WS2 moiré superlattices derived from mechanical flexibility for hydrogen evolution reaction

The discovery of moiré superlattices (MSLs) opened an era in the research of ‘twistronics’. Engineering MSLs and realizing unique emergent properties are key challenges. Herein, we demonstrate an effective synthetic strategy to fabricate MSLs based on mechanical flexibility of WS2 nanobelts by a facile one-step hydrothermal method. Unlike previous MSLs typically created through stacking monolayers together with complicated method, WS2 MSLs reported here could be obtained directly during synthesis of nanobelts driven by the mechanical instability. Emergent properties are found including superior conductivity, special superaerophobicity and superhydrophilicity, and strongly enhanced electro-catalytic activity when we apply ‘twistronics’ to the field of catalytic hydrogen production. Theoretical calculations show that such excellent catalytic performance could be attributed to a closer to thermoneutral hydrogen adsorption free energy value of twisted bilayers active sites. Our findings provide an exciting opportunity to design advanced WS2 catalysts through moiré superlattice engineering based on mechanical flexibility.

· In Figure 1d, it is not clear to me what is actually being represented. Is this a schematic? Or is it based on calculations? The figure caption (and associated in-text discussion) is not very helpful in this regard.

·
In Figure 2a/b, the authors present evidence of the WS2 MSLs in the form of STEM images. However, presumably this was taken at one point across a macroscopic sample (e.g., in the electrochemical tests, the electrode seems to be on the cm2 scale). It would be useful if the authors comment on how representative this image is? Also, is the 13.82° twist angle observed for all WS2 MSLs? Or does each nanobelt/nanocone have a unique twist angle? · In Figure 3a: What is the scan rate? Electrolyte? Geometric electrode area? Also, have these been iR corrected? None of this is mentioned in the figure caption.

·
In Figure 3a (and SI), potential is spelled wrong on the x-axis. · In Figure 3b, it appears as if the Tafel slope for the green and yellow traces has been estimated over a very narrow range of current densities (not even 1 order of magnitude). Why is this the case when the author have clearly measured the currents over a larger range of current densities (i.e., in Figure 3a)? Also, how can the authors be confident of their assignment of Tafel slopes over such a narrow range?

·
My major concern with this work is how the data are normalised. Currents appear to be normalised to area (i.e., current density), but it is not clear to me what the area corresponds to or how it was measured. Is this the actual area of the active material? If yes, how was this estimated? Alternatively, if this is geometric area, then the statement "MSLs show superior catalytic activity for HER" is very misleading. Electrochemical activity can be benchmarked by measuring the exchange current density (j0, with units of mA cm-2) or the heterogeneous electron-transfer rate constant (k0, with units of cm s-1). Due to the complexity of multi-electron processes, j0 is almost always adopted for complex catalytic reactions such as the HER. In principle, j0 can be estimated from the Tafel plots in Figure 3b (by extrapolating the linear region back to zero overpotential), with the caveat that the active electrode area must be known. If this is not known or it cannot be measured, then the difference in the curves shown in Figure 3a/b could EITHER be due to changes in activity (i.e., a legitimate difference in j0) or could simply be due to differences in the roughness/porosity of the electrode. In principle, even a very 'inactive' catalyst could present a lower overpotential (hence higher apparent "activity") at a given geometric current density if the exposed surface area is high enough. Misleading statements about the "activity" of rough/porous electrodes is a major problem in the electrocatalysis literature, as it makes it very difficult to compare results from different labs. There have been many reviews that discuss this fact (e.g., 10.1021/cs500923c).

·
In Figure 3a/b, Pt/C is presented but never mentioned/discussed in the main text.

·
The form of analysis performed in Figure 3d assumes that the electrodes behave as "ideal" electrochemical capacitors (i.e., current scales linearly with scan rate). However, all three plots do not pass through the origin, (0,0), and some even show clear deviations from linearity (i.e., see the 1T'-WS2 NSs). Consulting the source data (Supplementary Figure 14), all three electrodes exhibit asymmetric I-E plots (i.e., the reduction current is larger in magnitude than the oxidation current) that are somewhat sloped. This may suggest that a charge-transfer reaction (i.e., Faradaic process) may also be occurring in this potential range (i.e., at E < 0.20 V, where the baseline is clearly sloped). Is it appropriate to perform such a simple form analysis with such a complex electrode architecture? Perhaps the authors could investigate how closely the electrodes mimic an "ideal capacitor" with electrochemical impedance spectroscopy? In addition, In Figure 3d, there is no explanation as to what the units on the y-axis (DJ0.2V) actually means.

·
The authors state: "The much smaller Tafel slopes of WS2 MSLs (40 mV decade−1) indicated that the kinetics of the electrochemical hydrogen evolution on WS2 MSLs was much faster than those of the 2H-WS2 NSs and 1T'-WS2 NSs (Fig. 3b)." Tafel slopes DO NOT indicate on kinetics, rather they can indicate on the reaction mechanism under some circumstances (i.e., for a well-defined, dimensionally stable electrode). As noted above, j0 would be an indicator of kinetics, but this has not been calculated.

