Non defect-stabilized thermally stable single-atom catalyst

Surface-supported isolated atoms in single-atom catalysts (SACs) are usually stabilized by diverse defects. The fabrication of high-metal-loading and thermally stable SACs remains a formidable challenge due to the difficulty of creating high densities of underpinning stable defects. Here we report that isolated Pt atoms can be stabilized through a strong covalent metal-support interaction (CMSI) that is not associated with support defects, yielding a high-loading and thermally stable SAC by trapping either the already deposited Pt atoms or the PtO2 units vaporized from nanoparticles during high-temperature calcination. Experimental and computational modeling studies reveal that iron oxide reducibility is crucial to anchor isolated Pt atoms. The resulting high concentrations of single atoms enable specific activities far exceeding those of conventional nanoparticle catalysts. This non defect-stabilization strategy can be extended to non-reducible supports by simply doping with iron oxide, thus paving a new way for constructing high-loading SACs for diverse industrially important catalytic reactions.


Line 14Start line with 'The'
Response: Thank you for the suggestion, "The" has been added in the revision.
Action: Manuscript amended 4. For the computational work, explain how they get free energies using VASP as this requires vibrational partition functions for solid and gas phase species. This is not trivial to do. Please The standard Gibbs free energies of G O2(g) , G Pt(g) , and G PtO2(g) were calculated using the following equations, taking into account the individual translational E t and S t , vibrational E v and S v , rotational E r and S r , and ZPE contributions: For slab models, the entropy and enthalpy corrections to free energies are neglected in this work.

G = H -TS = U + k b T -TS S = S t + S v + S
The resulting corrections to the ZPE, H, S, and G at various temperatures are given below: and nanoparticles (0.1 s -1 vs 0.08 s -1 ), indicating a common active site, and hence the superior specific activity of the single-atom catalyst reflects its improved atom efficiency (every Pt atom directly activates methane). 5 We respectfully disagree that methane combustion to CO 2 is not an 'interesting' reaction; natural gas power plants accounted for ~25 % of global energy production in 2015 6 and their contribution is expected to rise over coming decades. Methane combustion is thus one of the primary means of human energy production, and least polluting of fossil fuels. Although the reaction temperature of 700 °C required for in situ genesis of our single-atom catalyst (SAC) is high compared to those employed in commercial catalytic methane combustion, the specific activity of our  Figure 5, which was subjected to 6 h methane combustion at 700 °C. The resulting STEM confirmed that the initial NPs were completely dispersed into single atoms and remained atomically-dispersed for extended periods at high temperature ( Figure 5, and Supplementary Figure S14). We also compared our catalyst with a previously reported Pt/Al 2 O 3 catalyst, 7 which reveals our Pt 1 /Fe 2 O 3 SAC is 20 times more active (Supplementary Table S4). These aspects are now described in the manuscript and supplementary information. Statement on pg 2 that SAC stability is intrinsically linked to the density and stability of defects is not justified, ignores lots of literature and is not supported by literature generally.

Response:
We thank the reviewer for the comments and suggestions, although we respectfully disagree with some of these as discussed below.
Experimental evidence for Pt adatoms coordinated to 4 surface oxygens in our SAC is in fact strong and clearly evidenced by EXAFS fitting (Supplementary Figure S5B and Supplementary  Fig. 2 of Science 2017, 358, 1419 9 , and Fig. 1e of ACS Catal. 2017, 7, 887 10 ). Any small variations in the precise local coordination environment of isolated Pt atoms in our high loading Pt/Fe 2 O 3 SAC are therefore of far less scientific significance than the extremely high specific activities offered by our new synthetic protocol.
We agree that methane conversion data alone are not sufficient to describe the intrinsic reactivity of our SAC, and now report and briefly describe the corresponding TOFs in the manuscript and a new Supplementary  14 5 Pt/C defect 15 6 Pd/C defect 16 7 Pt/C vacancy 17 8 Fe,Co,Ni/C vacancy 18 9 Pd/C vacancy 19 10 Co-Pt/C defect 20 11 Pd/mpg-C 3 N 4 "six-fold Cavities" 21 12 Ir,Au,Pd,Ag,Pt/C defect 22 13 Pd/graphene vacancies 23 14 Pt/graphene defects 24 15 Pt,Co, In/graphene vacancies 25  Preparation, characterization and catalytic performance of single-atom catalysts. 58 2 Increasing the range of non-noble-metal single-atom catalysts. 59 3 Two-dimensional materials confining single atoms for catalysis. 60 4 Atomically Dispersed Supported Metal Catalysts. 61 5 Single-Atom Catalysts: A New Frontier in Heterogeneous Catalysis. 62 6 Catalysis by Supported Single Metal Atoms. 63 7 The Power of Single-Atom Catalysis. 64 8 Atomically dispersed supported metal catalysts: perspectives and suggestions for future research. 65 9 Single-Atom Electrocatalysts. 66 10 Metal Catalysts for Heterogeneous Catalysis: From Single Atoms to Nanoclusters and Nanoparticles. 67

Response:
We thank the reviewer for the suggestion and now include EXAFS data for both samples in Figure 1E and a new Figure 2. EXAFS fits and fitted parameters for the 0.3Pt/Fe 2 O 3 -C800 sample appear in a new Figure S5 and Supplementary Table S1. All these EXAFS data evidence the complete dispersion nanoparticles as isolated Pt atoms, with only Pt-O nearest neighbor scatters (no Pt-Pt contributions).

