Sulfur stabilizing metal nanoclusters on carbon at high temperatures

Supported metal nanoclusters consisting of several dozen atoms are highly attractive for heterogeneous catalysis with unique catalytic properties. However, the metal nanocluster catalysts face the challenges of thermal sintering and consequent deactivation owing to the loss of metal surface areas particularly in the applications of high-temperature reactions. Here, we report that sulfur—a documented poison reagent for metal catalysts—when doped in a carbon matrix can stabilize ~1 nanometer metal nanoclusters (Pt, Ru, Rh, Os, and Ir) at high temperatures up to 700 °C. We find that the enhanced adhesion strength between metal nanoclusters and the sulfur-doped carbon support, which arises from the interfacial metal-sulfur bonding, greatly retards both metal atom diffusion and nanocluster migration. In catalyzing propane dehydrogenation at 550 °C, the sulfur-doped carbon supported Pt nanocluster catalyst with interfacial electronic effects exhibits higher selectivity to propene as well as more stable durability than sulfur-free carbon supported catalysts.


Reviewer #1 (Remarks to the Author):
In their manuscript Yin et al. reported about the effect of sulfur on stability of metal nanoclusters against their sintering. The authors have shown that metal nanoclusters (Pt, Ru, Rh, Os, or Ir) supported on sulfur-doped carbon support exhibit remarkable sinterresistant properties unlike their counterparts supported on the undoped support. It has been concluded that the enhanced stability of the former arises from strong metal-sulfur interaction. This conclusion is not novel (see comment 1). Practical relevance of Pt/S-C catalyst was also not justified (see comments [2][3][4][5]. Specific comments are given below. 1. The novelty of the idea of stabilizing metal nanoparticles against sintering by interaction with sulfur on carbon support is doubtful. Several papers (J. Mater. Chem., 2009, 19, 5934-5939; J. Mater. Chem. A, 2017, 5, 19467-19475;Scientific Reports, 2019, 9, 12704) have shown that sulfur-doped carbon materials are promising supports for Pt-based catalysts. The addition of sulfur into support was shown to improve thermal stability and dispersion of Pt particles due to the strong metal-support interaction between the sulfur and Pt atoms. Although there is an obvious similarity between the current work and the works listed above, the present authors did not refer to previous studies.
Response : We agree with the reviewer that the above mentioned literatures have reported sulfur-doped carbon materials as promising supports for Pt-based catalysts, which were not cited in the original manuscript. Actually we cited another two papers that discussed the stabilization of metal clusters based on the interactions between metal and nitrogen or sulfur atoms (refs. 32 and 33). Here we would like to explain why we cited refs. 32 and 33 but not the mentioned literatures in the original manuscript.
In all the mentioned literatures, the S-doped carbon supports were reported to stabilize Pt nanoparticles of 3~5 nm, instead of Pt nanoclusters. For example, in the first paper (J. Mater. Chem., 2009, 19, 5934-5939), the authors demonstrated that Pt nanoparticles supported on sulfur-containing ordered mesoporous carbon (S-OMC) exhibited thermal stability at 600 °C under N2 flow and that the sizes of Pt nanoparticles increased from 3.14 nm to 4.50 nm. Another two papers (J. Mater. Chem. A, 2017, 5, 19467-19475;Scientific Reports, 2019, 9, 12704) studied the electrochemical stability (instead of thermal stability) of about 3 nm Pt nanoparticles on sulfur-doped carbon.
In contrast, our current work focuses on the thermal stability of metal nanoclusters of around 1 nm rather than nanoparticles of 3~5 nm, as we highlighted in the Title (Sulfur stabilizing metal nanoclusters on carbon at high temperatures), Abstract, and Introduction parts. The motivation of our work is that metal nanoclusters have some distinct features compared to nanoparticles but are extremely difficult to stabilize even on high-surface-area supports, owing to the remarkably improved surface free energy. This is particularly challenging for metal nanocluster catalysts in the applications of high-temperature reactions, where thermal sintering induced catalyst deactivation is often the key issue.
Therefore, in the original manuscript, we cited two papers that discussed the stabilization of metal clusters based on the interactions between metal and nitrogen (ref. 32) and sulfur atoms (ref. 33).
