Two-dimensional antimonene single crystals grown by van der Waals epitaxy

Unlike the unstable black phosphorous, another two-dimensional group-VA material, antimonene, was recently predicted to exhibit good stability and remarkable physical properties. However, the synthesis of high-quality monolayer or few-layer antimonenes, sparsely reported, has greatly hindered the development of this new field. Here, we report the van der Waals epitaxy growth of few-layer antimonene monocrystalline polygons, their atomical microstructure and stability in ambient condition. The high-quality, few-layer antimonene monocrystalline polygons can be synthesized on various substrates, including flexible ones, via van der Waals epitaxy growth. Raman spectroscopy and transmission electron microscopy reveal that the obtained antimonene polygons have buckled rhombohedral atomic structure, consistent with the theoretically predicted most stable β-phase allotrope. The very high stability of antimonenes was observed after aging in air for 30 days. First-principle and molecular dynamics simulation results confirmed that compared with phosphorene, antimonene is less likely to be oxidized and possesses higher thermodynamic stability in oxygen atmosphere at room temperature. Moreover, antimonene polygons show high electrical conductivity up to 104 S m−1 and good optical transparency in the visible light range, promising in transparent conductive electrode applications.

Is there any supporting references or data for this statement?
ii) Figure 4. Did the authors run ab-initio calculations up to 2000 ps? In the main text, the authors state that the simulation is up to 2000 ps, while in Figure 4, it is 2000 fs.
iii) Reference 25 predicts  and -phases to be degenerate. Can the authors comment on it?
iv) The results of a recent paper for oxygen in phosphorene should be bench-marked with the results obtained previously (Wang et al. 2D Mater. 3, 025011, 2016). In the present study, the calculated dissociation energy of O2 on phosphorene is high. Did the authors use triplet or singlet O2 to get the dissociation energy?
v) The dissociation energy of O2 on antimonene is calculated to be -1.62 eV/molecule, which implies that O2 prefers to dissociate on antimonene. XPS measurement could confirm whether O2 is dissociated or not. Can the authors address this discrepancy?
iii) Pl correct he citation for references and figures in the main text.
(b) Page 10. "Raman spectra of typical antimonene polygons stored in air for one month after sample synthesis are shown in Figure S5." Figure S5 is not Raman spectra..

Point-by-point replies to the reviewers' comments
Reviewer #1 Q1: The authors claimed that antimonene nanoribbon can be found in Figure 1b. However, the width of the sample is even larger than 10 um, how can it be called "nanoribbon"? Answer) Thanks for your reminder or thank you for your commenting. Indeed, this expression is not accurate. It has been changed into "ribbon". Furthermore, 85 ribbons were counted, and their width distribution is shown in Figure R1.
From the statistics, we can see the lateral size of 70% ribbons is less than 10 um and that of 5.4 % ( 27%*20% =5.4% ) is less than 1 um. Therefore, a small portion of ribbons can be called 'nanoribbon' while lateral size of most ribbons are more than 1 um.
We have made corresponding corrections in the revised manuscript.
(line 26~28 page 7) Q2: The authors mentioned that bulk antimony can be viewed as ABC stacking of monolayer anitmonene. What is the stacking order of the few-layer anitmonene in this manuscript? In addition, the crystal constants of few-layer antimonene should also be given.

Answer)
Thanks for your comments.
1) As we know, stacking order significantly affects the properties of 2D materials. In our manuscript, we have added the description into the TEM characterization. Figure S8 (Supplementary Information).

Obviously, there is very good agreement on the arrangement of bright points between the experimental and the simulated atomic images, which confirms the ABC stacking of the synthesized antimonene layers. '
Besides, comparing the line scanning profiles of experimental image with those taken directly from the calculated (dash line) images along the dashed lines, we can also find atomic image obtained experimentally in good agreement with the simulated ABC stacking.
( Line 3~7 Page 10) 2) As to the cystal constants, β-Sb possesses a rhombohedral structure though theoretical workers usually view it as hexagonal structure. And relative parameters ( a=b=c= 4.418, α=β=γ=56.979 º) is added into Figure 2a   Q3: The antimonene sheets in Figure 1g and Figure S2 are the same sample, why the thickness of the former is ca. 4 nm, while the later is 1 nm? Did you put the picture wrong in Figure S2? Answer) Thanks for your commentsreminding. We errantly put Figure 1g in Figure S2 carelessly and we have just deleted the the wrong picture.