·
The authors state: "Additionally, compared to 2H-WS2 and 1T'-WS2 NSs, the lower charge transfer resistance and rapidly electron transportation capability of the WS2 MSLs are confirmed by the electrochemical impedance spectroscopy (EIS) measurements ( Supplementary Fig. 11)." However, they do not offer any justification/discussion as to why the EIS measurements indicate this. They do not even fit the spectra or include an equivalent circuit. Given the high porosity of the electrodes in question, interpreting the EIS spectra is not straightforward.

·
The authors state: "The electrochemical double-layer capacitances (Cdl) were calculated to contrast the electrochemical surface area (ECSA) of 2H-WS2 NSs, 1T'-WS2 NSs and WS2 MSLs (Fig. 3d). The Cdl of WS2 MSLs (33.7 mF cm−2) was much higher than that of 1T'-WS2 NSs (21.2 mF cm−2) and 2H-WS2 NSs (7.2 mF cm−2), indicating that the WS2 MSLs possessed more fully exposed active sites for electrochemical hydrogen evolution." How were these values normalised? Geometric area? In principle, if the specific capacitance (Cspecific in F cm-2) of these materials was known, then the exposed surface area (or ECSA) could be estimated from these values (i.e., AECSA = Cdl / Cspecific). The ECSA however does not necessarily indicate on the number of exposed "active sites", as all sites, regardless of 'activity' (e.g., the basal and edge planes of 2H-WS2 plus the underlying carbon support) contributes to the non-faradaic current, whereas only certain sites (e.g., the edge plane of 2H-WS2) may dominate the HER catalysis. How was the carbon support corrected for when calculating the ECSA?

·
The authors use the terms "superhydrophilic" and "superaerophobic" throughout. It would be useful if they provide a definition of these terms for the reader (i.e., what distinguishes hydrophilic from superhydrophilic?).

·
In Figure 5, it is not clear to me what the numbers "1,2,3,4,5,6" refer to or what the arrows are indicating. It is also not clear why a 3D plot is necessary here? The pictures of the various active sites are also too small to see clearly, so overall it is very difficult to determine anything meaningful from this figure.

·
The authors state "We ascribe the activity enhancement to a combination of electronic, geometric, superaerophobic and superhydrophilic effects." Again, as stated above, in electrochemistry "activity" refers to electron-transfer kinetics. This statement is misleading, as the aforementioned "superaerophobic and superhydrophilic effects" influence the mass transfer of bubbles, rather than enhancing electron-transfer kinetics · The authors state "ΔGH is insensitive to the MSLs" which is counter to the argument that they presented in Figure 5 (e.g., "Computational predictions for the MSLs effect on the HER activity indicated that the active sits of W-edge and S-edge of twisted bilayers WS2 have much more appropriate ΔGH compared with normal bilayers WS2 in Fig. 5.")

·
The "Electrochemical Measurements." Section of the SI is not sufficiently detailed to enable the reader to repeat the measurements. For example, in the "supplementary methods section" the authors state "The full description of linear sweep voltammetry (LSV)and cyclic voltammetry (CV) tests have been shown in supplementary methods." Also "The value of electrochemical double layer capacitance (Cdl) Electrochemical active surface areas (ECSA) was calculated by measuring CV curves of samples." is very ambiguous. Further "Nyquist plots of three samples were measured in the frequency range from 100 Hz to 0.1 kHz at an open circuit potential of -350 mV." According to Figure 3a of the main text, -350 mV is well beyond the onset potential of the HER on all considered electrodes. Therefore it is simply impossible that -350 mV could correspond to the "open circuit potential" of the catalysts.

Reviewer #3 (Remarks to the Author):
In this manuscript, the authors reported the synthesis of large-scale WS2 Moiré superlattices (MSLs) through a one-pot hydrothermal approach, and demonstrated their catalytic hydrogen production performance. However, the evidence of Moiré superlattices is not sufficient and doesn't support the main theme of this article. Thus, the main claim in this work is weak. Upon a careful examination, I cannot recommend its publication in Nature Communications. Some detailed comments are listed as follows: 1) HRTEM images in Figure 2a and 2b seems have been applied too much filters during the data recording and processing, and the images didn't show a clear Moiré period even in a small region.
2) From the XRD results in Fig. S5, we can see the 1T′-WS2 NSs sample has a low crystallinity, to make a comparison, the standard PDF card should be presented in the same image.
3) The EIS measurements carried out at a large overpotential (-350 mV), which is not appropriate. The EIS measurements are better carry out at a small catalytic current region. 4) In line2, page 7. The "S-Mo-S" should be S-W-S.
We sincerely thank the referees for carefully reviewing our manuscript and their valuable comments, which certainly help us improve our manuscript. We also thank the editor for giving us such an opportunity to address the comments and revise the manuscript. The changes in the revised manuscript have been highlighted in red for your review. The point-by-point responses are presented below.
Finally, we sincerely express our gratitude to the referee for all the constructive comments and suggestions, which really help us to improve our understanding of the relation between structures and properties of WS2 MSLs. Following the referee's suggestions, we have conducted more experimental and theoretical analysis to provide more convincing evidences of WS2 MSLs and correspondingly modify the theoretical models to reflect the most possible sample condition. We believe that the quality of the revised manuscript has reached the requirements of Nature Communications.