Further, although some doped carbon supports (including the literatures mentioned by the reviewer) have been reported to stabilize Pt nanoparticles of 3~5 nm up to 600 °C, to our knowledge, so far no one has reported that metal clusters of 1 nm could be stabilized on carbon supports up to 700 °C. We thus believe that these published works on the nanoparticles do not weaken the novelty of our work on metal nanoclusters and that our work represents a solid step toward the applications of metal nanocluster catalysts under realistic technical conditions.
According to the reviewer's comments, we also cited the mentioned three papers in the revised manuscript to better clarify the background as below: Page 4: "Although some sulfur-doped carbon supports have been reported to stabilize Pt nanoparticles of 3~5 nm up to 600 °C, to our knowledge, so far no one has reported that metal clusters of 1 nm could be stabilized on carbon supports up to 700 °C." 2. It is well-known that unpromoted supported Pt species tend to sinter under propane dehydrogenation (PDH) conditions. Therefore, a typical industrial catalyst contains Sn as a promoter for stabilizing Pt particles. In order to demonstrate the practical potential of Pt/S-C in propane dehydrogenation the authors should compare their catalyst with Sn-Pt/Al2O3 under the same reaction conditions. Response: Many thanks for the reviewer's valuable comment here. We have supplemented the PtSn/Al2O3 catalysis data in the revised manuscript. The commercial γ-Al2O3 support was purchased from Alfa-Aesar. The PtSn/Al2O3 catalyst was prepared by the conventional impregnation method. The prepared PtSn/Al2O3 was then tested for the propane dehydrogenation under the same reaction conditions (Fig. R1). The selectivity of PtSn/Al2O3 (98%) was comparable to Pt/S-C, but the conversion on PtSn/Al2O3 catalyst gradually decreased from 39% to 32% (corresponding to 17.9% deactivation) after 600 min, then further slowly decayed to 30% after 950 min. The deactivation rate of PtSn/Al2O3 was 0.031 h -1 within 600 min, which is also much higher than that of Pt/S-C (0.005 h -1 ). Meanwhile, sintering of PtSn nanoclusters happened by monitoring the morphology change of PtSn/Al2O3 catalysts before and after the PDH reaction (Fig. R2), which results in the attenuation of activity despite of its high selectivity.
The above results and discussion on the PtSn/Al2O3 catalyst have been added in the    Table 3) is not possible. Moreover, unlike the state-of-the-art catalysts listed in Table 3, Pt/S-C was tested using a quite diluted mixture (only 10 vol%C3H8) that may be a reason for the high selectivity and low deactivation constant. Moreover, the present authors do not provide degrees of propane conversion. In general, catalysts used in the PDH reaction can be compared at same propane conversion obtained at same temperature and using same reaction feed. Thus, the superior performance of Pt/S-C is not justified.
Response: Thanks for the reviewer's valuable comments here. As suggested by the reviewer, it is impossible to make a more fair performance comparison between different catalysts due to the various experimental conditions. We quite agree with the reviewer on this issue and have deleted the comparison table and relevant discussion from the revised manuscript to make the discussion more objective. Meanwhile, we further supplemented the catalytic experiments of high-concentration C3H8 with the ratio of propane of 33 vol% and 50 vol% in feeding (Fig. R3). As a result, when the C3H8 concentration was increased from 10 vol% to 33 vol% and 50 vol%, the conversion of Pt/S-C decreased from 40% to 20% and 13%, respectively. Despite of a slight decrease in selectivity (from 98% to 95%) at 33 vol% C3H8 feeding, the Pt/S-C showed a high stability in activity with a comparatively lower deactivation rate of 0.003 h -1 , which clearly demonstrates the superior stability of Pt/S-C. The selectivity dropped to around 88% at 50 vol% C3H8 feeding, but the Pt/S-C still showed a stable activity.
In contrast, for the Pt/S-free-C reference catalyst, the propane conversion dramatically dropped from 30% to 18% after 1000 min and the selectivity also showed a sharp decline from 80% to 50% with the increase of propane concentration.
