CO oxidation activity of non-reducible oxide-supported mass-selected few-atom Pt single-clusters

Platinum nanocatalysts play critical roles in CO oxidation, an important catalytic conversion process. As the catalyst size decreases, the influence of the support material on catalysis increases which can alter the chemical states of Pt atoms in contact with the support. Herein, we demonstrate that under-coordinated Pt atoms at the edges of the first cluster layer are rendered cationic by direct contact with the Al2O3 support, which affects the overall CO oxidation activity. The ratio of neutral to cationic Pt atoms in the Pt nanocluster is strongly correlated with the CO oxidation activity, but no correlation exists with the total surface area of surface-exposed Pt atoms. The low oxygen affinity of cationic Pt atoms explains this counterintuitive result. Using this relationship and our modified bond-additivity method, which only requires the catalyst–support bond energy as input, we successfully predict the CO oxidation activities of various sized Pt clusters on TiO2.

The authors performed a combined experimental and computational study on CO oxidation reaction on a series of Ptn/Al2O3 systems (and partially extended to Ptn/TiO2). They find that the undercoordinated Pt atoms located at the very edges of the first cluster layer are more positively charged (aka cationic) due to direct contact with the Al2O3 support. The overall CO oxidation activity is found to correlate with Nnutral/Ncation/Ntotal, "the ratio of neutral to cationic Pt atoms normalised by the total number of Pt atoms". Using a bond-additivity model (BAM) that based on computationally or empirically estimated bond energies between a single catalyst atom and the support, they constructed the thermodynamically stable morphologies of the clusters, and predicted the CO oxidation activities of such clusters.
Overall, I think the experimental results of the CO oxidation activity on a series of size-selected Ptn cluster on Al2O3 and TiO2 surfaces provide interesting data for constructing a meaningful model in predicting the catalytic properties. I am not an expert on the experimental aspects, thus my comments are mainly on the computational results. Many of the findings in this work are consistent with the general understanding of catalysts with nano-particles, thus are not surprising or completely new. For example, the findings that the interfacial atoms are more ionic and the oxidation state of Pt increases with the particle size decreases are such cases. However, they provide clear evidence that the CO oxidation activity is correlated with the ratio Nnutral/Ncation/Ntotal, which is interesting and provides insight for understanding the role of various metal sites with different oxidation states or charge states. I therefore think the manuscript might become publishable for Nature Communications after taking into account of the following comments and suggestions:

1)
When Ptn clusters are supported on oxides, the difference of the chemical potentials decides how much amount of charges are transferred, as have been discussed in the literature (e.g. J. Am. Chem. Soc., 2017, 139, 6190). They need to clarify this point because when the metal cluster and the support have similar chemical potentials, there is no cationic atoms, even at the interfacial region.

2)
The undercoordinated, interfacial Pt atoms are more positively charged if the charge transfer flows from metal to support. Since the authors performed DFT calculations, I am surprised that they did not provide the Bader charges for these atoms at different layers of the cluster. The charges should be included to provide more direct data on the "chemical state" change of the Pt atom.

3)
Conceptually, valence state, oxidation state, and atomic charge are quite different concepts. The authors should avoid mixing them together. The high oxidation state metal (e.g. Pt(II) or Pt^+2) often carries more positive charge than a low oxidation state metal (e.g. Pt(0)). But one cannot confuse oxidation state with charge state. In p. 7, "Pt within the Pt7 cluster has an oxidation state between Pt0 and Pt2+". Here Pt2+ is the charge state, while Pt(II) or Pt+2 is the oxidation state.

4)
The key finding of CO oxidation activity to correlate with Nnutral/Ncation/Ntotal needs some further explanation of its physical meaning. In fact, Nnutral/Ncation/Ntotal = Nnutral/(NcationNtotal) = (Ntotal -Ncation)/(NcationNtotal) = 1/Ncation + 1/Ntotal ; Because the Ntotal is usually far larger than Ncation, one can imagine that this simply says that the CO oxidation activity is inversely proportional to the Ncation. On the other hand, when making this correlation of CO oxidation activity correlating with Nnutral/Ncation/Ntotal , have they excluded those atoms that are inaccessible (e.g. not on the surface) for CO oxidation?

5)
The manuscript correlates the cationic state with the coordination number. In fact, both the coordination number and the bond distances matter for the bond energy, as discussed in Nature Chem. 2015, 7, 403. The authors missed this important work.

6)
One should realize that the metal clusters might be dynamically changed in structure when CO and O2 are adsorbed and react (e.g. Nature Commun., 2015, 6, 6511). The intrinsic structure of Ptn/support may not be the same as it started before the reaction. This should be pointed out in the manuscript to readers.

