyze the Pt edge in post-reaction catalyst by EXAFS for all the reactions to confirm the stability and the coordination structure proposed according to the as-received sample.
3) For the CO oxidation reaction, a rather high contact time has been used (100 mg sample, 80 ml/min flow). Firstly, this does not well reflect the real application (suggest: 100 mg sample, 1-5 L/min flow). The more serious concern is that the key catalyst--0.2Pt/m-Al2O3 has minimal activity below 200 oC in repeated cycles, even at such a high contact time. This is the same issue encountered in Ref.  Fig. 10,11). This makes me wonder if the method of preparation used here is really limited to very low Pt loadings. This is a drawback that must be made clear in the paper. On the discussion that follows still on p.8, another control sample for the authors to consider would be the addition of Pt by incipient wetness impregnation on their m-Al2O3. Can this Pt "anchor" on the special Al(3+) pentacoordinated sites? How much Pt can thus be stabilized? The m-Al2O3 has very high content (1/3) of these special sites.. 6) How do the findings link to the classical debate of the structure sensitivity found in Pt/Al2O3 catalysts for CO oxidation? Are there two mechanisms for the reaction? One at low-temperature and one at high-temperature. How about kinetic measurements over the two different structures of Pt catalysts presented here? As for the Pt dispersion, measured here by H2-O2 titration and CO pulse chemisorption (never saturated?! Fig. S7), the authors do not have a table to summarize these results, and never use the standard term "dispersion" to describe these findings. I did a few calculations myself, and found that the relation between the kinetic rate and the total exposed Pt surfaces of the m-, p-, and commercial alumina supported samples does not have a clear trend. The authors need to ponder this important issue. 7) For the given CO oxidation and calcination temperatures up to 400 oC, I could not judge if the single-atom platinum is more stable than the alkali ion-stabilized single platinum atoms reported in J. Am. Chem. Soc. 137, 3470-3473 (2015). To demonstrate the point that the single atoms in alumina prepared as shown here do have a superior stability, the main premise of this paper, the samples treated at 600 oC should be fully analyzed to show the exclusive presence of single atoms, and the activity should also be reported. The results for the samples treated at 800 oC are actually puzzling to me. The authors need to be aware that the platinum may have already vaporized at this temperature (see Science 353, 150-154 (2016), and multiple other reports discussing Pt catalysts deactivation). How much Pt remain in the alumina after the 800 oC-treatment? 8) What is reason that leads to the incremental (not breakthrough) improvement for the hexane reactions? Again, the total amount of the exposed Pt surfaces in various samples that I estimated from the H2-O2 titration results could not explain the initial reaction rates. Does the chemical valence (metal vs. cation) matter to the hydrogenation and reforming reactions? Will the isolated metal Pt atom stabilized in an alternative substrate be a better solution? As in other places in the paper, the authors report good activity, but do not get to the heart of the chemistry involved. Potentially, the section on hexane can be removed, and published separately in a specialized journal. Already, the paper has too many figures, and parts for a Communications paper.
9) The stability of the m-Al2O3 seems to be poorer compared to many conventional alumina supports, although the Pt doping is found to stabilize the alumina framework. I suggest adding another reference sample by using the m-Al2O3 and loading the platinum by incipient wetness impregnation method. The commercial Pt/Al2O3 sample may have too many differences besides the state of Pt to make fair comparisons with the in-house prepared Pt catalysts. For example, the unusual light-off curve shape for the commercial Pt/Al2O3 sample at high conversions may due to the mass transfer related issues rather than the intrinsically modified chemistry.
10) "The strong metal support interaction" has its unique original meaning in catalysis beyond the expression of "the interaction between the metal and support is strong". The authors need to be cautious about this and modify their wording in the section of discussion. 11) Finally, the title of the paper should be more specific and spell out the Pt/m-Al2O3 being investigated. A more appropriate title would be: Thermally stable single atom Pt/m-Al2O3 for CO oxidation and the selective hydrogenation of 1,3 butadiene Response: We agree that many factors, including the electronic state, particle size, coordination environment, and metal-support interaction, will influence the CO absorption characteristics. Thus, CO-DRIFT alone could not provide an unambiguous picture of the metal species on the support. The peak assignments varied from case to case. For example, 0.18 wt% Pt/θ-Al 2 O 3 single atom catalyst exhibits two CO adsorption peaks at room temperature: 2108 cm -1 and 2056 cm -1 (J. Am. Chem. Soc. 2013, 135, 12634-12645), the latter of which has been assigned as CO adsorbed at Pt(0) single-atom sites.
This is different from both the example mentioned by the reviewer and our case.
In addition, CO absorbed to positively charged Pt species does not necessarily exhibit stretching In literature, a shift of CO adsorption band as a function of coverage is well-known for nanoparticles.

2011, 3, 634-641).
We propose the Pt 1 species in our catalyst are not identical, which may stay on slightly different coordination sites on the support. This leads to relatively wide FWHM and induces asymmetry to the CO adsorption peak. It also results in the shift of CO adsorption band when coverage changes.
We have added an auxiliary line to indicate the slight shift of CO adsorption peak as coverage decreases (see below), and the some text in the MS (see Response to c).
c. In the recent Science papers by Stair's group and Datye's groups CO was observed to stick very strongly for stretches that were assigned as single Pt atom adsorption sites. This does not agree with the results presented here. This should be discussed.  We thank the reviewer for providing excellent comments related to DRIFT-IR spectra. 3. It is mentioned that the reactivity of the 0.2Pt-m/Al 2 O 3 actually increases with time on stream.
This is suggesting there may be some agglomeration of Pt atoms to form small metallic clusters that are more active for this reaction. Some suggestions of what structural or chemical changes may be occurring to the catalyst that induce this reactivity change should be added.