The above results and discussion on the catalytic experiments of high-concentration  4. It is also not clear how the authors determined the rate of propene formation. Was it determined in a differential reactor? What was the degree of propane conversion? Response: Thanks for the reviewer's valuable comments here. According to the reviewer's comments, we have changed the ordinate to the propane conversion in revised manuscript. As described in the original manuscript (Page 13), the propane dehydrogenation reaction was performed in a fixed-bed quartz tube reactor at the atmospheric pressure and the products were analyzed by an online GC-14C gas chromatograph equipped with a flame ionization detector and a thermal conductivity detector. The propene formation was defined as the moles of C3H6 formation per g Pt per hour from the following formula: where F (C3H8) represents the flow rate of propane, ( ) and are the yield of propylene and the weight of catalysts, respectively, and is the percentage of Pt weight loading in the catalyst.

5.
Another serious concern about the applicability of the developed catalysts for propane dehydrogenation is the absence of any data related to anti sintering performance under air conditions. Do Pt species sinter in air? This is actually a problem of commercial Pt-containing catalysts. Part of above discussion has been added in the revised manuscript (page 4).

Reviewer #2 (Remarks to the Author):
This manuscript discusses a new method of stabilizing small metal nanoclusters for heterogeneous catalysis using sulfur-doped carbon matrix support. This approach allows high thermal stability of 1 nm metal nanoclusters of Pt, Ru, Rh, Os, and Ir at elevated temperatures up to 700 °C, preventing thermal sintering and consequent deactivation. The S-C bond in the matrix enhanced the adhesion strength between the nanoclusters and the support, which arises from the interfacial metal-sulfur interaction that retards metal atom diffusion and ripening. A highly efficient and stable catalyst for propane dehydrogenation was developed based on sulfur-doped carbon-supported Pt nanocluster with 98% selectivity and extended durability for 1800 minutes. The work presented in the manuscript is scientifically sound and presented concisely and clearly.
This work provides new synthetic approaches for heterogeneous catalysis by solving an important issue of metal sintering and deactivation during high-temperature catalysis.
I therefore recommend its publication after addressing the following issues: As suggested by the reviewer, we have reinforced the relevant discussion to make the novelty and advantages of our methods more clear. TiO2 as support?
Response: Thanks for the reviewer's valuable comments here. We believe that the primary function of carbon is to supply a thermally stable matrix to stabilize sulfur sites.
According to the reviewer's comments, we tried to apply sulfur-stabilizing method to   Pt/TiO2 catalysts after annealing at 700 °C in 5% H2/Ar for 120 min.
The above results and related discussion have been added in the revised manuscript (Page 5) and Supplementary information (Figs. S16 and S17). Fig. 1 to a more demonstrative schematic illustration.

I suggest the authors change
The current figure is a bit confusing and does not give precise overall info of the manuscript story.
Response: Thanks for the reviewer's valuable comments here. As suggested by the reviewer, we have revised the schematic illustration to give clear description of the manuscript story (Fig. R7).
The above figure have been changed in the revised manuscript ( Fig. 1).   7. I also suggest that the authors demonstrate the applicability of their catalysis in other high-temperature reactions, such as methane or CO conversion.
Response: Many thanks for the reviewer's comments on this issue. As suggested by the reviewer, we further evaluated the stability of the Pt/S-C and Pt/S-free-C for water-gas shift (WGS) reaction (Fig. R10). The catalytic performance of the catalysts in the WGS reaction was evaluated in a fixed-bed flow reactor. The catalyst (10 mg) was pretreated by 5% H2/Ar at 200 °C for 2 h. After that, the temperature was increased to 400 °C, and the catalyst was exposed to the WGS reaction mixture. The reactant gas consisted of 5% CO (flow rate: 30 mL min −1 ) and water vapor at 46 °C (water vapor pressure: 10.094 kPa) balanced with Ar that yielded the PCO/PH2O ratio of 1:2. All catalysts were heated to the desired reaction temperatures at a rate of 1 K min −1 , and the steady state compositions of the effluent gas were analyzed with an online gas chromatograph (FULI 9790II) with a TCD attached to a TDX column. The catalytic activity was calculated by the change in the CO concentrations of the inlet and outlet gases. The WGS rate was calculated based on the total Pt content. As shown in Figure R10, Pt/Sfree-C exhibited a higher initial activity (286.7 ) than Pt/S-C, which could be ascribed to the high oxidation state of Pt on S-free-C. The electron-deficient Pt is beneficial to weaken CO binding and promote reaction (Nature 2017, 544 (7648)

Reviewer #3 (Remarks to the Author):
Yin et al. reported a sulfur doped carbon matrix to stabilize the various metal nanoclusters (Pt, Ru, Rh, Os, and Ir). Among the synthesized nanoclusters, Pt was found to perform high selectivity in propane dehydrogenation. Despite the topic may be interesting to the community, several uncertainties are required to be included in order to draw a more rigorous scientific conclusion for the current experimental findings.
