Ordered clustering of single atomic Te vacancies in atomically thin PtTe2 promotes hydrogen evolution catalysis

Exposing and stabilizing undercoordinated platinum (Pt) sites and therefore optimizing their adsorption to reactive intermediates offers a desirable strategy to develop highly efficient Pt-based electrocatalysts. However, preparation of atomically controllable Pt-based model catalysts to understand the correlation between electronic structure, adsorption energy, and catalytic properties of atomic Pt sites is still challenging. Herein we report the atomically thin two-dimensional PtTe2 nanosheets with well-dispersed single atomic Te vacancies (Te-SAVs) and atomically well-defined undercoordinated Pt sites as a model electrocatalyst. A controlled thermal treatment drives the migration of the Te-SAVs to form thermodynamically stabilized, ordered Te-SAV clusters, which decreases both the density of states of undercoordinated Pt sites around the Fermi level and the interacting orbital volume of Pt sites. As a result, the binding strength of atomically defined Pt active sites to H intermediates is effectively reduced, which renders PtTe2 nanosheets highly active and stable in hydrogen evolution reaction.


Li and coworkers' manuscript presents a nice research work regarding the ordered clustering
of single atomic Te vacancies in atomically thin PtTe2 for electrocatalytic application. In this manuscript, the authors first prepare atomically thin 2D PtTe2 nanosheets with well-dispersed single atomic Te vacancies by electrochemically exfoliating bulk PtTe2 crystals. Then, heat treatment induces the migration of the random Te-SAVs to form ordered Te-SAV clusters. This discovery is interesting and crucial because accurately tailoring the vacancy structure at the atomic level to improve the intrinsic catalytic activity is generally quite difficult. When used as electrocatalysts for HER, the heat-treated PtTe2 materials with Te-SAV clusters reveal significantly improved catalytic activity, much better than that of the state-of-the-art Pt (1) What is the purpose of presenting C mapping in Fig. 1g? If this figure shows that TBAB or its decomposition products on the surface of exfoliated PtTe2 are removed, the authors should give the corresponding description in the main text.

Response:
We thank the reviewer for the positive comments and helpful suggestion. According to STEM elemental mapping results displayed in Fig. 1g, we can see that only C signal from the carbon-coated copper grid is detected after long-time signal collection. No signal of C is detected on the surface of exfoliated PtTe2. The purpose of presenting C mapping in Fig. 1g is to show that TBAB and/or its decomposition products on the surface of exfoliated PtTe2 are removed. According to the reviewer's suggestion, the related descriptions have been added in the revised manuscript and marked as blue.
(2) Descriptions and evidence regarding single vacancies are Te vacancies rather than Pt vacancies need to be added into the manuscript because this point is vital for the whole work.
Response: We thank the reviewer for the comment. The evidences regarding single vacancies are Te vacancies rather than Pt vacancies come from the following aspects: (1) The molar ratio of Pt : Te in the prepared PtTe2 catalysts (0.61) is much larger than the theoretical stoichiometric ratio of PtTe2 (0.5) according to the ICP-OES results (Supplementary Table 1), from which the atomic percentage of Te vacancies in the prepared PtTe2 catalysts is calculated to be 18 at.%.
(2) Vacancies can be directly visualized with distinguishable contrast in the few-layer PtTe2 NSs at the lower left corner of Fig. 1f. The corresponding line intensity profile (inset in the upper portion of Fig. 1f) combined with the atomic model (inset at the lower right corner of   Supplementary Figs. 3 and 4), the bulk PtTe2 crystals are exfoliated into the atomically thin two-dimensional PtTe2 nanosheets with fully exposed undercoordinated Pt sites (Fig. 1d). As a result, PtTe2 NSs show substantially larger geometric HER current density than bulk PtTe2 crystals at the same overpotential. Similar results are also reported in other materials, such as NiPS3 7 and Sb 8 . According to the reviewer's suggestion, the related references to support this conclusion have been added in the revised manuscript.
(4) Defects engineering is a common strategy to improve the catalytic activity of nanomaterials.
Previously published work on defect engineering in TMDs materials usually focuses on improving the number of vacancies and doping vacancies via heteroatoms. Accurately tailoring vacancy structure in TMDs materials at the atomic level to improve the intrinsic is seldomly reported. The authors can further highlight this point in the main text.