Response:
The  preventing the Pt species from forming larger particles. At higher concentrations, the Pt tends to form larger particles. And it appears that the reactivity of these catalysts is not superior to that of the conventional catalysts. The work is not suitable for publication in its present form and needs major revisions to correct the errors in the interpretation of the data, as listed below. And the reactivity comparisons for CO oxidation need TOF data and for the selective hydrogenation, need to be compared at similar conversion.
Specific comments 1) The authors have synthesized mesoporous alumina (m-alumina), and then another one they call porous alumina (p-alumina). Since alumina tends to crystallize when heated, I would like to see evidence that the mesoporous alumina structure is preserved after their high temperature treatments. Supplementary figure 7 shows ordered structure in the as-prepared state. Figure S8 and S9 shows them after treatments, but the temperature is not stated, and the samples do not exhibit any order. Hence, the authors need to state whether or not the pore structure is retained.
Otherwise, I would infer that the m-and p-aluminas are not probably very different after being heat treated. Response: The samples shown in Figure S7, S8 and S9 were 0.2Pt/m-Al 2 O 3 -O 2 and 0.2Pt/m-Al 2 O 3 -H 2 , both of which were treated at 400 °C. We did not mention the temperature because "400 °C" is the "standard" treating temperature in this study, while any sample treated at a different temperature is clearly labelled. The HAADF-STEM images in Figure S8 and S9 were taking from ultra-thin areas at very high magnification to observe single Pt atoms, at which condition the ordered mesoporous structure of the support could not be seen clearly.
In fact, m-Al 2 O 3 preserves its porous structure much better than p-Al 2 O 3 in heating condition. From BET analysis (provided in Figure S26c,e), m-Al 2 O 3 with 0.2 wt% Pt can maintain its surface area (> 250 m 2 /g) after 600 °C and 800 °C treatment. In sharp contrast, the BET surface area of 0.2 wt% Pt on p-Al 2 O 3 decreased monotonically with increased treating temperature, dropping from 265 m 2 /g (treated at 400 °C) to 114 m 2 /g (treated at 600 °C), and further to 61 m 2 /g (treated at 800 °C). We also provided TEM images for the 0.2Pt/m-Al 2 O 3 -O 2 and 0.2Pt/p-Al 2 O 3 -O 2 after 600 and 800 °C treatment in Figure S27, further demonstrating the ordered structure of the 0.2Pt/m-Al 2 O 3 -O 2 was well preserved.
2) The N 2 adsorption isotherms in Fig. S5 do not suggest a broad pore size distribution, not what one would expect based on the TEM images (which of course sample very small regions of the specimen).
Hence, the authors need to show the low angle XRD region which will show clearly the extent of ordered mesoporosity in their structure. This is important since the authors claim that their samples are stable under extreme conditions and the m-alumina is better, so establishing the stability of the alumina pore structure is important.

Response:
We have conducted small-angle XRD analysis on both 0.2Pt/m-Al 2 O 3 catalyst (treated at 400 °C, 600 °C, and 800 °C, respectively) and 0.2Pt/p-Al 2 O 3 control sample (treated at 400 °C), which provided strong evidence of the presence of hexagonally ordered mesopores (p6mm symmetry) for 0.2Pt/m-Al 2 O 3 samples ( Figure below). The XRD patterns show two reflections, (100) and (110) 3) The manuscript should omit reporting data from the literature, for example figure S1 on utilization of Pt and Figure 1 in the manuscript, which is not reporting any original data from this study.

Response:
We have removed Figure 1a, 1b and Figure S1 in the revised MS and SI. A reference is given to Johnson Matthey's report reflecting the production and demand of Pt in recent years. Fig. 2 are incorrect. First, to image single atoms of Pt you need a probe diameter of sub-Angstrom size. Each bright dot will then correspond to the size of the probe. I cannot tell from the scale of the image if each of those dots meets this criterion. I do not understand the Gaussian fit of image g, which should show a single peak corresponding to the size of the probe (since the atom is much smaller in size). I cannot read the scale very clearly in this rather fuzzy image, but establishing this is important to convince the reader that they truly have isolated single atoms.

4) The interpretation and analysis of the images in
Response: Yes, the observed STEM intensity is a convolution of the probe function and object function, and the probe size is a decisive factor for the image resolution. In this study, a probe of ~ 1 Å was used and the FWHM of the Gaussian fitted peaks for single atoms are around 1.  Fig. 2e is incorrect in asserting this to be the interparticle distance, since the STEM image in Figure 2 is a projection of the 3-D sample. Hence what you measure as the interatom distance is not really the true distance, it is a projection of this distance on to a plane. It is clear that the bright dots don't have the same contrast, so they are not in one plane. And to make any inferences from the interatomic distances in the cluster seen in figure 2g is simply incorrect, since metal clusters tend to fragment and fall apart when subjected to the intense electron beam dose that is used to generate images such as the STEM images shown in figure 2 d-g. This is one reason where no ordered porosity is seen in Fig. 2  Response:

5) Secondly,
We thank the reviewer for raising insightful comments. We totally agree that strictly speaking, the measured "inter-atom" distances here are projected distances.
We did not distinguish "projected distance" from the true, 3-D "inter-atom distances" in our original submission because of the following considerations: i) The HAADF-STEM images were taken from ultra-thin areas in the specimen to gain reasonable contrast of single atoms. Therefore, the difference in height between a single atom and its nearest neighbors is insignificant; and ii) HAADF-STEM with a large convergent angle (~ 27 mrad used here) exhibits high depth-sensitivity and this phenomenon is pronounced for single heavy atoms on a light support (PNAS 2006, 103, 3044-3048).
The contrast of a single atom would decrease significantly upon a height change of only a few Ångstroms and completely blur out when larger height variation is introduced (Appl. Phys. Lett. 2005, 87, 034104). Therefore, the identified single atoms under a given focus condition should not differ much in height. Taken together, we thought that it would not bring much error to approximate the "projected distances" as "inter-atom distances". In the revised manuscript, we follow the reviewer's comments to make our statements more precise as follows. Note that the distances between neighboring single atoms measured from HAADF-STEM are "projected distances". The true threedimensional "inter-atom distances" should be larger. However, given that the HAADF-STEM images were taken from ultra-thin areas in the specimen to gain reasonable contrast of single atoms and that the image contrast is very sensitive to the vertical position of single atoms, the identified single atoms should not differ much in height. Therefore, it would not bring much error to approximate the "projected distances" as "inter-atom distances" in this specific case. This paragraph of clarification has been provided in the caption of Supplementary Fig. 9, on page S10.
The reviewer is correct that most single atoms and small clusters exhibit considerable degree of structural dynamics upon the scanning of electron probe and that the cluster image in Fig. 2f ~ h should be more appropriately regarded as a "snapshot" of the dynamical structure. It is difficult to decide whether the observed loosely-packed cluster is an intrinsic feature or induced by the electron beam. According to this comment, we revised the corresponding statement as follows: "Apart from the dominant amount of isolated atoms, a few small clusters that exhibit considerable structural dynamics under electron beam were also observed. A snapshot image (Fig. 2g, h) shows that the cluster has loosely-packed atoms with the interatomic distances longer than those observed in metallic Pt. 35 However, it is difficult to affirm whether these small clusters are formed by loose packing of single Pt atoms, or by electron beam-induced fragmentation of the close packed structure." (MS Page 4).
We thank the reviewer for the comments on the identification of single atoms. In this study, we identified Pt single atoms in the HAADF-STEM image by searching the local intensity maxima using the algorithm reported by I. F. Sbalzarini etc. (J. Struct. Biol. 2005, 151, 182-195). This method is not so sensitive to the contrast variation of substrates and works well for locating single atoms. The positions of single atoms are further refined and their coordinates are extracted for the nearest neighbor distance analysis. Specifically, we found that by using a aperture size of ~ 1.6 Å, a cut-off score of 0 for the non-particle discrimination, and an accepted percentile of 0.01 ~ 0.05, single atoms can be well identified in most HRSTEM images. Those located and refined single atoms are carefully inspected to avoid spurious identification before proceeding to the distance analysis. The detailed method for identifying single atoms from HAADF-STEM images has been added in the method section in SI (Page S3).
6) An example of over interpretation or imprecise characterization is on line 165 of page 7. What is 'fast' desorption of CO? When I study their FTIR spectra, I see the bands being fairly resistant to desorption in flowing N 2 at 30 C. The CO desorbs only after heating to 100 C. This is true of all of their spectra, which means the CO is quite strongly bound, unlike what they state in the text. Response: This is a nice comment. The CO adsorption on our catalyst is not particularly strong compared to some earlier Pt single-atom systems. For instance, no decrease of CO adsorption band was observed on 0.5Pt/HZSM-5 catalyst, even when the temperature increased to 100 °C for 15 min (Science, 2015, 350, 189-192 7) On line 171 page 7, they refer to 'embedded' morphology, another ill-defined term. How do they know the atoms are embedded and not on the surface? And if they were embedded, they might not be catalytically active. For this reason they need to report turnover frequency (TOF). They should base this per atom of Pt, since in this manner their catalyst reactivity can be compared with those of other workers. As it stands, they only show lightoff curves which depend on heat and mass transfer effects and do not represent kinetics. They need to report TOF at low conversions so they are free from mass and heat transfer limitations.

Response:
We thank the reviewer for raising insightful comments. The word "embedded" has been removed in the revised manuscript.
We also agree that TOF is an important parameter to benchmark catalyst performances. CO oxidation over 0.2Pt/m-Al 2 O 3 -H 2 catalyst has been carefully measured between 180 and 250 °C, with 5 °C increment each step. The conversion of CO was kept low in the entire region to get rid of mass transfer limitations. TOF was 0.010 s -1 at 180 °C and steadily increased to 0.17 s -1 at 250 °C. There were five papers using Pt single-atom catalysts for CO oxidation, and our catalyst is comparable with most values. We have compiled these TOF data and related catalyst/reaction information in Supplementary