Therefore, the current form is suggested to undergo a major revision subject to further review with the additional improvement.  We noted that the PDH reaction was rarely reported by other metals besides Pt.
According to the reviewer's comments, we also tested the catalytic activity of other S-C supported metal cluster catalysts, but their activity and selectivity was poor (Fig.   R12). For examples, Os/S-C showed a very low activity with a conversion of less than 4%, while Ru/S-C and Rh/S-C exhibited a slightly higher conversion of around 10%.
Though the activity of Ir/S-C (~35% conversion) is comparable to that of Pt/S-C, the selectivity is poor (~80%). Further exploration of the catalytic applications of these metal cluster catalysts for other reactions can be carried out in future. very close to that of cuboid-shape, which means that the truncated octahedron is also energy favored.
Meanwhile, The HAADF-STEM image show that the exposed crystal faces of Pt/S-C are (111) and (200) (Fig. R13), which is in good agreement with the atomic array and exposed crystal plane of the proposed Pt38 structure. Overall, considering the computational and experimental results, we chose the Pt38 clusters with truncated octahedron shape in the current simulation work. 4. The DFT results of Figure 4 are also confusing. Figure 4(a) should be renamed as Pt38 cluster desorption since the FS geometry of Pt38/S-C is substantially away from the graphene plane. The individual atom "escape energy" in Figure 4(b) is ambiguous due to the selection of the "single atom" is not well-defined, and the "escape energy" is believed to be substantially subject to the atom selection.
Response: Thanks for the reviewer's valuable comments here. We have corrected the description of Figure 4(a) to "Energy barrier of Pt38 cluster desorption on S-Graphene and Graphene" to make our manuscript more precise. Moreover, we further selected several other atoms at different positions to calculate the average escape energy of individual atom (Fig. R14). As can be seen, the escape energy of all individual atoms on S-Graphene was larger than that on Graphene, which can be ascribed to the enhanced strength of interfacial adhesion at the Pt38/S-Graphene owing to the metal-S bonding.
The above results and related discussion have been added in the revised manuscript (Page 7, Fig. 4). to C3H5*, Pt/S-C exhibits a higher C3H6 selectivity. As for the absorbed geometric effect, we have repeatedly optimized the molecular adsorption sites before the PDH reaction paths (Fig. R15-R17), and the final system is spontaneously stabilized at the sites shown in Figure R15.
As suggested by the reviewer, we further study the charge transfer by the density of state (DOS) analyses for the two model (Fig. R18). For the C3H8* adsorption stage, the     I have carefully read the author's reply and the emended manuscript as well as the supporting information. I understand the arguments related to the novelty of this study. Nevertheless, I am not sure if it is a significant difference if the presence of S in support is decisive for stabilization of Pt nanoparticles of 3-5 nm or 1 nm as has been previously reported in several previous studies on in the present manuscript. The practical relevance of the developed has not been justified due to the following reasons. i) As seen in Fig.5(c) and Figure R3(a,c), the selectivity decreases with an increase in propane concentration in reaction feeds. Nevertheless, the authors reports the values obtained with a diluted feed, where the highest propene selectivity was achieved. In addition, the selectivity was calculated on the basis of detected gas-phase products without considering coke formation. The latter is typically a major side product in propane dehydrogenation.
ii) The fact that carbon support will burn upon oxidative catalyst regeneration is known. Why did the authors not apply their approach using an industrially relevant support? iii) No tests at higher reaction temperatures relevant for industrial applications have been performed.