Reply to Referee 1 and revisions made accordingly:
In this manuscript, the author reported a MSLs material WS2 by a facile one-step hydrothermal method. They also found this material has high HER activity. The study on the HER of WS2 are not novel, but the MSLs material for HER is interesting. Therefore, I would recommend publication after major revisions. My specific comments are as follows.
Response: Thanks for the reviewer's positive evaluation of this work and suggestion.
We have made revisions according to each comment, as summarized below.
1. The author called the WS2 they synthesized as moiré superlattices (MSL) material, but they didn't provide sufficient evidence to support it is the MSL material. The superlattice structure is not very obvious. Further experimental results are supposed to give.
Response: Thanks for the comments and suggestions from the reviewer. Moiré patterns can arise under two conditions, either when the two lattices have slightly different parameters or when identical lattices are twisted at an angle θ with respect to each other.
HRTEM is widely used to characterize moiré patterns in 2D materials. In order to demonstrate that the synthesized WS2 was moiré superlattices (MSL) material, we provide HRTEM images and the simulated HRTEM image of WS2 as the sufficient evidence to support that it is the MSL material. We collected low-magnified TEM images and estimated the moiré superlattices as shown in Figure R1. Low-magnified TEM image of WS2 MSL in Figure R1a exhibits a well-arranged hexagonal lattice structure which is attributed to the twist of bilayer WS2 with a twisted angle. As can be seen, moiré superlattices are found throughout the measured region. The corresponding FFT patterns contain double sets of 6-fold symmetry diffraction spots. According to the measurement of the splitting spots in the FFT patterns, the misorientation angle of ∼13.8°could be calculated from the fast Fourier transformed (FFT) images as shown in Figure R1b. Herein, the twisted angle θ also could be obtained via the formula: θ = 2arcsin a/λ, where a = 0.322 nm is the lattice constant of WS2 and λ ≈ 1.34 nm is the moiré wavelength depicted in Figure R1c   To address the reviewer's concern, we replaced some typical images with regular hexagonal MSLs domains in Figure 2a, b as sufficient evidence to support that it is the MSL material (page 7 in the revised manuscript). The relevant discussions have been added into the revised manuscript (Page 7, 2-4; page 8, line 5-11).
2. They used the whole Figure 4 to explain the good hydrophilic and aerophobic characteristics from the morphology. However, the author needs to provide some intrinsic explanation for this specifically twistronics WS2 HER properties, not simply attribute to the morphology of material.
Response: Thanks for the comments and suggestions from the reviewer. The two primary categories of activity measurements are "total electrode" activity (i.e., geometric electrode area-normalized measurements) and "intrinsic" activity (i.e., persite turnover frequency, TOF). Total electrode activity measurements are useful for practical device performance comparisons, but they are not ideal for fundamental studies of novel catalyst materials because they do not reveal the physical or chemical origins of an electrode's activity. Intrinsic activity measurements provide the activity of the catalyst on a per-site basis, and therefore contribute to the molecular-level structure-property-function relationships necessary to guide catalyst development (ACS Catal. 2014, 4, 3957-3971).
The prominent "total electrode" activity of WS2 MSLs is due to the systematic optimization of electronic structure and geometric structure principally-that is, high "intrinsic" activity, abundant electrocatalytic active sites, excellent conductivity, and the hydrophilic and aerophobic characteristics of electrode surface for fast mass transfer.  Taking the reviewer's suggestions seriously, the first-principles method was used to study the twist effect of bilayer WS2. The superlattice with θ = 13.2° was selected for further calculations of the total potentials in moiré superlattices. As shown in Figure   R2, obviously, compared with the referenced bilayer structure, the potential barrier between two adjacent layers in the moiré superlattice has much smaller height and width, implying the increased interlayer coupling between electron orbitals. Because of the reduced potential barrier and increased interlayer coupling in the WS2 moiré superlattice, electrons can transfer easily from the conductive substrate to the active sites, thus leading to superior HER electrocatalytic performance. According to the reviewer's suggestion, we have included the above discussion into new version as intrinsic explanation for this specifically twistronics WS2 HER properties in our case. We added the Figure R2  3. In the DFT calculation section, what is the basis of the author modeling the initial periodic superlattice periodic structure MSLs WS2? How to prove the modeling geometry are consist with the experimental structure? Did the author fully relax the initial constructed structure to provide the optimization? Did the author give any strain or else vectors to keep the structure not relax to the initial geometry? How does the author know the activity only occurs on the edge not the basic plane? Anyway, the author needs to give more intrinsic explanation from the electronic structure, not only calculated the adsorption energy of H + .
Response: Thanks for the comments and suggestions from the reviewer.
(1) The twisted bilayer WS2 MSLs can be formed when the two identical monolayers are stacked with a relative twist angle of θ. The lattice structure of the twisted bilayer can be commensurable, such that the twisted bilayer forms a crystallographic superlattice whose superlattice period is determined by θ (ACS Nano 2017, 11, 11714-11723). Additionally, moiré patterns of twisted bilayer can also introduce a periodic potential acting on the interlayer coupling between two monolayers of the twisted bilayer, which results in the formation of a moiré superlattice. The period of moiré superlattice can be equal to or smaller than that of the crystallographic superlattice in the twisted bilayer. For the rotating model, we fixed a W atom and rotated the whole layer atom by a certain angle. The optimized structures of bilayer 2H-WS2 and 1T/1T'-WS2 with the moiré unit cell were shown in Figure R3.   Although it is still a great challenge for theoretical calculation to provide a complete description for the real and complicated experimental system, especially for the surface electrocatalysis, the theoretical analysis based on simplified model system still can help understand the surface reaction mechanism (PNAS 2011, 108, 937-943).
(3) We confirmed that the structures are indeed relaxed fully, i.e., both atomic positions and lattice constants are relaxed, to achieve a fully optimized stable structure. The structural optimization was performed until the total energy difference was less than 10 −5 eV. In order to investigate the minimum energy path of S atoms in the structural phase transition of WS2 stacking layer structures, the NEB method, as implemented in formalism (Phys. Rev. B 1999, 59, 1758). A vacuum region larger than 15 Å and perpendicular to the WS2 sheets (along the c axis) was applied to avoid the interaction between the sheets in neighboring cells caused by the periodic boundary condition. In our calculation, a kinetic-energy cutoff for plane-wave expansion was set to 500 eV. All atoms in each unit cell were fully relaxed until the force on each atom was less than 0.005 eV/Å. Electronic energy minimization was performed with a tolerance of 10 −6 eV.
The DFT-D3 approach (J. Chem. Phys. 2010, 132, 154104) was used in order to take into account the effect of the van der Waals interaction. In order to test the HER activity on the edge sites of WS2 MSLs, we modeled a nanoribbon structure with a vacuum of 15 Å in the y-direction. Two kinds of edge structures are created by the above procedure, including the one that ends with the exposed W atoms and the other ending with S only.
We have added the details of DFT calculations in the Supplementary Note 4. Table   R1 lists the binding energy per WS2 unit and the interlayer distance for different system.
Comparing to the AA stacking, the twisted bilayer is energetically more favorable, as evidenced by the reduction in binding energy and the decrease in interlayer distance. (4) There was no strain given to keep the structure not relax to the initial geometry. The twisted bilayers were modeled using accidental angular commensurations. In a hexagonal lattice whose basis vector is a1, a2, a skewed supercell with basis vector (na1 + ma2) has a corresponding skewed angle. We have added these data as a new      Figure 5. This seems to be contradictory.