Response:
The reviewer has raised an excellent point. Indeed, the low Pt loading and the significant overlap with Al signal made XPS interpretation unconvincing. We have removed XPS spectra (original S23 and S24), experimental details and discussions in the revised MS and SI.
9) The EXAFS and XANES are convincing that their samples show absence of Pt-Pt coordination. But these are air exposed samples, where the Pt is exposed to air. I did not see any in-situ XPS demonstrating the nature of the working catalyst. For example, Figure 4 shows the EXAFS of the 0.2 m-alumina sample after CO oxidation, but I am not sure if the sample was exposed to air during transfer to the EXAFS? Response: XAS analysis was conducted in Japan so the sample was inevitably exposed to air during sample transfer and mailing. This may not be a big concern, however, due to the following reasons. First, the CO oxidation reaction was conducted under O 2 -rich atmosphere, so that the catalyst was always under oxidative conditions (the same as when exposed to air). Secondly, our catalyst is stable with O 2 in air. We have conducted CO-DRIFT study both in-situ and ex-situ, since the CO adsorption peak is sensitive to electronic and geometric characteristics of Pt species. For the in-situ measurement, the sample was pre-reduced with H 2 at 400 °C before CO adsorption (Figure 3f), whereas the ex-situ measurement was done with the sample transferred from the reactor to the DRIFT cell (Supplementary Figure 16a). The CO adsorption peaks were identical, suggesting exposure to air did not induce appreciable changes to Pt on m-Al 2 O 3 .
On the other hand, we fully agree with the reviewer that in-situ XAS technique would provide direct information concerning the key features of working catalyst. The collaboration between NUS team in Singapore and Kyoto U team Japan is continuing, and we have prepared and submitted a proposal to request beamtime for in-situ XAS analysis in the future.
10) Ultimately the evidence for the single atom nature of the 0.2 Pt/m-alumina lies in the CO FTIR which shows that the band position is different from that of 0.2 Pt/p-alumina. But it is a difference in degree, which means the same bands are present but in different proportions. So, what I see is that the authors have a sample that is kinetically trapped into a state where its high dispersion is preserved (due to the low loading). But this is a metastable state, since we see some larger clusters in this sample. This means that if the Pt atoms come into contact with each other, they will grow to form clusters. Which is why their sample retains its characteristics only at low loadings (as they admit in line 132 on page 5). Response: p-Al 2 O 3 indeed provided a portion of Pt single-atoms in the fresh catalyst. However, its initial catalytic performance (both activity and selectivity) in all three probe-reactions were lower than Pt supported on m-Al 2 O 3 . Moreover, the long-term stability of Pt/p-Al 2 O 3 were far worse than Pt/m- In the area of single-atom catalysis, the ultimate goal may not be synthesizing thermodynamically stable atoms, since these species are likely to be catalytically less active or inactive. A balance between stability and the catalytic activity is important. In our study, a system in which Pt single atom species exhibit high stability without comprising activity has been developed, and therefore holds potential to be expanded into a broad range of single-atom catalysts with improved stability.

11) Despite the high dispersion of the Pt/m-alumina, the reactivity is comparable to the other
catalysts. This is why they need to show TOFs to establish whether the reactivity is truly superior. For the other probe reaction, Fig. 5g, I noticed that the m-alumina catalyst is actually lower in reactivity than the commercial sample. And the high selectivity (Fig. 5a) is only seen at near zero conversion.
The catalysts need to be compared at similar conversion.

Response:
We agree with the first part of the comment. TOF values have been provided and compared with literature data in the revised submission (see Response to Comment 7).
Concerning the second part of the comment, there seems to be a misunderstanding from the reviewer on Fig. 5g. This figure was meant to provide a comparison of on-stream stability between 0.2Pt/m-Al 2 O 3 catalyst and control samples. The x-axis refers to time on stream, whereas the y-axis refers to the rate of deactivation, measured by the drop of n-hexane conversion every hour. From reviewer's comment, we realized the original Fig. 5g is not reader friendly, and may be a bit miss leading. We have replotted it (see below), using the ratio between activity at any time and initial activity as the y value, which highlights how fast and to which extent the catalysts loss activity under extreme conditions (550 °C, in the presence of H 2 ). From the new figure, it becomes easier to find that 0.2Pt/m-Al 2 O 3 only lost 10% activity whereas control samples lost over 30% initial activity after Revised Figure 5g: the catalyst deactivation at 550 °C.