As the authors did not provide units for terms used for calculating the rate of propene formation, it is not easy to check if this formula correct. It is also not explained if the rate was determined under integral or differential conditions. Reviewer #3 (Remarks to the Author): I am satisfied with the changes made by the authors and their explanations on the scientific comments/concerns risen by reviewers. It could be a even better presentation if couple sentences of the writing can adjusted to be more academic style. This manuscript is recommended for publication after minor revisions noted.
There are a few examples that can be further revised. At page 4, the sentence -"no one has reported ...." could be further revised. At page 4, the sentence -"we must clarify ...." could be further revised.

Reviewer #1 (Remarks to the Author):
I have carefully read the author's reply and the emended manuscript as well as the supporting information. I understand the arguments related to the novelty of this study.
Nevertheless, I am not sure if it is a significant difference if the presence of S in support is decisive for stabilization of Pt nanoparticles of 3-5 nm or 1 nm as has been previously reported in several previous studies on in the present manuscript.
The practical relevance of the developed has not been justified due to the following reasons.
i) As seen in Fig.5(c) and Figure R3(a,c), the selectivity decreases with an increase in propane concentration in reaction feeds. Nevertheless, the authors reports the values obtained with a diluted feed, where the highest propene selectivity was achieved. In addition, the selectivity was calculated on the basis of detected gas-phase products without considering coke formation. The latter is typically a major side product in propane dehydrogenation.
Response: Thanks for the reviewer's valuable comment here. According to the reviewer's comments, we have deleted the high selectivity values in the Abstract section and further supplemented the description of selectivity changes under different propane concentration feeding ( Figure R1), which make the discussion more objective.
Though the selectivity decreased with an increased propane concentration in reaction feeding, the S-contained catalyst was still superior to S-free catalyst, which further suggested the positive role of strong metal-sulfur interaction in promoting the selectivity of propane dehydrogenation.
As for the calculation of selectivity, we took the experiment of Pt/S-C under 33% C3H8 feeding as example (Table R1). The reaction product gas (CH4, C2H4, C2H6, C3H6, and C3H8) were continuously detected by online flame ionization detector (FID) and thermal conductivity detector (TCD) gas chromatograph equipped downstream. Under sampling at intervals of six minutes, we could obtain the corresponding peak area of the reaction product (column 2-6 of Table R1). Using the external standard method, we could quantify the molar content of each product by peak area and obtain the relative molar ratio (column 7-11 of Table R1), and further calculate the carbon balance (Eqs. 1, column 12 of Table R1). Carbon balance typically ranged between 95% and 105% for all the reactions, which allows for ignoring the loss of carbon deposition to some extent. Figure R1. The above results and discussion catalyst have been added in the revised manuscript (page 9) and Supplementary Information (Fig. S26). ii) The fact that carbon support will burn upon oxidative catalyst regeneration is known.
Why did the authors not apply their approach using an industrially relevant support?
Response: Many thanks for the reviewer's comments on this issue. According to the reviewer's comments, we tried to apply sulfur-doped TiO2 support to regeneration for PDH reaction under oxidation conditions. The sulfur-doped TiO2 support Pt nanoclusters also exhibited outstanding thermal stability up to 700 °C owing to the strong Pt-S interactions (Fig. S16).  and Supplementary Information (Fig. S28).
iii) No tests at higher reaction temperatures relevant for industrial applications have been performed.
Response: Thanks for the reviewer's comments on this issue. According to the reviewer's comments, we further tried a higher temperature PDH reaction at 600 °C.
When the reaction temperature increased to 600 °C, the C3H8 conversion of Pt/S-C gradually decreased from 55% to 40% after 1200 min with a 0.030 h -1 deactivation rate and the selectivity was around 92%, indicating a slightly worse catalytic performance than that under 550 °C. Under the same high-temperature condition, the Pt/S-free-C showed rapid deactivation at the beginning of the reaction, which further demonstrated the pivotal role of the Pt-S interactions in catalyzing PDH reaction. The above results and discussion catalyst have been added in the revised manuscript (page 9) and Supplementary Information (Fig. S27).
As the authors did not provide units for terms used for calculating the rate of propene formation, it is not easy to check if this formula correct. It is also not explained if the rate was determined under integral or differential conditions.