Response:
Thanks for the comments from the reviewer. We appreciate the valuable comments and are pleased to clarify this issue. Two surface models through trigonal prismatic and octahedral coordination of W atoms were built to simulate 2H-WS2 MSLs and 1T'-WS2 MSLs, respectively. As shown in Figure R5, the hydrated cation preferentially adsorbs onto the surface of the 1T' phase, evidenced by a much lower adsorption energy (−3.45 eV) as compared to that of the 2H phase (−1.82 eV). Actually, the 1T' phase is formed due to the filling of electrons to the d orbitals of the W atom in the 2H phase. This is to say, as compared to 2H, the 1T' phase is an electron-rich phase that exhibits a high affinity to the positively charged ions in the electrolyte solution and results in enhanced adsorption. The ΔGH close to zero at 1T'-WS2 MSLs basal plane site (−0.24 eV) validates the high HER activity of WS2 MSLs in this study ( Figure R6).
Wettability is a function of the specific free energy for any given surface (i.e., higher surface free energy causes increased wettability) (ACS Nano 2016, 10, 9145-9155). It has also been reported that active edge sites with the moderate adsorption energy of proton could also act as hydrophilic species (   important and helpful for improving the quality of our work. As will be shown below, we have carried out a series of new measurements following Reviewer 2's comments.
According to lots of articles by Thomas F. Jaramillo et al, all the data were standardized and analyzed to ensure the accuracy of the conclusion in the revised manuscript. The main conclusion of our manuscript is further strengthened, and we believe the quality of the paper is significantly improved.
1. In the abstract, the authors mention "electron-catalytic activity", is this supposed to be electro-catalytic activity?
Response: Thanks for the comments from the reviewer. According to the reviewer's suggestion, after careful consideration, we believe that the "electrocatalytic activity" in our article is more accurate. We have changed "electron-catalytic activity" to "electrocatalytic activity" in the revised manuscript (Page 2, line 10).
2. Also in the abstract, the authors mention "ascribed to much more appropriate ΔGH of twisted bilayers WS2 active sits". Without context, "more appropriate" is meaningless.
Also, the typo "active sits" has been repeated throughout the manuscript (perhaps unsurprising, given that this sentence has been copy/pasted from later in the manuscript).
Response: Thanks for the comments and suggestions from the reviewer. We have used "ascribed to appropriate ΔGH of twisted bilayers WS2 active sits" instead of "ascribed to much more appropriate ΔGH of twisted bilayers WS2 active sites" in the revised manuscript (Page 2, line 11-12). Furthermore, we have checked the manuscript thoroughly and make the corresponding revisions. The typo "active sits" have been corrected in the revised manuscript.
3. In Figure 1c the authors mention finite element calculations, but as far as I can tell they have not provided any details on the nature of these calculations?   Response: Thanks for the comments from the reviewer. We think that there are too many factors (temperature and pressure, etc) that can induce errors during the synthetic process, and it is almost impossible to ensure an identical twist angle of every nanobelts.
However, the synthesized WS2 nanocones in this work are very uniform. We collected ten HRTEM images and their corresponding FFT images taken from different WS2 nanocones to ensure the twist angle in Figure R10.  MSLs, (Ⅱ) 1T'-WS2 NSs, (Ⅲ) 2H-WS2 NSs.
Moreover, a more detailed description was presented in the electrochemical test section (Supplementary Information, page 4-6, marked by red). As shown in Figure   R12, we also put all raw data without iR corrections and iR corrected data in the revised SI to give comprehensive information (Supplementary Figure 13). 7. In Figure 3a (and SI), potential is spelled wrong on the x-axis.
Response: Thanks for the comments from the reviewer. We have changed the "potencial" to "potential" in Figure 3a (and SI).     with the caveat that the active electrode area must be known. If this is not known or it cannot be measured, then the difference in the curves shown in Figure 3a/b could EITHER be due to changes in activity (i.e., a legitimate difference in j0) or could simply be due to differences in the roughness/porosity of the electrode. In principle, even a very 'inactive' catalyst could present a lower overpotential (hence higher apparent "activity") at a given geometric current density if the exposed surface area is high enough. Misleading statements about the "activity" of rough/porous electrodes is a major problem in the electrocatalysis literature, as it makes it very difficult to compare results from different labs. There have been many reviews that discuss this fact (e.g., To further demonstrate the enhanced intrinsic activity of WS2 MSLs, the polarization curves were normalized to electrochemically active surface area (ECSA), which was derived from the double-layer capacitance (Cdl, Figure R14-17). As shown in Figure   R18 and In addition, we also calculated the activities per mass of WS2 MSLs, and compared it with the activities per mass of WS2-based electrocatalysts in other literature (Table   R5)