Q4:
The authors pointed out that the absence of dangling bonds on the substrate surface is critical for the successful growth of antimonene polygons. However, one antimonene sheet grown on mica in Figure S3 shows irregular shape. Is it oxidized or formed by some other reasons?
Answer) We appreciate the reviewer's comments. Raman spectra, AFM files and TEM-EDS files are areadded in the latest manuscript, confirmed the reliability and stability of our experiments and instruments. Instead of oxidization, we attribute the irregular shape to growth kinetics. Normally, the growth of 2D crystal can be divided into two stages: nucleation and growth. As shown in Figure 1h(the latest manuscript), samples, obtained after one-hour growth, usually possess a average lateral size of 6um. However, the sample possesses the irregular shape with a ca. 100 nm lateral size. It may be attributed to the poor crystallinity, further resulting in growth arrest.

Q5:
The authors mentioned that the migration energy barrier of antimony atoms on mica substrate is very small, resulting in a high migration and fast growth rate. Are there any references or calculation results to support this piont? In addition, after 60-min growth, the lateral size of the sample is only ca. 10 um. Such growth rate can hardly be called "fast". Answer) We appreciate the reviewer's comments.
1) As to the "low migration barrier" and "fast growth rate", Yes, they have been considered as two features of Van der Waals Epitaxy. For example, in reference 23, the authors clearly pointed out that the absence of dangling bonding resulted a low interaction between adatoms and substrate. Therefore, the migration barrier is 'low' and growth rate is 'fast'. Figure R2. Descriptions regarding "low migration barrier" and "fast growth rate" in reference 23.
2) As to the accurate growth rate of antimonenes, we implemented additional time-dependent growth experiments, as shown in Figure S4. According to Figure S4, the growth rate of antimonenes is about 0.5 um·min -1 and actually the growth has been finished in the first 10 minutes even for longer durations.
We have added the clear description on the growth rate in the revised manuscript .

( line 10~22 Page 8)
Q6: The characterization of the stability of the antimonene sheet is not convincing only by Raman analyses. Optical, AFM and TEM-EDX measurements of the freshly-made antimonene sheet and that stored one month should be given.

Answer)
We appreciate the reviewer's comments.
Optical, AFM and TEM-EDX was employed to characterize the stability as the reviewer advised. Both the measurements indicates the stability, detailedly illustrated in 'High stability of mutilayer antimonene'. This suggestion is very helpful. These additional experiments were implemented as shown in

Answer)
We appreciate the reviewer's reminder. Strictly, the JAP paper is indeed the first experimental work of antimonene. And, we added the paper into the citations of our paper.
Lei et al. firstly achieved the synthesis of antimonene via MBE( molecular beam epitaxy).
Samples obtained via MBE and research the band structure of monolayer antimonene. The spin-orbit coupling phenomenon was observed in monolayer antimonene. The work is worth a good evaluation.
However, plenty of necessary measurements, such as Raman spectroscopy, AFM and TEM, were absent, which are actually critical was bad for further study on antimonene. and greatly prevented the work from higher evaluation.
In our manuscript, we reported the physical vapor transport synthesis of multilayer antimonene, together with a systematic characterizations.
Anyway, we believe that both of our research will promote the further research and application of antimonene.
2. The authors mentioned that van der Waals epitaxy can be achieved even when the lattice constant mismatch is as high as 50% and the epitaxial layer is completely relaxed without excessive strain. If the interaction between the epitaxial layer and the substrate is totally van der Waals force, authors can only say that it is without excessive strain. But they cannot say that the epitaxial layer is completely relaxed. The "relax" of epitaxial layer is defined as the emergence of threading dislocations from the interface through the epitaxial layer in order to release the excess strain energy. The excess strain energy is caused by the lattice mismatch between the epitaxial layer and the substrate as the critical thickness of epitaxial layer is proceeded. If so, the interaction between the epitaxial layer and the substrate should be chemical bonding. However, the authors claimed that the interaction is totally van der Waals force. Since the van der Waals epitaxy can overcome the limitation of lattice mismatch in conventional heterogeneous epitaxy, the scale of epitaxial thin film should be reached to wafer level. However, the scale of antimonene layers in this study is less than 10 μm. In fact, the quality and scale of heterogeneous epitaxial materials depend on several factors, not only lattice mismatch. Due to the micrometer scale, it might not be the van der Waals epitaxy. If the authors believe that the interaction is totally van der Waals force, could they provide further evidence? For example, the XPS analysis could be used to confirm that there are not any chemical bonds between Sb and other elements in the substrate. Moreover, the cause of the limited scale of antimonene layers needs to be well explained, if the totally van der Waals force between the antimonene layers and mica substrate is proved.
Answer) We appreciate the reviewer's comments.
As the reviewer reminded, we did mix up some basic concepts. The "relax" in epitaxial layer doesn't corresponds to van der Waals epitaxy. According to the kind suggestion, we have corrected out initial expression to avoid the confusion.(line22 Page 6) And, the interaction force between Sb epitaxy layer and mica is van der Waals force undoubtedly, reasons listed as follows: Firstly, mica is completely passivated by F ions. Therefore, there is no possibility to form chemical bonding between epitaxial layer and mica. Secondly, lots of research on van der Waals epitaxy based on mica was published and typical papers have been cited in the manuscript, such as Referemce16, 18 and 23 in the manuscript. 16 Koma, A. Van der Waals epitaxy-a new epitaxial growth method for a highly latticemismatched system. Thin Solid Films 216, 72-76 (1992 Reference 16 firstly claimed the van der Waals epitaxy on the mica. The growth of several layered materials was mentioned for demonstration purposes. Later, a series of research based on exitaxy on mica was conducted. And, van der Waals epitaxy on mica is explicitly pointed out in Reference18 and 23.
Furthermore, as suggested, to confirm the van der Waals epitaxy, XPS was employed. As shown in Figure S2a , peaks of Sb 3d3/2 and Sb 3d5/2 appears at ca.528 eV and 538 respectively, in good agreement with nonvalent Sb while O 1s peak at ca.532 eV is attributed to mica. It's proved that there are not any chemical bonds between Sb and other elements on the substrate.
Another intuitionistic evidence is the morphology difference between samples fabricated on different substrates, which we have mentioned in the manuscript. As shown in Figure R5, antimony sampleson grown both SiO2/Si and sapphire possess irregular shape and higher thicknee while antimonene layers have well-defined shapes. These differences is attributed to surface induction.