Action: Manuscript+ESI amended
2) It is found that, Pt nanoparticles cannot be redispersed on Al2O3 after hightemperature calcination in air. If comparing the initial H2PtCl6/Al2O3 (shown in Figure S7) and H2PtCl6/Fe2O3 (shown in Figure S13), the sizes of Pt particles are different in those two cases. The initial size of Pt particles in the H2PtCl6/Fe2O3 sample is smaller. It can be expected that, the redispersion dynamic of Pt nanoparticles into Pt single atoms are related with their particle size, which has also been reported in a recent paper (Nature Communications 9 (1), 574.). Therefore, to have a fair comparison, I would like to suggest to author to load the Pt-PVP nanoparticles on Al2O3 (as done with Fe2O3, in Figure   S3) and then treat the sample by calcination in air at 800 oC to see if those Pt nanoparticles can be re-dispersed on Al2O3.

Response:
The reviewer makes an excellent suggestion. We have therefore loaded 0.3 wt% of colloidal Pt NPs on Al 2 O 3 and calcined these at 800 °C in flowing air precisely as for Fe 2 O 3 (synthetic procedure in the Methods-Pt/Al 2 O 3 -NP and Pt/Al 2 O 3 -C800). The resulting TEM images of this Pt/Al 2 O 3 -NP material before and after calcination now appear in Figure S7 E-H.
The initial Pt particle size was about 2-3 nm (the same as our Pt/Fe 2 O 3 -NP catalyst). Following calcination, severe nanoparticle aggregation was observed resulting in Pt agglomerates >10 nm.
This experiment confirms that Pt nanoparticles cannot be dispersed on Al 2 O 3 by high temperature calcination, irrespective of their initial particle size. We note that dispersion of <1 Those unsaturated sites can be anchoring sites for Pt atoms. According to the data presented in this manuscript, the positions of Pt atoms on Fe2O3 cannot be determined. Therefore, I will suggest the authors not to use "non-defect" to describe the support and the corresponding redispersion behaviour unless they can provide more proof or more discussion on that point.

Response:
The reviewer raises an important consideration. We must clarify that our title "Non defect-stabilized," is not intended to suggest that our supports are defect-free, but rather that any defects present do not play a significant role in stabilizing single-atoms. This assertion is made on the basis that the defect concentration is far lower than the concentration of single-atoms. We agree that our as-synthesized Fe 2 O 3 will contain some defects; however, after 800 °C calcination the intrinsic defect concentration is exceedingly small (~10 -11 level) 72 far below our Pt metal loading level (~1 %). If Pt single-atoms were solely stabilized by defects, then the maximium metal loading at which a SAC could be prepared would be extremely small (many orders of magnitude lower than we experimentally observe). Hence in our system, Pt single-atoms are not associated with defects. In any event, we have attempted to quantify the oxygen anion vacancy The discovery that single-atom formation can be decoupled from a requirement for surface defects is highly significant, and underpins this work, and hence our preference is to retain the existing title. The fourth sentence of the Abstract clarifies any possible ambiguity "Here we report that isolated Pt atoms can be stabilized through a strong metal-support interaction that is not associated with support defects".

Action: Manuscript+ESI amended
4) The in situ TEM experiments have been carried out to investigate the redispersion behaviour of Pt nanoparticles on Fe2O3, as presented in Figure 2. Please, make a careful look into the images, it seems that some of the nanoparticles (especially in the middle area) slightly grow.
Does it mean the Ostwald ripening mechanism also occurs? The size distributions of the Pt nanoparticles should be provided for the fresh sample and the sample after 20 min at 800 oC to see the structural evolution.

Response:
The reviewer raises an interesting question and suggestion. The size distribution of Pt nanoparticles is now added to Figure 3C and F by measuring every particle in Figure 3A and 3D according to this suggestion. Although two NPs in these images grew slightly after heating at 800 °C for 20 min, the majority shrank or even vanished. Consequently, the total particle number dropped from 300 to 200, and the average particle size decreased by 0.3 nm (from 3.0 nm to 2.7 nm). These changes were indeed accompanied by the growth of a few particles, suggesting that Ostwald ripening may compete with particle dispersion. This is unsurprising since migrating molecular PtO 2 species can be either trapped by the support or through encounters with large Pt NPs. However, prolonged heating resulted in the complete dispersion of all Pt NPs as singleatoms, evident in ex situ TEM images ( Figure S8A and B) and from EXAFS ( Figure 2, Figure   S5, and Table S1). Control experiments using Na-free Fe2O3 as the support are recommended.

Response:
The reviewer raises an excellent question related to recent work from Prof. Flytzani- Stephanopoulos's group which show that alkali cations can stabilize surface hydroxyls and in turn stabilize metal (Au and Pt) single-atoms and subnanometer clusters. [73][74][75] We have not identified such a role for alkali in our systems, 76 Figure S16A and B). After calcination at 800 °C in air for 5 h, Pt NPs were dispersed as Pt single-atoms ( Figure S16C and D), confirming that the reducible Fe 2 O 3 support, and not presence of Na + , is responsible for stabilizing Pt single-atoms. These results are now summarized in the manuscript.
However, the surface area of our alkali-free Fe 2 O 3 (N) support was very small, < 1 m 2 /g, and so only able to fully disperse a lower Pt loading (0.2 wt%) as single-atoms compared with the Fe 2 O 3 prepared using a Na 2 CO 3 . Alkalis may therefore be beneficial in preventing structural collapse of the oxide support during high temperature calcination.