2015, 25, 1127-1136
As shown in Figure R19,  10. In Figure 3a/b, Pt/C is presented but never mentioned/discussed in the main text.
Response: Thanks for the comments and suggestions from the reviewer. The relevant discussions have been added into the revised manuscript (Page 12, line 3-10).
11. The form of analysis performed in Figure 3d assumes that the electrodes behave as "ideal" electrochemical capacitors (i.e., current scales linearly with scan rate). However, all three plots do not pass through the origin, (0,0), and some even show clear deviations from linearity (i.e., see the 1T'-WS2 NSs). Consulting the source data ( Supplementary   Figure 14), all three electrodes exhibit asymmetric I-E plots (i.e., the reduction current is larger in magnitude than the oxidation current) that are somewhat sloped. This may suggest that a charge-transfer reaction (i.e., Faradaic process) may also be occurring in this potential range (i.e., at E < 0.20 V, where the baseline is clearly sloped). Is it appropriate to perform such a simple form analysis with such a complex electrode architecture? Perhaps the authors could investigate how closely the electrodes mimic an "ideal capacitor" with electrochemical impedance spectroscopy? In addition, In     Table R6). The double-layer capacitance we measured by EIS is within 15% of that measured from the scan ratedependent CVs (Table R7) Table R6. Impedance parameters for the equivalent circuit that was shown in Figure   R24.

Samples
Rs CPE Rct  Table R7. Impedance parameters for the equivalent circuit that was shown in Figure   R24.