Response: Thanks for the reviewer's comments on this issue. The propene formation rate was determined under differential conditions. Each time node could correspond to the yield of propene at a current time. Then the propene formation was calculated by the formula that was defined as the moles of C3H6 formation per gPt per hour. These details have been described in the Experimental Section in the revised manuscript.

Reviewer #3 (Remarks to the Author):
I am satisfied with the changes made by the authors and their explanations on the scientific comments/concerns risen by reviewers. It could be a even better presentation if couple sentences of the writing can adjusted to be more academic style. This manuscript is recommended for publication after minor revisions noted.
There are a few examples that can be further revised.
At page 4, the sentence -"no one has reported ...." could be further revised.
At page 4, the sentence -"we must clarify ...." could be further revised.
Response: Many thanks for the reviewer's comments on this issue. According to the reviewer's comments, we have adjusted the expression of the above sentences. We also checked the whole manuscript again and revised accordingly.

<b>REVIEWER COMMENTS</B>
Reviewer #1 (Remarks to the Author): The authors have provided some additional data and revised their manuscript. The reported results show that the presence of S in the support is important for higher propene selectivity and on-stream stability in comparison with a S-free reference material. For industrially relevant feeds, the selectivity values are rather "standard" when considering relative low degree of propane conversion. For a feed with 33viol% C3H8, the selectivity is about 93% at only 18% propane conversion (Table R1). The selectivity becomes even lower, when performing PDH with a feed containing 50 % propane. Although, the degree of propane conversion is not given, I assume it is not higher than 18%. These values are not surprising when even comparing with the state-of-the-art catalysts based on metal oxides (DOI: 10.1039/d0cs01140a). In addition, the space time yield of propene formation calculated on the basis of Figure 5(a) is also not significantly higher that the corresponding values reported for metal oxide catalysts.
In terms of the rates of propene formation and deactivation, this catalyst does not show any unexpected results with industrially relevant feeds when comparing with previously tested Pt-based catalysts ( Fig.8 and Table 2 in DOI: 10.1039/d0cs00814a). Figure R2. It is not written which feed was used. In the case of feed with 10vol% C3H8, 0.1 wt% Pt/S-TiO2 deactivates significantly quicker in comparison with Pt/S-C

Reviewer #1 (Remarks to the Author):
Comments: The authors have provided some additional data and revised their manuscript. The reported results show that the presence of S in the support is important for higher propene selectivity and on-stream stability in comparison with a S-free reference material.
For industrially relevant feeds, the selectivity values are rather "standard" when considering relative low degree of propane conversion. For a feed with 33viol% C3H8, the selectivity is about 93% at only 18% propane conversion (Table R1). The selectivity becomes even lower, when performing PDH with a feed containing 50 % propane. Although, the degree of propane conversion is not given, I assume it is not higher than 18%. These values are not surprising when even comparing with the state-of-the-art catalysts based on metal oxides (DOI: 10.1039/d0cs01140a). In addition, the space time yield of propene formation calculated on the basis of Figure 5(a) is also not significantly higher that the corresponding values reported for metal oxide catalysts.
In terms of the rates of propene formation and deactivation, this catalyst does not show any unexpected results with industrially relevant feeds when comparing with previously tested Pt-based catalysts ( Fig.8 and Table 2 in DOI: 10.1039/d0cs00814a).

Response:
We really appreciate the reviewer's patience during the two rounds of reviewing. Specially, the reviewer's professional comments on the propane dehydrogenation are very helpful to promote the quality of our work.
In the last version of manuscript, we have supplied the conversion and selectivity results under different C3H8 feed ( Figure S22) and we also mentioned these data in the main text (page 9). The conversion of Pt/S-C indeed decreased to only 13% at 50 vol% C3H8 feed. In addition, the space time yield (STY) of propene formation also decreased from 0.88 to 0.41 and 0.33 with the C3H8 feed increasing from 10 vol% to 33 vol% and 50 vol% (Table R1).