7)
Well-defined clusters embedded or supported on oxides surface provide opportunity for precision control of catalytic reactions. This kind of "single-cluster catalysts" are widely studied lately and should be discussed in the introduction or as perspective.

8)
There are some typos, for instance, p. 9 "high electron negativity" should be "high electronegativity"; Pt-support interaction energy should be called "binding energy" not "adsorption energy". The latter is reserved for molecules (e.g. CO, O2, CO2) that adsorb and desorb on the Ptn/support surface.

Reviewer #2 (Remarks to the Author):
This is a high-quality manuscript on the CO Oxidation reaction, where the authors try to provide a structure-activity relationship. Such knowledge is important and may be useful for purposeful catalyst design. Nevertheless, there are several issues requiring clarification. i) A volcano-type dependence of catalyst activity on the number of Pt atoms in different clusters was established (Fig. 1b, Fig.3 g,h and Fig. 5 b,c). It is, however, not correct to relate the activity to the total number of Pt sites for the catalysts with 3-D structured Pt species. In addition, it is not clear why the ratio of neutral to cationic Pt sites should be related to the total number of Pt atoms (See the right Y axis in Fig.3 h and Fig. 5 b,c). ii) To support their conclusions, the authors should report surface coverage by adsorbed carbon monoxide and oxygen species before starting the reaction. Can the authors exclude the fact that the ratio of adsorbed carbon monoxide to oxygen species depend on the size of Pt clusters? What is about the kind of adsorbed oxygen species? Does oxygen exist in molecular or atomic forms or their mixtures?
iii) The authors reported desorption profiles of CO in Fig. 1d. What is about oxygen? iv) A weak point of the experimental part of this manuscript is that the authors do not provide the rate of CO oxidation. Can the authors calculate such values from their data? v) Will the authors observe the same activity-size dependence under steady-state conditions? We thank the reviewers for their comments and insightful remarks. In this point-by-point response letter, we respond to these comments based on the experimental and calculated data in manuscript, including new information that has been added in response to the reviewers All the points made here are reflected in the revised manuscript.

Responses to the comments from reviewers CO oxidation activity of non-reducible oxide-supported mass-selected few-atom Pt nanoclusters
The reviewers' comments are shown in italic bold text. In our responses, the text shown in red indicates added or revised sentences. Significant changes have been made in the manuscript text, including updated figure numbering; thus, in this letter, we refer to figures using the numbering in the revised manuscript. The references cited in this letter are listed at the end of this document.

However, they provide clear evidence that the CO oxidation activity is correlated with the ratio Nnutral/Ncation/Ntotal, which is interesting and provides insight for understanding the role of various metal sites with different oxidation states or charge states. I therefore think the manuscript might become publishable for Nature Communications after taking into account of the following comments and suggestions:
Response: We are grateful that the reviewer understands the importance of our work. Thanks to the reviewer's comments, the clarity of our work has been improved, as explained in the following point-by-point responses. (e.g. J. Am. Chem. Soc., 2017, 139,

6190). They need to clarify this point because when the metal cluster and the support have similar chemical potentials, there is no cationic atoms, even at the interfacial region.
Response: We thank the reviewer for this comment regarding the charge transfer between the clusters and the support. Vila et al. have demonstrated charge transfer from Pt to an Al 2 O 3 support, providing evidence for the existence of cationic atoms as shown in below 1 .
As we did not discuss charge transfer in the previous manuscript, to improve clarity, we revised the manuscript as follows. On line 182, 'This suggestion is reasonable because the coordination number of adsorbates has been shown to affect the adsorption energy 2 and because charge transfer from Pt to Al 2 O 3 has been demonstrated 1 .'

3) Conceptually, valence state, oxidation state, and atomic charge are quite different concepts. The authors should avoid mixing them together. The high oxidation state metal (e.g. Pt(II) or Pt^+2) often carries more positive charge than a low oxidation state metal (e.g. Pt(0)). But one cannot confuse oxidation state with charge state. In p. 7, "Pt within the Pt7 cluster has an oxidation state between Pt0 and Pt2+". Here Pt2+ is the charge state, while Pt(II) or Pt+2 is the oxidation state.
Response: We thank the reviewer for this comment on the use of the terms oxidation state and charge state. We revised all the relevant descriptions in the manuscript accordingly.