The authors state: "The much smaller Tafel slopes of WS2 MSLs (40 mV decade −1 ) indicated that the kinetics of the electrochemical hydrogen evolution on WS2
MSLs was much faster than those of the 2H-WS2 NSs and 1T'-WS2 NSs (Fig. 3b)." Tafel slopes DO NOT indicate on kinetics, rather they can indicate on the reaction mechanism under some circumstances (i.e., for a well-defined, dimensionally stable electrode). As noted above, j0 would be an indicator of kinetics, but this has not been calculated.   Fig. 11)." However, they do not offer any justification/discussion as to why the EIS measurements indicate this. They do not even fit the spectra or include an equivalent circuit. Given the high porosity of the electrodes in question, interpreting the EIS spectra is not straightforward.  Figure R26). The simulation of the EIS spectra using an equivalent circuit model allowed us to determine the charge transfer resistance, Rct, which is a key parameter for characterizing the catalyst-electrolyte charge transfer process.  Table R8).

As shown in
We attribute this measured small charge transfer resistance (Rct) to its distorted nanobelt structure. The edge-terminated feature can ensure an isotropic electron transport from CFC substrate to WS2 edges and significantly decrease the resistance for traversed layers (Nat. Commun. 2015, 6, 7493; Nano Lett. 2014, 14, 553-558). In addition, the misorientation-induced lattice strain and the reduced interlayer potential barrier in the moiré superlattice ( Figure R27)   The inset is the equivalent circuit model that contains the electrolyte resistance (Rs), constant phase element (CPE) and charge-transfer resistance (Rct). Z' is the real impendence and Z'' is the imaginary impedance. Table R8. Impedance parameters for the equivalent circuit that was shown in Figure   R26.  14. The authors state: "The electrochemical double-layer capacitances (Cdl) were calculated to contrast the electrochemical surface area (ECSA) of 2H-WS2 NSs,

Rs
1T'-WS2 NSs and WS2 MSLs (Fig. 3d). The Cdl of WS2 MSLs (33.7 mF cm −2 ) was much higher than that of 1T'-WS2 NSs (21.2 mF cm −2 ) and 2H-WS2 NSs (7.2 mF cm −2 ), indicating that the WS2 MSLs possessed more fully exposed active sites for electrochemical hydrogen evolution." How were these values normalised? Geometric area? In principle, if the specific capacitance (Cspecific in F cm −2 ) of these materials was known, then the exposed surface area (or ECSA) could be estimated from these values (i.e., AECSA = Cdl/Cspecific). The ECSA however does not necessarily indicate on the number of exposed "active sites", as all sites, regardless of 'activity' (e.g., the basal and edge planes of 2H-WS2 plus the underlying carbon support) contributes to the non-faradaic current, whereas only certain sites (e.g., the edge plane of 2H-WS2) may dominate the HER catalysis. How was the carbon support corrected for when calculating the ECSA?
Response: Thanks for the comments and suggestions from the reviewer. As shown in where Cdl is double-layer capacitance, Cspecific is specific capacitance.
In general, the ECSA estimates to be accurate within about an order of magnitude, It is well known that bare carbon fiber cloth (CFC) preferentially increases the conductivity and dispersity of the samples so that higher current densities and lower overpotentials for electrocatalysis will be exhibited. But the bare CFC we used, Phychemi (HK) Company Limited-W0S1010, can hardly be polarized for HER (seeing the curve in Figure R28). Therefore, the intrinsic performances of the bare carbon for HER can be neglected. Moreover, the same electrochemical capacitance test methods (see the part of Electrochemical Measurements in Supporting Information) were used to determine the double-layer capacitance (Cdl) of bare CFC (Figure R29). The Cdl values of the different samples for HER were shown in Table R9. The Cdl of all WS2 samples are more than two orders of magnitude greater than that of bare CFC. Therefore, the bare CFC exhibits negligible effect on the electrochemical surface area measurement of electrode materials. The background capacitance of the bare CFC electrode was subtracted from the obtained double layer capacitance to compensate for the low substrate coverage (Nat. Commun. 2019, 10, 2650).     15. The authors use the terms "superhydrophilic" and "superaerophobic" throughout.
It would be useful if they provide a definition of these terms for the reader (i.e., what distinguishes hydrophilic from superhydrophilic?).

Response:
Thanks for reviewer's kind suggestion. According to the comments of reviewers, in the revised manuscript, we try to clarify some basic concepts and facilitate readers to better understand the unique wetting phenomenon on the electrode material surface. We have added some terms interpretation in the revised manuscript, such as "hydrophilic", "hydrophobic" and "superhydrophilic" (Page 16, line 13-16). In addition, we have added the Supplementary Figure 25 (Figure R30) to help readers better understand these terms, showing a detailed schematic diagram and corresponding textual annotations.   17. The authors state "We ascribe the activity enhancement to a combination of electronic, geometric, superaerophobic and superhydrophilic effects." Again, as stated above, in electrochemistry "activity" refers to electron-transfer kinetics. This statement is misleading, as the aforementioned "superaerophobic and superhydrophilic effects" influence the mass transfer of bubbles, rather than enhancing electron-transfer kinetics.
Response: Thanks for reviewer's kind suggestion. The two primary categories of activity measurements are "total electrode" activity (i.e., geometric electrode areanormalized measurements) and "intrinsic" activity (i.e., per-site turnover frequency, TOF). We have carefully polished this statement to "We ascribe the total electrode activity enhancement to a combination of electronic, geometric, superaerophobic and superhydrophilic effects." We have made the corresponding changes in the revised manuscript (Page 20, line 22; page 21, line 1-2).
18. The authors state "ΔGH is insensitive to the MSLs" which is counter to the argument that they presented in Figure 5 Figure R32 and Table R10). The inset is the equivalent circuit model that contains the electrolyte resistance (Rs), constant phase element (CPE) and charge-transfer resistance (Rct). Z' is the real impendence and Z'' is the imaginary impedance. Table R10. Impedance parameters for the equivalent circuit that was shown in Figure   R32. Our point-by-point response is presented as follows.