Response:
We thank the reviewer for the helpful suggestion. Vacancy engineering is a vital strategy to modulate the surface electronic structure of electrocatalysts to improve their catalytic activities [9][10][11][12] . Many electrocatalysts with different vacancies have been prepared and applied in various electrocatalytic reactions [9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24] . Nevertheless, the diversity and complexity of vacancies in these catalysts prevent the in-depth understanding of the vacancy-catalysis relationship. As a result, there are still many open questions that have not been answered in this field. For example, how to precisely create atomic vacancies with a well-defined structure?
How to rationally regulate vacancy structure to improve the intrinsic electrocatalytic activity?
What is the underlying correlation between vacancy and catalytic activity? and etc.
Many synthetic methods, such as controlled growth 13,14 , etching [15][16][17][18][19]24 , heat and plasma treatment [20][21][22] , and template synthesis 23 , have been applied to prepare electrocatalysts with vacancies. However, up to now, it is still challenging to accurately regulate specific vacancies with atomic precision in materials as the conventional synthesis methods typically have no selectivity to particular sites 9,16,19,25 , making the study of vacancy-catalysis relationship very challenging. Furthermore, the vacancies in the materials are usually randomly dispersed without forming a well-defined structure 13 . Different from traditional works, our work shows that a large amount of uniformly dispersed single atomic Te vacancies can be created in PtTe2 crystals by chemical vapor transport method. Besides, these uniformly dispersed Te vacancies with the well-defined structure can be fully exposed after electrochemically exfoliating the bulk PtTe2 crystals. More importantly, previous works on vacancy engineering mainly focus on regulating the number of vacancies as well as doping vacancies via heteroatoms 17,18,20,[25][26][27][28][29] .
Accurately tailoring vacancy structure at the atomic level to improve the intrinsic catalytic activity remains a grand challenge. Our work shows that single atomic vacancies can migrate, and form ordered atomic vacancy clusters with the trigonal structure, which effectively regulates the bindings between the reaction intermediates and catalytic sites and improves the intrinsic catalytic activity. According to the reviewer's suggestion, the related descriptions have been added in the revised manuscript and marked as blue.  Response: We appreciate the reviewer for the insightful comments and suggestions. According to frontier orbital (FO) theory, the interacting intensity between orbitals of catalyst and adsorbent can be determined by three factors, namely accessible energy, matchable symmetry and sufficient overlapping. If the energy and symmetry maintain the same, the overlapping of two interacting orbitals from catalyst and adsorbent will become a decisive factor to impact their interacting intensity. In this work, the interacting orbitals of Pt atoms are represented by partial electron density around the Fermi level of PtTe2 system 32 . A larger iso-surface (i.e. the larger volume enclosed by the surface) indicates a higher electron density around the corresponding atoms when the iso-value is fixed, thus can give rise to more effective overlapping with the electron density (i.e. the FO orbital) of adsorbent with stronger adsorption.
In our work, the adsorption of H on Pt atoms in defective PtTe2 gradually decreases when the neighboring Te-SAVs approach to each other owing to their gradually decreased orbital volume.
According to the reviewer's suggestion, the related descriptions have been added in the revised manuscript and marked as blue.
Although the d-band center theory has been commonly used to predict reactivity of metal or cluster systems, its adaptation in single-atom and compound system is still dubious. The distinction may come from different orbital contribution of metal atoms (adsorbing site) in different systems [33][34][35] . Actually, this descriptor can even produce contradictory results for some metal alloys with complicated structures 36 . In general, for most metal (or simple surface alloy) systems, the d-orbitals of metal atoms will hybridize with each other strongly. It also can simultaneously participate in the interaction with orbitals of adsorbent. Thus, an average weight of d-band (d-band center) can be useful to evaluate the reactivity of metal atoms 37 . However, for substrate supported single-atom system, the orbital distribution will be affected by latticefield of substrate, which usually results in the slight hybridization with each other. As a result, the orbital overlapping between metal atom and adsorbent has to consider their wave function matching, and always only part of d-orbitals show dominant role in the interaction, under which the d-band center may not be suitable any more 33 . The compound containing M-X covalent bond also follows similar principle as single-atom system, since d-orbitals with different spatial directions show distinctive contribution to the interaction strength.  We agree with the reviewer that PtTe based catalysts have been studied as electrocatalysts for HER over recent years [53][54][55]  However, as stated in this paper, the v-Pd3Pt29Ru62Te6 AS consists of PtPdRuTe alloy and a small amount of PtTe2, rather than the pure PtTe2 material, which is demonstrated by their XRD result as shown in Fig. S1. Thus, the atomic structure of this material is very complex, making it difficult as a model catalyst to investigate the structure-property relationship and study the catalytic mechanism at the atomic scale. Completely different from the traditional PtTe-based materials, Pt and Te atoms in PtTe2 in our work crystallize in an atomically precise CdI2-type trigona (1 T) structure (P3m1, a = b = 4.03 Å, c = 5.33 Å, Supplementary Fig. 1) 56 , rather than are randomly distributed, which have been demonstrated by the XRD in Fig. 1c, highresolution HAADF-STEM image and atomic model in Fig. 1f. Consequently, this atomicallythin 2D PtTe2 material offers an ideal model catalyst for providing atomic-level insights into the Pt active site-catalytic performance relationship due to its well-defined atomic structure.
For the aspect of catalytic performance, we agree that our PtTe2 catalyst's HER performance is not the best in the literature. However, as shown in Fig. 3e 76 , which is obviously a mistake because it has been reported that the Pt electrode can be dissolved from the oxidation reaction (Pt 0 to Pt 2+ ) and the dissolved Pt 2+ can transfer to the working electrode surface [77][78][79] . Apart from the activity, catalytic stability is equally important. Accelerated cyclic voltammetry (CV) test elucidates the robustness of PtTe2-600 NSs in HER catalysis with virtually unchanged polarization curves even after 20,000 CV cycles (Fig. 3h). Furthermore, long-term chronopotentiometry measurement shows a slight change of overpotential after 24 h at a constant current density of 200 mA cm -2 for PtTe2-600 NSs, much better than that of Pt/C catalyst (Fig. 3i).
The novelty of our work is that we designed and prepared atomically thin 2D PtTe2 nanosheets with exposed and stabilized undercoordinated Pt sites as an atomically controllable Pt-based model catalyst to understand the correlation between electronic structure, adsorption energy, and catalytic property of atomic Pt sites. Also, we found that thermal treatment could drive migration of the well-dispersed single atomic Te vacancies in PtTe2 to form thermodynamically stabilized, ordered trigonal well-dispersed single atomic Te vacancy clusters. This finding provides a new strategy to engineer geometrically well-defined active sites via clustering of atomic defects. What's more, the bulk PtTe2 crystals synthesized by chemical vapor transport technique have closely stacked lamellar architecture with interlamellar spacing of 3.52 Å (Supplementary Fig.   2 Supplementary Figs. 3 and 4), the bulk PtTe2 crystals are exfoliated into atomically thin two-dimensional PtTe2 nanosheets (average thickness: ~3 nm, Supplementary Fig. 5) with a lot of exposed coordinatively unsaturated Pt atoms. Only in this electrochemical exfoliation stage, the exposed surface area can be significantly enhanced. Such a result has been reported in other materials, such as NiPS3 7 and Sb 8 . The following heat treatment will not further exfoliate the PtTe2 sheets so that the heated PtTe2 nanosheets will remain the similar surface area to the original PtTe2 sheets. In addition, in the experimental part for HER, we used the same catalyst loading amount (0.336 mg cm -2 ) for all PtTe2 based materials.