Response:
This error has been corrected.
Reviewer My detailed comments are as follows: 1) The materials investigated do not reflect the industrial significance advocated in the introduction.
Indeed, the autocatalyst uses a lot of platinum every year as the authors cited, but NOT as Pt/Al 2 O 3 catalysts. The formulation of Pt/Al 2 O 3 has long been phased out from the mainstream three-way catalysts that are being used in conventional gasoline engine emission controls. Platinum is too volatile to survive the high-temperature aging, and palladium (major) and rhodium (minor) are the dominant precious metals being used. For diesel and other leanburn gasoline engine emissions, platinum has been used widely, but again not as Pt/Al 2 O 3 catalysts. The Pt-Pd alloy catalysts are now being widely used worldwide for this purpose. The alloy catalysts generally have a multifold better performance than the Pt-only catalysts for CO and HCs oxidation. Therefore, the catalyst development from the current work does not go into the heart of the technical challenge that we are facing today. The discussion in this section could be changed to reflect the practice more accurately. Response: The reviewer is a real expert in catalysis. Indeed, the new catalyst developed in this paper does not directly tackle any immediate technical challenge in industry. We have removed all descriptions on catalytic converters in the introduction. We have also modified the section on CO oxidation (MS Page 10).
2-1) According to the EXAFS results, it is claimed that the single-atom centric Pt-O4 planar structure will fully survive the H 2 reduction at 400 o C, and that this is indeed the versatile catalytic center for various reactions. However, if one compares Fig. 3D with Fig. 4F, a clear decrease of the Pt-O peak intensity in the R-space of EXAFS data can be observed for the same "best" 0.2Pt/m-Al 2 O 3 -H 2 catalyst.
Doesn't this indicate the evolution of the catalytic center? Response: We sincerely thank the reviewer for raising such an insightful comment. After inspecting the data, we find a mistake in the data processing procedure for this particular sample. The sample was measured under fluorescence mode but the box "Natural log" was mistakenly ticked. Since "mt = ln(I_fluorescence / I_0)" was used instead of "mt = I_fluorescence / I_0", the EXAFS oscillation was attenuated, thus the Fourier transform was also attenuated. That is a major reason for the significant decrease of Pt-O peak from 6.8 (Figure 3d) to 4 (original Figure 4f).
After processing the appropriate data with proper parameter selection, we obtained a revised Figure   4f as shown below. Since the difference between the fresh catalyst (Pt-O height of 6.8) and spent catalyst (Pt-O height of 5.5) may still not be considered as insignificant, we further conducted curve fitting analysis. To our surprise, the Pt-O coordination number before and after CO oxidation are almost identical. The difference in Pt-O peak height is cancelled by the increase of the Debye-Waller factor. While it remains unclear where the increase comes from, the change of Debye-Waller factor reflects some evolutions of the catalyst center as the reviewer suggested. In fact, the slight enhancement of the catalytic activity during the first few cycles also points to the same direction (Supplementary Figure 34a).
We added the following discussion to the text "……In the EXAFS spectrum, Pt-O contribution located at approximately 1.7 Å remains as the only prominent shell, unarguably proving that Pt largely maintains single-atom identity (Fig. 4f, Supplementary Fig. 32e explain the reaction mechanism, the alumina support was found not the part of the CO oxidation catalysis. There is a critical inconsistency here, which the authors must address.

Response:
We conducted the 2 nd and 3 rd cycles of catalyst reduction, without exposure to air between cycles.
Negligible amount of H 2 consumption was observed in these additional cycles. These new data were added into Supplementary Figure 1.  The carbon, hydrogen and nitrogen content in the catalyst all decreased considerably after reduction.
Based on these, we propose both the adsorbed surface oxygen species and the removal of C, N elements contribute to hydrogen consumption. The change of C, H and N contents in m-Al 2 O 3 (presumably on support surface) after reduction is a possible origin for the slightly different catalytic behaviour of 0.2Pt/m-Al 2 O 3 -O 2 and 0.2Pt/m-Al 2 O 3 -H 2 .
2-3) The authors should also analyze the Pt edge in post-reaction catalyst by EXAFS for all the reactions to confirm the stability and the coordination structure proposed according to the asreceived sample. Response: We analysed the EXAFS of Pt catalyst after CO oxidation (see response to 2-1). Pt fully maintained single-atom identify despite of some structural evolutions. We also managed to conduct XAS experiments for spent catalyst after n-hexane reforming reaction at 400 and 550 °C.  Table   4). These suggested the formation of Pt 0 nanoparticles/nanoclusters after reaction, but there is still a substantial amount of isolated Pt atoms in the catalyst…" (Page 15-16).
Since the new XAS experiments were conducted in Spring-8 in Japan, we have modified the "Catalysts characterization" section and the "Acknowledgements" section to include the information (Page 18 and 22, respectively).
3-1) For the CO oxidation reaction, a rather high contact time has been used (100 mg sample, 80 ml/min flow). Firstly, this does not well reflect the real application (suggest: 100 mg sample, 1-5 L/min flow).

Response:
Our mass flow controller has a maximum capacity of 100 ml/L. To increase the GHSV by a factor of ten, we decreased the catalyst to 10 mg while maintain the total flow rate at 80 ml/min. The TOF values are comparable with those obtained at 100 mg catalyst loading ( Supplementary Fig. 31a).
Furthermore, the apparent activation energy and reaction order measurements (high temperature range, 235-280 °C) were also conducted at this loading, vide infra.