Figure R5. Antimony grown on different substrates.
We also noticed that there is a size limitation during the van der Waals epitaxy growth of 2D crystals on mica and epitaxial layer can't fully cover the substrate. It has never been pointed out by clairvoyants. Herein, we'd like to share our speculation drawn from paper reading and analysis.
As the reviewer pointed out, crystals growth on mica, usually show a limited lateral size, which is a common phenomenon in this field, although mica was applied widely to van der Waals epitaxy. This confused us too. 3. The authors mentioned that the space between two sets of lattice fringes is measured to be 0.204 nm, corresponding to the (100) plane of antimony. The angle between two fringes is measured to be 120°, characteristic of hexagonal structure. However, grey antimony possesses rhombohedral structure. Its d spacing of (100) plane group should be 0.354 nm, the included angle between (100) and (-1-10) planes of it should be around 54° (or 126°) in theory. The material identification must be precisely checked.
Answer) We appreciate the reviewer's comments. Figure R7. Base vectors selection in hexagonal structure and rhombohedral structure(left) and Base vectors in Reference 34.
Firstly, as shown in Figure R7, hexagonal structure and rhombohedral structure can interconvert via different selection of base vectors in crystal.
As the reviewer pointed out, grey antimony possesses rhombohedral structure. However, up to now, experimental research is rare and a great portion of research on antimony is theoretical calculation. Usually, theoretical workers can conveniently build primitive cell when grey antimony is viewed as ABC stacking of monolayer antimonene along [111] axis with a, b and c as the base vectors, shown in Figure R5. And, Reference 34 is a typical example. Initially, we adopted this opinion to calculate the crystal parameters. Obviously, such an operation results strict confusion since grey antimony is known to possess rhombohedral structure. This has been corrected in rewritten 'Atomic scale microstructure of antimonene single-crystal polygons' section.(line9 Page 9~line 11 Page 10) Also, the reviewer pointed out some incorrect calculations in initial manuscript, which have been corrected as suggested, too.
For the absence of crystal parameters of rhombohedral structure in powder diffraction files(PDF) database, we calculate the diffraction spot of grey antimony along [111] axis. According to the simulated diffraction spots, we remark the SAED in the manuscript and rewrite TEM characterization section, as suggested. And, both of (100) and (-1-10) can not be found along [111] axis. Figure R6. Simulated diffraction spot of grey antimony(along [111] zone axis).
4. The crystal structure of β-Sb belongs to rhombohedral lattice and that of α-Sb belongs to orthorhombic lattice. The description, "the image clearly demonstrates a typical rhombic structure (along [001] axis), completely different from the hexagonal structure of β-phase antimonene," may confuse the readers. The descriptions regarding the crystal structures in the manuscript need to be rewritten for clear expression.
Answer) We appreciate the reviewer's comments.
In the initial manuscript, the we selected a, b and c as base vectors and viewed grey antimonene as a hexagonal structure, a confusing description regarding the crystal structure.
As the reviewer suggested, the descriptions regarding the crystal structure have been rewritten for the rhombohedral structure of grey antimony. (line9 Page 9 ~line 11 Page 10) 5. In thermodynamic simulation, the authors cannot only consider the difference of exothermic energy (Q), which is equal to enthalpy (H), between grey antimony and black phosphorus oxidations. The entropy should be also considered. Therefore, the difference of Gibbs free energy (G) between grey antimony and black phosphorus oxidations should be estimated in order to make sure which oxidation occurs preferentially.  (2) Both have positive value of ΔG, 59.394 and 15.993 kcal/mol, respectively. Though the enthalpy of the two reactions is less than zero, showing an exothermic process, the entropy of those reactions is reduced. From the calculated results, it suggests that both reactions are not spontaneous reactions at room temperature, and the product of O=Sb 2D =O is more difficult to generate than that for O=P 2D =O because of the higher value of ΔG. Thus, the calculated results and our strong experimental evidence also confirmed that the oxidation of 2D grey antimonene cannot occur preferentially, which is more difficult than that of black phosphorene. Figure R7. Thermodynamic simulation of Gibbs free energy (G) between greyantimony and black phosphorus oxidations 6. Essentially, antimony is a semimetal unless its thickness reduces under bilayer in theory. The antimony films obtained in this study cannot be called "antimonene" which is single layer antimony. It should be called multilayer antimonene or antimonene layers. It is unnecessary to use a known semimetal material as the channel of transistor to prove that it is a semimetal. The scale of the antimonene layers obtained in this study is too small to be commercialized for industrial device fabrication.