and 3). After electrochemical exfoliation (detailed in
Indeed, the total electrode activity is not only determined by the quantity of active sites, but also by the intrinsic activity of active sites 93 . For HER, TOF and overpotential based on electrochemically active surface area (ECSA) can be employed to reflect the intrinsic activity of the active sites in catalysts 60,62,67,70,82,91,[94][95][96]  NSs migrate to form ordered Te-SAV clusters, which effectively reduces the adsorption strength of hydrogen on the Pt sites and thus drastically enhances the hydrogen evolution performance. More importantly, we also experimentally performed scanning tunneling spectroscopy (STS, dI/dV versus V) to probe the local DOS over different vacancy structures.
The results show that near-Fermi local DOS of defect-free site, Te-SAV and trigonal Te-SAV in PtTe2 (marked to red, orange, and blue symbols, respectively, in Fig. 2d) gradually decreases ( Supplementary Fig. 28c). The result is consistent with the established theory calculation, which indicates a decreased DOS of Pt sites after the clustering of regular Te-SAVs ( Fig. 4d and Supplementary Fig. 27).
From the above results and discussion, it can be concluded that 18 at.% of Te defect sites and 8.7 wt.% of coordinatively unsaturated Pt atoms in PtTe2 nanosheets are obviously enough to influence the materials' overall catalytic activity. More importantly, experimental results also show that the intrinsic activity of coordinatively unsaturated Pt atoms in PtTe2 is significantly enhanced after the migration of the single atomic Te vacancies to form ordered Te vacancy clusters. Thus, the theory we have built up to explain the enhanced activity is reasonable.  (Fig. 3) clearly indicates the enhancement of HER performance with respect to temperature. It is a clear indication of enhanced crystallinity and conductivity, which enhances the activity. The improved activity is not related to defect sites; instead, it is related to crystallinity and conductivity (As the crystallinity is enhanced after annealing at 600  C), as Pt is a well-known the most active HER catalyst.