3-2)
The more serious concern is that the key catalyst--0.2Pt/m-Al 2 O 3 has minimal activity below 200 o C in repeated cycles, even at such a high contact time. This is the same issue encountered in Ref.
42. I did a calculation of turnover frequency (TOF) myself, and found that the TOF numbers from the present work and in Ref. 42 are surprisingly close. This reinforces the fact, from both papers, that the conventional Pt/Al 2 O 3 catalyst with platinum particles and clusters present is seemingly more active for CO oxidation, even from the perspective of TOF per Pt atom. To deal with this issue, the authors simply cited the findings from Ref. 38, and claimed the particles are responsible for the lowtemperature CO oxidation. This is convenient but unconvincing and controversial. If the Pt particles contain the low-temperature catalytic sites, how does the commercial Pt/Al 2 O 3 with a lot of particles lose its low-temperature activity over the cycles? And why the 0.2Pt/p-Al 2 O 3 -H 2 sample that has (some and growing) clusters never became active for CO oxidation below 200 o C? Response: During paper revision, we have carefully measured the TOF of our catalyst between 180 and 280 °C, We believe commercial Pt/Al 2 O 3 catalyst lose low temperature activity over the cycles due to size increase of Pt nanoparticles over time. The fresh catalyst contains Pt nanoparticles with size centered at 3.9 nm, whereas the average size increased to 9.1 nm after 10 cycles. Moreover, the size distribution became much broader after CO oxidation. The histogram of Pt nanoparticles before and after reaction have been added to Supplementary Figure 29e,f and 38g,h, respectively.
The exact reason why 0.2Pt/p-Al 2 O 3 -H 2 sample never became active for CO oxidation below 200 °C is not clear. CO oxidation is a structure sensitive reaction. Presumably, the percentage of clusters/nanoparticles that are highly active for low temperature CO oxidation during the cycles, statistically, is always low during reaction on stream.
3-3) Nonetheless, the single-atom Pt catalysts can be very active for low-temperature CO oxidation as a few earlier papers have pointed out [Nat. Chem. 3, 634-641 (2011). Angew. Chem. Int. Ed. 53, 8904-8907 (2014)]. However, one the conditions for activity are very different; e.g. in the former paper the PROX reaction is examined. Dry CO oxidation may be very different. What is the effect of water (-OH groups)? Is the catalyst rendered more active in the presence of water? This is an important question both from a fundamental and practical viewpoint that must be considered by the authors.

Response:
We are aware of literature reports that water could promote CO oxidation activity by the formation of surface hydroxyl group, which subsequently modified the reaction pathway (e.g., ACS Catal. 2017, 7, 887−891). To identify whether such effect exists in our system, we conducted experiments to examine water effect during paper revision (10 mg catalyst mixed with 90 mg γ-Al 2 O 3 , flowing rate 80 mL/min). Water was introduced into the reactor by a syringe pump at an infusion rate of 44.1 μL/h, equivalent to 50 mol% of CO in the flowing gas. No significant change of activity was observed.
The TOF of the catalyst at 150, 200, and 250 °C were 0.012, 0.039, and 0.26 s -1 without adding water, whereas these numbers became 0.015, 0.043, and 0.22 s -1 , respectively, in the presence of water.
The effect of water on CO reactivity has been added into the text (MS Page 11). Ref. 42, which the present work employs to offer mechanistic interpretations, it is reported that the highly active Pt 1 -O x structure only has two oxygen atoms from the substrate coordinating with the Pt directly. However, what the authors prepared experimentally in this paper is a very stable (meets 16 e rule) planar Pt-O4 species. In the section of discussion, the author mentioned that "overly strong interaction leads to catalytically inactive species". Is it possible that "over stabilization" applies to the authors' own work? If the single atom 0.2Pt/m-Al 2 O 3 is not intrinsically more active for CO oxidation, what is the benefit of the new stabilized structures? Here the authors may invoke this stability as an attribute for some practical applications to be defined.

4) From the DFT results in
Pertaining to autocatalysts, good Pt catalysts should have significant CO conversion below 200 o Cotherwise the subsequent HCs and NO oxidation will be greatly hindered. Response: Based on XAFS and XANES analysis, our Pt catalyst stay in a 16-e stable configuration with four oxygen atoms from the support. This only represents the rest state of Pt catalyst, and the structure of Pt maybe dynamic during catalytic cycles. Although the Pt catalyst is not highly active in dry CO oxidation, its hydrogenation activity is remarkable, indicating the stable 16-e structure does not significantly compromise catalytic reactivity. We speculate that the coordination between one or two surface oxygen and Pt is labile, and may be replaced by reagents during the catalytic cycle. As such, the Pt center always remain in a stable 4-coordinate, 16-e state. One of the best techniques to experimentally prove that is conducting in-situ XAFS analysis. The collaboration between NUS team in Singapore and Kyoto U team Japan is continuing, and we have prepared and submitted a proposal to request beamtime for in-situ XAS data in Spring-8.
As the reviewer pointed out, our catalyst is not suitable as an autocatalyst. Considering its excellent activity, selectivity and stability in hydrogenating a wide range of substrates, it may be used in reduction reactions under both mild and harsh conditions.  Fig. 10,11). This makes me wonder if the method of preparation used here is really limited to very low Pt loadings. This is a drawback that must be made clear in the paper. On the discussion that follows still on p.8, another control sample for the authors to consider would be the addition of Pt by incipient wetness impregnation on their m-Al 2 O 3 . Can this Pt "anchor" on the special Al(3+) pentacoordinated sites?
How much Pt can thus be stabilized? The m-Al 2 O 3 has very high content (1/3) of these special sites. Response: Penta-coordinated Al 3+ on the surface has been proven to be the anchoring site for Pt ions (Ref 35), and we ascribe the high stability of Pt species to be associated with surface penta-coordinated Al 3+ .
Nevertheless, there is a significant difference between our m-Al 2 O 3 material and γ-Al 2 O 3 used in Ref 35. In the latter case, the spin-lattice relaxation times of tetrahedral and octahedral alumina were almost identical (120 ms) whereas that for penta-coordinated Al 3+ ions was much shorter (8 ms).
Based on that, the authors concluded that most of tetrahedral and octahedral alumina locate in the crystalline γ-Al 2 O 3 framework whereas penta-coordinated Al 3+ ions locate on the surface of the alumina support. In our system, the spin-lattice relaxation times for all three types of Al 3+ species are similar ( Supplementary Figure 22b), indicating a majority of penta-coordinated Al 3+ ions stay in the bulk. 0.2 wt% may be related to the number of penta-coordinated Al 3+ ions on the surface, which is only a small fraction. Based on this analysis, Pt content should be able to be increased to a higher value if the surface density of penta-coordinated Al 3+ sites increases.
As we understand, the current system only work well with 0.2 wt% loading. When increased to 0.5 wt% loading, the major species were still Pt single-atoms but Pt-Pt bond became observable in XAFS analysis. It is a drawback considering practical applications. We have added the following statement in the final paragraph of the discussion "…An apparent limitation of the system is that the singleatom identify of Pt was achieved only at a low loading (0.2 wt%), while mixed active sites were observed at more practical Pt loadings…" We have used pre-synthesized m-Al 2 O 3 as support to prepare Pt catalyst via wet-impregnation method. A mixture of Pt single atoms and nanoparticles were obtained even at 0.2 wt% (Supplementary Figure 21), highlighting Pt precursor has to be introduced prior to m-Al 2 O 3 formation.
When Pt(IV) precursor is mixed with Al precursor and ethanol, it is reduced to Pt(II) (based on ESI-MS analysis) and plausibly coordinates with Al through oxygen linkages, which may be critical for the formation of four-coordinate, 16-e Pt structure during the formation of mesopores.
6) How do the findings link to the classical debate of the structure sensitivity found in Pt/Al 2 O 3 catalysts for CO oxidation? Are there two mechanisms for the reaction? One at low-temperature and one at high-temperature. How about kinetic measurements over the two different structures of Pt catalysts presented here? As for the Pt dispersion, measured here by H 2 -O 2 titration and CO pulse chemisorption (never saturated?! Fig. S7), the authors do not have a table to summarize these results, and never use the standard term "dispersion" to describe these findings. I did a few calculations myself, and found that the relation between the kinetic rate and the total exposed Pt surfaces of the m-, p-, and commercial alumina supported samples does not have a clear trend. The authors need to ponder this important issue.