According to the reviewer's comments, we tried to compare the performance of our catalyst with the reported monometallic Pt, Pt alloy, as well as metal oxide catalysts, in terms of the space time yield of propene formation, the selectivity to propene, and deactivation rate. Taking into account the difference in experimental conditions in many studies, the absolutely fair comparison between different catalysts is very challenging. To make the comparison as reasonable as possible, we restricted the test conditions as follows except for a few samples: i) the temperature of 550~600 °C; ii) the propane content of 10~50 vol%; and iii) the WHSV of 1~5 h -1 . We can conclude from the comparison summarized in Table R1, Figs. R1 and R2 that: i) At at high C3H8 feed, the Pt/S-C catalyst exhibited a lower C3H6 STY than some metal oxides catalysts and most Pt alloy catalysts and acquired a higher C3H6 STY under high WHSV (5 h -1 ) at 10 vol% C3H8 feed, which is superior to most monometallic Pt and even some Pt alloy catalysts.
ii) As for the selectivity, there is little difference between Pt alloy catalysts and the Pt/S-C (above ~90% for both kinds of catalysts), while some metal oxide and monometallic Pt catalysts showed low selectivity of 75~85%.
iii) In the term of deactivation rate, the Pt/S-C is obviously in the top position with outstanding stability and even better than most Pt-alloys catalysts.
Metal oxides catalysts show considerable activity, yet suffer from the loss of oxygen under reaction conditions and thus rapid deactivation in a short time. Over the course of a catalytic cycle, due to the coke deposition, frequent regeneration operations are needed 1 . We agree with reviewer that our monometallic Pt/S-C catalyst did not show unexpected C3H6 STY compared to the state-of-the-art catalysts metal oxide and Pt alloy catalysts. But the Pt/S-C catalyst indeed exhibited remarkable sintering resistance and low deactivation rate, which is actually the core story of this work, as highlighted in the title, abstract, and introduction part. As we have demonstrated, the sulfurstabilization strategy could be extended to prepare a wider range of noble metal catalysts (Pt, Ru, Rh, Os, and Ir) and applied to other high temperature reaction (such as WGS reaction, as shown in Figure  S29). All these results demonstrate the validity of our concept of "Sulfur stabilizing metal nanoclusters on carbon at high temperatures".
Inspired by the reviewer's comments, we additionally synthesized bimetallic Pt-Sn/S-C catalyst with targets of further promoting C3H6 STY and selectivity meanwhile maintaining low deactivation rate, in particular at high 50 vol % C3H8 feed. Fortunately, we got positive results. The bimetallic Pt-Sn/S-C catalyst exhibited an enhanced activity with a conversion of 25% and also a promoted selectivity of 94% compared to monometallic Pt/S-C catalysts (13% conversion and 88% selectivity) under the identical reaction conditions. Meanwhile, the stability was well maintained with a low deactivation rate constant of 0.003 h -1 . The systematic optimization of the Pt-Sn/S-C catalyst and comprehensive screen of bimetallic combination of alloys may further improve the performance, but it has seriously deviated from the core concept of this work.
The above data and related discussion have been added in the revised manuscript (page 9) and supplementary materials (Fig. S28-29) Comments: Figure R2. It is not written which feed was used. In the case of feed with 10vol% C3H8, 0.1 wt% Pt/S-TiO2 deactivates significantly quicker in comparison with Pt/S-C Response: For the Figure R2 in last response letter, 10 vol % C3H8 was used for the Pt/S-TiO2 catalyst.
We have updated such information.
The Pt/S-TiO2 catalyst indeed deactivated quickly in comparison with Pt/S-C. The rapid deactivation was ascribable to the strong acid centers on the TiO2 supports, which would lead to the formation of a significant amount of coke even under high selectivity 2 . The coke deposition could be rapidly removed by the regeneration under oxidation atmosphere; the Pt clusters showed no sintering and the activity was recovered.
The above discussion has been added in the revised manuscript (Fig. S30). Figure R1: Comparisons of the C3H6 STY, the selectivity to propene, and deactivation rate of the Pt/S-C, Pt-Sn/S-C, reported metal oxides, monometallic Pt and Pt alloy catalysts for PDH. To make the comparison as reasonable as possible, the test conditions were restricted as follows except for a few samples: the temperature of 550~600 °C, the propane content of 10~50 vol% and the WHSV of 1~5 h -1 . The details of these catalysts were summarized in Table S3.