Response:
We thank the reviewer for the comments, but it is too arbitrary for the reviewer to draw such a conclusion. The LSV curves of the samples recorded in 1.0 M KOH (Fig. 3) indicate enhancement of HER performance with respect to temperature. However, this result is completely not an indication of enhanced crystallinity and conductivity. Indeed, our bulk PtTe2 crystals are synthesized at 1000 ˚C for 48 h and then 1150 ˚C for another 1 h in furnace (details in experimental part). The obtained crystals already have good crystallinity as demonstrated by the XRD result as shown in Fig. 1c. Electrochemical exfoliation method also has been proved as an efficient strategy to produce atomically thin 2D materials with high quality and crystallinity [98][99][100][101][102] . Thus, after exfoliating the bulk PtTe2 crystals, the obtained atomically thin PtTe2 NSs still have good quality and crystallinity, which was already demonstrated by the high-resolution HAADF-STEM image in Fig. 1f. The following heat treatment which causes migration of the random Te-SAVs to form ordered Te-SAV clusters is below 600 o C, much lower than the crystal growth temperature (1000-1150 o C).
To further exclude the possibility that the improved activity is related to crystallinity, we conducted the following experiments. Firstly, 4 mg of the exfoliated PtTe2 NSs were dispersed in 1 ml of ethanol with 8 μl of Nafion 117 solution and sonicated for ~2 minutes to obtain a homogeneous catalyst ink. Then, 25 μl of the catalyst ink was dropped onto two carbon papers (Toray Paper 060, 1cm  1cm) with the same area (1cm  0.5cm) and loading amount (0.198 mg cm -2 ). After completely drying, one carbon paper with exfoliated PtTe2 NSs was heated under vacuum (1  10 -9 mbar) in the STM equipment at only 120 o C for 2 h. The randomly dispersed Te-SAVs in PtTe2 can also partially migrate and form ordered Te-SAV clusters after the low-temperature treatment in this condition (1  10 -9 mbar, 120 o C for 2 h), as demonstrated in Fig. R3a. At last, we compared the HER performance of the two materials. Here, the treated sample is denoted as PtTe2 NSs-120 o C in vacuum, and the original sample is denoted as PtTe2 NSs. Obviously, these two samples have nearly similar crystallinity. As shown in Fig. R3b, the sample PtTe2 NSs-120 o C in vacuum obviously exhibits the significantly higher HER activity as compared to PtTe2 NSs. From the above results, it can be concluded that the improved activity should be related to the ordered clustering of single atomic Te vacancies, rather than the crystallinity.
For conductivity, it should be noted that when the thickness increases from monolayer to bilayer, PtTe2 evolves from semiconductive (bandgap: ~0.84 eV) to metallic. The electronic structure of trilayer PtTe2 has already resembled to its bulk material (Supplementary Figs. 19 and 20). Since metallic nature is of pivotal importance for hydrogen adsorption, we have performed the atomic force microscopy measurements, which show that the PtTe2 NSs have thicknesses about 0.6-6 nm (average thickness: ~3 nm) ( Supplementary Fig. 5). Most of the exfoliated PtTe2 NSs have more than 3-layer structures as the theorical thickness of the monolayer PtTe2 is about 0.67 nm, suggesting that PtTe2 NSs have good conductivity. Thus, the conductivity should not be the decisive factor for the significantly imported HER performance.
As we mentioned in our reply to question 1, the intrinsic activity of coordinatively unsaturated Pt in PtTe2 is significantly improved after the clustering of single atomic Te vacancies, which has been demonstrated by the TOF and overpotential based on electrochemically active surface area (ECSA) 60,62,67,70,82,91,[94][95][96]  Additionally, the Hupd peak of PtTe2-600 NSs is even more negative than that of the Pt/C catalyst. More importantly, we experimentally performed scanning tunneling spectroscopy (STS, dI/dV versus V) to probe the local DOS over different vacancy structures. The results show that near-Fermi local DOS of defect-free site, Te-SAV and trigonal Te-SAV in PtTe2 (marked to red, orange, and blue symbols, respectively, in Fig. 2d) gradually decreases ( Supplementary Fig. 28c). The result is consistent with the established theory calculation, which indicates a decreased DOS of Pt sites after the clustering of regular Te-SAVs ( Fig. 4d and Supplementary Fig. 27).
From these mutually corroborating data, it can be concluded that during heat treatment, the Te-SAVs in atomically thin PtTe2 NSs migrate to form ordered Te-SAV clusters, which effectively reduces the adsorption strength of hydrogen on the Pt sites and thus drastically enhances the hydrogen evolution performance. The improved activity is strongly related to the significantly improved intrinsic activity of coordinatively unsaturated Pt in PtTe2, and rather than crystallinity and conductivity (as they are already good enough). 3. The straightforward comparison between Pt/C (20%) with PtTe2 will not be fair, as the amount of Pt in PtTe2 is much higher than the Pt/C. It means the per unit mass activity of Pt/C is much better than the PtTe2.