Response:
We conducted measurements of apparent activation energy and reaction order at both lower temperature range (180-250 °C) at 100 mg catalyst loading and higher temperature range (235-activation energy for CO oxidation between 180-250 °C was 80.3 kJ/mol, whereas it was 77.5 kJ/mol between 235-280 °C. There does not seem to be a two-mechanism scenario in our system.  Am. Chem. Soc. 137, 3470-3473 (2015). To demonstrate the point that the single atoms in alumina prepared as shown here do have a superior stability, the main premise of this paper, the samples treated at 600 o C should be fully analyzed to show the exclusive presence of single atoms, and the activity should also be reported.

Response:
The JACS paper mentioned by the reviewer is from Flytzani-Stephanopoulos's group, entitled "A Common Single-Site Pt(II)−O(OH)x− Species Stabilized by Sodium on "Active" and "Inert" Supports Catalyzes the Water-Gas Shift Reaction". It presented an elegant study on waster-gas shift reaction over alkali ion-stabilized single platinum atoms, not on CO oxidation reaction. It is not possible to conduct a direct comparison to our work due to different reaction systems. Nevertheless, we have inspected that paper to get some clues related to high temperature stability. In that study, the harshest reaction condition was to expose Pt catalyst at 400 °C, 350 °C and 300 °C for 2 h each, whereas we have maintained our catalyst working at the 400 °C for 220 h. Aggregation of Pt species into nanoparticles (~ 2 nm) was observed in their system ( Figure S11 in the JACS paper) after reaction, but it did not happen to our catalyst based on CO-DRIFT and XAS study.
As we are aware, the stability of single-atom catalysts have seldom been scrutinized at temperature above 320 °C (Nat. Chem., 2011,  could not fully survive reforming reaction condition, m-Al 2 O 3 seems to be able to prevent over- Due to the instability of isolated Pt species under reaction condition, the intrinsic activity of Pt (II) single-atom species compared with Pt(0) on nanoparticles in n-hexane reforming reaction remains unclear. It is very good advice to use a support having stronger anchoring effect with Pt ions to evaluate the intrinsic activity of isolated Pt atoms in reforming reaction and compare that with Pt nanoparticles, which, to our knowledge, has not done before, and we will investigate it in future work.
At present, we prefer to keep this part in the manuscript, as it tells the community the limit of the system, and points out a direction of future research.
9) The stability of the m-Al 2 O 3 seems to be poorer compared to many conventional alumina supports, although the Pt doping is found to stabilize the alumina framework. I suggest adding another reference sample by using the m-Al 2 O 3 and loading the platinum by incipient wetness impregnation method. The commercial Pt/Al 2 O 3 sample may have too many differences besides the state of Pt to make fair comparisons with the in-house prepared Pt catalysts. For example, the unusual light-off curve shape for the commercial Pt/Al 2 O 3 sample at high conversions may due to the mass transfer related issues rather than the intrinsically modified chemistry.

Response:
We have prepared another control sample as suggested by the reviewer, using m-Al 2 O 3 as the support to load 0.2 wt% Pt by wetness impregnation method (denoted as 0.2Pt/m-Al 2 O 3 -imp). A mixture of single-Pt species and Pt nanoparticles were generated (Supplementary Figure 21). When tested in CO oxidation reaction, its performance over 14 cycles lies between that of 0.2Pt/m-Al 2 O 3 -H 2 and commercial Pt/Al 2 O 3 catalyst. I.e., the lower temperature activity (< 200 °C) quickly decreased over the first few cycles whereas higher temperature activity (> 220 °C) slightly increased.
Temperature (  10) "The strong metal support interaction" has its unique original meaning in catalysis beyond the expression of "the interaction between the metal and support is strong". The authors need to be cautious about this and modify their wording in the section of discussion.