Nnutral/Ncation/Ntotal needs some further explanation of its physical meaning. In fact, Nnutral/Ncation/Ntotal = Nnutral/(NcationNtotal) = (Ntotal-Ncation)/(NcationNtotal) = 1/Ncation + 1/Ntotal ; Because the Ntotal is usually far larger than Ncation, one can imagine that this simply says that the CO oxidation activity is inversely proportional to the Ncation.
Response: In this study, we obtained experimental evidence for the existence of cationic Pt atoms on an Al 2 O 3 surface and we found that N n /N c is correlated with the CO oxidation activity. We also found that N n /N c divided by the number of atoms in the cluster shows a stronger correlation to the CO oxidation activity; however, the physical meaning of this parameter (division by n,) was unclear in the previous manuscript.
Through our attempts to find the physical meaning of this parameter (such as by transforming the equation as the reviewer also kindly attempted), we noticed that N n /N c is the atomic ratio of the neutral and cationic Pt atoms exposed on the Pt n /Al 2 O 3 sample surface. Thus, N n /N c should be simply compared with the amount of CO 2 produced from that Pt n /Al 2 O 3 surface according to the different cluster sizes.
We deposited a certain amount of Pt atoms (~0.02 ML for all the Pt n /Al 2 O 3 samples) on the Al 2 O 3 surface, and some of them appeared on the surfaces of clusters, which we defined as surface-exposed Pt atoms. It is logical to think that the CO 2 molecules detected by TPR were produced on these surface-exposed Pt atoms. Thus, if there are differences in the oxygen affinities of the neutral and cationic atoms, there should be some correlation between the atomic ratio of these sites and the amount of produced CO 2 . For this reason, we think N n /N c should be compared with the amount of CO 2 produced for different n values of Pt n /Al 2 O 3 (as shown in Fig. 3g in the revised manuscript). An excellent R 2 value of 0.99 was achieved for the curves of N n /N c and the amount of CO 2 produced.
This consideration should also be applied to predict the ideal cluster size, and thus, Figs. 5b and 5c were also amended to compare the amounts of CO 2 produced with the ratios of neutral to cationic Pt atoms theoretically predicted using our BAM. These plots also exhibited higher R 2 values than those in the previous manuscript.  Fig.3 h and Fig. 5 b,c).

In addition, it is not clear why the ratio of neutral to cationic Pt sites should be related to the total number of Pt atoms (See the right Y axis in
Response: We thank the reviewer for this comment. We agree with the reviewer that it is not correct to relate the activity to the total number of Pt atoms because there are inaccessible atoms in the 3D structured Pt clusters.
Thanks to this comment, we noticed that that N n /N c is the atomic ratio of the neutral and cationic Pt atoms exposed on the Pt n /Al 2 O 3 sample surface. Thus, N n /N c should be simply compared with the amount of CO 2 produced from that Pt n /Al 2 O 3 surface according to the different cluster sizes.
We deposited a certain amount of Pt atoms (~0.02 ML for all the Pt n /Al 2 O 3 samples) on the Al 2 O 3 surface, and some of them appeared on the surfaces of clusters, which we defined as surface-exposed Pt atoms. It is logical to think that the CO 2 molecules detected by TPR were produced on these surface-exposed Pt atoms. Thus, if there are differences in the oxygen affinities of the neutral and cationic atoms, there should be some correlation between the atomic ratio of these sites and the amount of produced CO 2 . For this reason, we think N n /N c should be compared with the amount of CO 2 produced for different n values of Pt n /Al 2 O 3 (as shown in Fig. 3g in the revised manuscript). An excellent R 2 value of 0.99 was achieved for the curves of N n /N c and the amount of CO 2 produced. This consideration should also be applied to predict the ideal cluster size, and thus, Figs. 5b and 5c were also amended to compare the theoretically predicted amount of CO 2 produced using our BAM and experimental one. These plots also exhibited higher R 2 values than those in the previous manuscript. Figures 3 and 5 and the corresponding figure captions were revised as follows. to cationic Pt atoms. The coefficient of determination (R 2 ) between these parameters was calculated to be 0.99 using the least-mean square method.

ii) To support their conclusions, the authors should report surface coverage by adsorbed carbon monoxide and oxygen species before starting the reaction.
Response: In accordance with the reviewer's comment, the coverages of both CO and O are explicitly reported in the revised supplementary information. We did not detect O 2 desorption during the TPR measurements, as in a previous report 10 . Thus, based on this previous report, the O coverage was calculated by assuming that all the adsorbed O was consumed by CO during the reaction 10 . To report the coverages, we revised the manuscript as follows.
On line 112, 'The conversion efficiency from adsorbed CO to CO 2 and the total amount of adsorbed CO and O for each cluster size are summarized in Fig. 1f and Supplementary Figs. 3a,b, respectively.'