Response:
We thank the reviewer for raising the question. Actually, when comparing the activity of Pt/C (20 wt.%) with PtTe2, we kept the amount of Pt in PtTe2 on glassy carbon electrode (GCE) the same as that in Pt/C, which means that the per unit mass activity of Pt/C is the same as that in PtTe2. As already stated in the experimental part, before loading catalyst, GCE was successively polished with 1.0, 0.3, and 0.05 mm Al2O3 slurry to obtain an ultraclean surface. Afterwards, 4 mg of catalysts were dispersed in 1 ml of ethanol with 8 μl of Nafion 117 solution and sonicated for ~2 minutes to obtain a homogeneous catalyst ink. For commercial Pt/C (20 wt.%), 40 μl of the catalyst ink was dropped onto the GCE and allowed to dry at room temperature. Thus, the total mass loading of Pt is 0.162 mg cm -2 . For PtTe2 (nPt:n Te=0.61), 16.6 μl of the catalyst ink was dropped onto the GCE and dried at room temperature.
The average loading of the catalyst is 0.336 mg cm -2 , and the Pt content is ~0.162 mg cm -2 , which is the same as that in Pt/C (20 wt.%) catalyst. The LSV curves based on mass activity of Pt in PtTe2-600 NSs and Pt/C are shown in Fig. R4a, we can see that the mass activity of Pt in PtTe2-600 NSs is obviously higher than that in Pt/C at the same potential. At -0.2 V vs. RHE, the mass activity of Pt in PtTe2-600 NSs is calculated to be 1.55 A mgPt -1 , while that is only 1.13 A mgPt -1 for Pt in Pt/C. Response: We thank the reviewer for the comments. We agree that our PtTe2 catalyst's HER performance is not the best in literature, and some Ru and Ir based catalysts exhibit better HER performance than our catalyst. Based on the reviewer's suggestion, we have added more Ru and Ir based catalysts with super HER activity in the Supplementary Table 2. However, as shown in Fig. 3e and Supplementary Table S2, PtTe2 catalyst still outperforms many of the reported highly active HER catalysts, especially Pt-based catalysts. It is commonly known that many factors can affect the activity measurement of HER, such as catalyst loading amount [57][58][59][60][61][62] , electrode preparation procedure, iR correction [63][64][65][66] , reference electrode calibration 60, 61, 67-70 , temperature [71][72][73][74] , scan speed 75 , and so on. However, for example, many published papers regarding catalysts with high HER activities even do not contain the information about reference electrode calibration, as marked in the Supplementary Table S2. In addition, some published works with very high HER activities even selected Pt wire as the counter electrode 76 , which is obviously a mistake because it has been reported that the Pt electrode can be dissolved from the oxidation reaction (Pt 0 to Pt 2+ ) and the dissolved Pt 2+ can transfer to the working electrode surface and is deposited on the electrode materials [77][78][79] . Apart from the activity, catalytic stability is equally important. Accelerated cyclic voltammetry (CV) test elucidates the robustness of PtTe2-600 NSs in HER catalysis with virtually unchanged polarization curves even after 20,000 CV cycles (Fig. 3h). Furthermore, long-term chronopotentiometry measurement shows a slight change of overpotential after 24 h at a constant current density of 200 mA cm -2 for PtTe2-600 NSs, much better than that of Pt/C catalyst (Fig. 3i).
As stated before, the novelty of our work is that we designed and prepared atomically thin 2D PtTe2 nanosheets with exposed and stabilized undercoordinated Pt sites as an atomically controllable Pt-based model catalyst to understand the correlation between electronic structure, adsorption energy, and catalytic property of atomic Pt sites. Also, we found that thermal treatment could drive the migration of the well-dispersed single atomic Te vacancies in PtTe2 to form thermodynamically stabilized, ordered trigonal well-dispersed single atomic Te vacancy clusters. This finding provides a new strategy to engineer geometrically well-defined active sites via clustering of atomic defects.

Reviewer #3 (Remarks to the Author):
My previous comments have been fully addressed. I strongly support the publication of this work.

Response:
We thank the reviewer for the recommendation to publish our work.