Response:
Thanks for pointing this out. Considering the unique meaning of "strong metal-support interaction (SMSI)" in catalysis, we have avoided the use of this expression in the revised manuscript. 11) Finally, the title of the paper should be more specific and spell out the Pt/m-Al 2 O 3 being investigated. A more appropriate title would be: Thermally stable single atom Pt/m-Al 2 O 3 for CO oxidation and the selective hydrogenation of 1,3-butadiene Response: We agree the title proposed by the reviewer sharpens the scope of the article. The title has been changed as suggested. 2) The 0.2 wt.% Pt single atom catalyst that was heated to 600 and 800 o C shows good cyclic performance up to 400 o C in subsequent CO oxidation reaction. The new figure S28 is convincing. Fig.  S27 shows NP formation after the high-temperature treatment and retention of some of the single Pt sites as well. The activity plot resembles that of samples heated to lower temperatures before the reaction. How are these findings reconciled? Response: We have plotted the cyclic performance of 0.2Pt/m-Al 2 O 3 -400 and 0.2Pt/m-Al 2 O 3 -600 in the same figure (see below), taking data from 1 st , 5 th and 10 th cycles, respectively. Indeed, the two catalysts exhibited almost identical activities (again, we are impressed by the reviewer's sharpness and meticulousness!). This could be rationalized by the fact that Pt single-atoms are still dominant in 0.2Pt/m-Al 2 O 3 -600, which is consistent with CO-DRIFT observations ( Figure S28). Pt nanoclusters and nanoparticles that are present in 0.2Pt/m-Al 2 O 3 -600 may have higher CO oxidation activity. However, these species account for a small percentage of Pt in the sample. Furthermore, their relative higher activity may be canceled by the smaller dispersion of Pt atoms. 3) The new Fig. S1 shows much higher reduction from the m-alumina sample. There may be H 2 O desorption from this special (hydrothermally prepared, mesoporous) alumina), which has nothing to do with H 2 O produced by H 2 -TPR. Need to check this point by simply running desorption in He.

Response:
Thanks for reviewer's comment. We considered the possibility of H 2 -TPR signal complexation by adsorbed H 2 O on Al 2 O 3 . Therefore, the samples were treated at 150 °C under N 2 for one hour before cooling down to 50 °C to start the H 2 -TPR experiments. This experimental detail has been added into the SI (page S4). To confirm whether this treatment is sufficient to remove adsorbed water in Al 2 O 3 , we have further conducted TPD experiments for 0.2Pt/m-Al 2 O 3 in He, both with and without pre-heating under He at 150 °C for one hour. A peak was present over the sample without pre-heating but the peak disappeared after pre-heating at 150 °C. This suggests our pre-heating step is sufficient to remove most of adsorbed H 2 O species on Al 2 O 3 . The new TPD profiles have been plotted and added as Figure S1b.
4) All the Pt samples (best and references) in this paper seemed to have darker color than the pure malumina (Fig. S3). I am not fully convinced that the best catalyst is Pt NP-free and the reference is simply too poor.

Response:
Thanks for reviewer's comment. We agree that the color of the material, to some extent, is an indicator of the existence of Pt NPs. Nevertheless, , the single-atom Pt is dominant in the 0.2 wt% Pt samples, while the portion of Pt NPs is below the detection limit of XAS analysis. We have added the discussion following Supplementary Figure 3.
5) The response 3-2 is not really convincing. Moreover, in the related response 3-6, the kinetics show that the single-atom Pt-O4 is behaving essentially the same as conventional Pt/Al2O3 NP catalysts in both activation energy and rxn orders. If they are catalytically the same, why are the single atoms inferior in low-temperature CO oxidation activity? In their original submission, the authors claimed there are two stages of kinetics for NPs and single atoms, which have now been disputed by their own kinetics. Table S6 does not answer the question either-it simply indicates that their atomic catalyst is as inactive as others. Otherwise, there must be a significant presence of Pt metal that hides the presence of cationic Pt in kinetics. For the final version of the paper, the authors should put the emphasis on the hydrogenation reaction, where the new catalyst is exciting and most promising; and limit the interpretation of CO oxidation data as a way of examining stability in oxidative atmosphere-not activity. Response: Thanks for the excellent point. We do not totally understand the catalytic behavior of the single-atom Pt catalyst in CO oxidation. What is certain from the experiments is that the 0.2 wt% Pt/m-Al 2 O 3 catalyst is not very active but stable under the reaction condition. Similarly, we could not fully rule out the possibility of activity contribution from a small portion of highly active NPs for CO oxidation. Due to these considerations, we have adopted the reviewer's suggestion to limit the interpretation of CO oxidation data as a way of examining STABILITY in oxidative atmosphere, while we have strengthened discussions on selective hydrogenation with more discussions provided. Consequently, the title of the paper has been changed as "Thermally stable single atom Pt/m-Al 2 O 3 for selective hydrogenation and CO oxidation", and the abstract has been modified. 6) Throughout the paper, in IR parts, please use CO absorption band-not adsorption band.

Response:
Thanks for pointing out the misspelling. Corrections have been made in the revised version.