Rs
1. HRTEM images in Figure 2a and 2b seems have been applied too much filters during the data recording and processing, and the images didn't show a clear Moiré period even in a small region.
Response: Thanks for the comments and suggestions from the reviewer. Moiré patterns can arise under two conditions, either when the two lattices have slightly different parameters or when identical lattices are twisted at an angle θ with respect to each other.
HRTEM is widely used to characterize moiré patterns in 2D materials. In order to demonstrate that the synthesized WS2 was moiré superlattices (MSL) material, we provide HRTEM images and the simulated HRTEM image of WS2 as the sufficient evidence to support that it is the MSL material. We collected low-magnified TEM images and estimated the moiré superlattices as shown in Figure R33. Low-magnified TEM image of WS2 MSL in Figure R33a exhibits a well-arranged hexagonal lattice structure which is attributed to the twist of bilayer WS2 with a twisted angle. As can be seen, moiré superlattices are found throughout the measured region. The corresponding FFT patterns contain double sets of 6-fold symmetry diffraction spots. According to the measurement of the splitting spots in the FFT patterns, the misorientation angle of ∼13.8°could be calculated from the fast Fourier transformed (FFT) images as shown in Figure R33b. Herein, the twisted angle θ also could be obtained via the formula: θ = 2arcsin a/λ, where a = 0.322 nm is the lattice constant of WS2 and λ ≈ 1.34 nm is the moiré wavelength depicted in Figure R33c (Nature 2018, 556, 80-84). The significant honeycomb-structured Moiré pattern in Figure R33c is consistent with the simulated HRTEM images of WS2 MSL (Figure R33d). To address the reviewer's concern, we replaced some typical images with regular hexagonal MSLs domains in Figure 2a, b as sufficient evidence to support that it is the MSL material (page 7 in the revised manuscript). The relevant discussions have been added into the revised manuscript (Page 7, 2-4, page 8, line 5-11). Fig. S5, we can see the 1T′-WS2 NSs sample has a low crystallinity, to make a comparison, the standard PDF card should be presented in the same image.

From the XRD results in
Response: Thanks for the referee's valuable comments and suggestion. For comparison, the standard PDF card (PDF#08-0237) was presented in XRD pattern ( Figure R34).
We added the Figure R34 Figure R35 and Table R11). The inset is the equivalent circuit model that contains the electrolyte resistance (Rs), constant phase element (CPE) and charge-transfer resistance (Rct). Z' is the real impendence and Z'' is the imaginary impedance. Table R11. Impedance parameters for the equivalent circuit that was shown in Figure   R35.

Response:
We appreciate the reviewer very much for the helpful suggestion. We have changed the "S-Mo-S" to "S-W-S", which has been highlighted in the revised We thank the referees for their valuable comments and positive endorsement to our manuscript. We have carefully considered the referees' comments and revised the manuscript accordingly. Our responses and corresponding revisions are as follows: Reply to Referee 1: The authors have answered all my questions and I accept it for publication.

Response:
We are very grateful to your encouraging and positive comments and really appreciate your agreement of acceptance with this revised manuscript.

I have read the revised manuscript and responses to my original comments in detail and
agree that there is a significant improvement in this version. However, I still have some queries about the electrochemical dataset. Please consider the following: Response: Many thanks for your positive comments on our manuscript. We have revised our manuscript accordingly. The low Tafel slope of 40 mV decade −1 for HER indicates the fast reaction kinetics." to "The as-synthesized WS2 MSLs electrocatalysts display an overpotential of 60 mV at a current density of 10 mA cm −2 and a Tafel slope of 40 mV dec −1 " in the revised manuscript (Page 4, line 9-11).
2. Figure  3. Lines 204-208: Exchange current density is only meaningful if normalised to the ECSA (see Reviewer 2, comment 9). I suggest moving this discussion to after the ECSA has been calculated (perhaps where the TOF is discussed?) and re-normalising j0 with the ECSA.
Response: Thanks for the comments and suggestions from the reviewer. As shown in Figure R1, the exchange current density was normalized to electrochemically active surface area (ECSA). Moreover, we moved the discussion about exchange current density to after the ECSA calculation and before the TOF calculation in the revised manuscript (Page 14, line 1-5). We added the Figure R1   4. In response to Reviewer 2, comment 2: In my view, the revised text is still ambiguous.
Ultimately, it is the decision of the authors, but my recommendation would be "ascribed to a closer to thermoneutral hydrogen adsorption free energy value (i.e., ∆GH --> 0) of twisted bilayers active sites" Response: Thanks for the comments and suggestions from the reviewer. After careful consideration, we believe that the statement suggested by reviewer is clearer and more accurate. We have used "ascribed to a closer to thermoneutral hydrogen adsorption free energy value of twisted bilayers active sites" instead of "ascribed to appropriate ΔGH of twisted bilayers WS2 active sites" in the revised manuscript (Page 2, line 11-13).
5. In response to Reviewer 2, comment 6: In Figure R11, "at a static overpotential of 0.2 V vs. RHE". Potential is measured versus a reference (RHE) whereas overpotential is measured relative to the equilibrium potential, by definition. This statement should either read: "at a static potential of −0.2 V vs. RHE" or "at a static overpotential of 0.2

V"
Response: Thanks for the comments and suggestions from the reviewer. We have used "at a static potential of −0.2 V vs. RHE." instead of "at a static overpotential of 0.2 V vs. RHE." in the revised manuscript (Page 11, line 5-6).
6. In response to Reviewer 2, comment 13: This remains a concern in the revisions. In Figure R26 was truly taken in a "non-faradaic" region, that how can the authors justify the use of a Randles circuit, which is used to describe a Faradaic process? Strictly speaking, if the electrode/electrolyte behaves as an "ideal" electrochemical capacitor (Reviewer 2, comment 11), then there should be no charge transfer (Faradaic) process taking place. What is the physical origin of RCT? In addition, given that this RCT value cannot be related to the HER, as it is thermodynamically impossible at 0.25 V vs. RHE, then why should RCT be related to HER activity? In my view, there is no physicochemical reason why the RCT measured for a charge-transfer process at 0.25 V vs. RHE should reflect the kinetics of the HER, which is a totally different process that takes place in a different potential range! Response: Thanks for the comments and suggestions from the reviewer. We agree with you that the Rct value fitted in the non-faraday process is not be related to the HER, let alone the dynamics of HER.
To illustrate the improved electron transport in HER, electrochemical impedance spectroscopy (EIS) was carried out at a potential of −50 mV (vs. RHE) for all WS2 samples, as shown in the Nyquist plots in Figure R2. The EIS measurements were performed in 0.5 M H2SO4 solution from 100 kHz to 0.1 Hz. The Warburg component of the Nyquist plots is not observed at frequencies as low as 0.1 Hz. This indicates that there was a sufficient supply of H + at the surface and mass transport impedance can be neglected 1 . The EIS data are characterized by two overlapping semicircles on the complex plane plot; one at high frequencies with small diameter and one at low frequencies with larger diameter. Several models have been developed to explain the two semi-circle EIS response of HER on various electrodes 1 .
We adopt the two time constant parallel model proposed by Armstrong and Henderson 2 (the equivalent circuit is displayed in Figure R2b). We added an inductor to account for the observed high frequency inductance from, e.g. the connection wires.
The use of constant phase elements instead of capacitors is required to account for the slight depression of the semi-circles caused by inhomogeneities at the atomic scale [3][4][5] .
The constant phase element impedance is described by: where i = √-1, ω is the angular frequency of the AC voltage, 0 < α < 1, and Q is the frequency-independent parameter. The case α = 1 recovers a perfect capacitor.
The capacitance associated with the CPS can be evaluated as 6 : This model is characterized by one high frequency time constant (τ1, CPE1-R1) and one low frequency time constant (τ2, CPE2-R2). The experimental impedance data were fitted to the equivalent circuit model by ZView software. The EIS data at a potential of −50 mV vs. RHE together with the corresponding fitted curve are displayed in Figure   R2a and the equivalent circuit model is seen to the EIS data very well. In addition, it is noted that equivalent circuits drawn for same electrochemical impedance spectroscopy are not unique.
In the two time constant parallel model the RS resistance element is attributed to the uncompensated solution resistance, the high frequency time constant (τ1, CPE1-R1) is related to the Faradaic resistance for the charge transfer process, Rct, and double layer capacitance, Cdl, while the low frequency time constant (τ2, CPE2-R2) is related to hydrogen adsorption, Rp and Cp.
The fitted Rct values for WS2 MSLs, 1T'-WS2 NSs and 2H-WS2 NSs are 1.6, 3.4, and 11.2 Ω, respectively ( Figure R2). Importantly, the WS2 MSLs exhibits the smallest Rct value among the tested samples, suggesting the superior interfacial charge-transfer kinetics on the surface of WS2 MSLs for HER catalysis.
We added the Figure R2