Answer) We appreciate the reviewer's comments. Samples in our research are inedeed more than bilayer. And, 'multilayer antimonene' is more accurate. We have corrected it in the latest version manuscript.
The reviewer mentioned that antimony is a semimetal unless its thickness reduces under bilayer in theory and it is unnecessary to use a known semimetal material as the channel of transistor to prove that it is a semimetal, consistent with the calculation in our previous paper(Reference 10).
Firstly, one purpose of this test to confirm that it's truly β-Sb instead of α-Sb since α-Sb, as a isologue of black phosphorous, is expected to be stable and semiconducting. Therefore, the test can further confirm the phase of our sample. ). There is a urgent necessity to verify the electronic properties of β-Sb since it's a isologue of β-As. Actually, during our revision, they claimed β-Sb was similar to β-As and has a bandgap in another paper(Reference 36). So, electrical tests in our paper are quite important to clarify the potential confusion.
As to the present sample scale, just as the reviewer pointed out, it is indeed far from industrial application. The demonstration shown in our paper was just based on the material properties to show a potential application. And, looking back on the the growth of 2D materials, we can find that the successful synthesis of 2D materials, responsible for practical commercial application, wasn't usually realized in a short term.
As the reviewer expected, we will devote more efforts to grow samples with higher quality and larger scale for further research and application.
7. The authors mentioned that the antimonene layers have potential to be used for the applications of the transparent conductive electrode. However, the scale of antimonene layers obtained in this study is too small to be a thin film. In the bending test, how do the authors confirm that the bending effect is indeed exerted on the antimonene layers with the scale less than 10 μm, since the scale of mica substrate is centimeter. In addition, the authors must enlarge the scale of antimonene layers up to centimeter scale for real applications.
Answer) We appreciate the reviewer's comments.
1) As the reviewer mentioned, antimony polygons was too small for us to confirm if the bending effect exerted on the antimony polygons. Therefore, we make some adjustments.
Mica substrate was attached onto the surface of iron rod to confirm that the bending effect is equably exerted on the whole substrate. Subsequently, we retest the conductivity change as a function of bending times and drew a more strict conclusion. As shown in Figure 5d, bending test affects little on the conductivity.
2) Just as the reviewer pointed out, the present sample scale can't satisfy real application. In fact, we urgently want to enlarge the sample scale, as the reviewer expected.
8. The authors need to correct many typos in the manuscript and the writing should be polished by a native English speaker.
Answer) We appreciate the reviewer's comments. We checked the manuscript carefully and correct the typos, as the review pointed out.

Reviewer #3
i) Page 7. The authors state that "The migration energy barrier of antimony atoms on mica substrate is very small, which results in a high migration rate along the mica substrate and a fast lateral growth of 2D antimonene polygons." Is there any supporting references or data for this statement?

Answer)
We appreciate the reviewer's comments. In fact, mica is widely applied as the van der Waals epitaxy substrate. Compared with other substrates, smaller atom migration energy barrier on mica is clarified in many papers, some of them listed in the manuscript: Furthermore, as suggested, to confirm the van der Waals epitaxy, XPS was employed. As shown in Figure S2a  iii) Reference 25 predicts □ and □ phases to be degenerate. Can the authors comment on it?
Answer) We appreciate the reviewer's comments. In this question, there are missing "two word": iii) Reference 25 predicts ? and ? phases to be degenerate. In our opinions, we guess that this question should be like: "iii) Reference 25 predicts γ and δ phases to be degenerate.
Can the authors comment on it?".In Reference 25, Prof. Pandey group studied antimonene, with four interesting phases, α, β, γand δ phases. In group V, α and β are typical phases. In fact, the counterpart bulk material of α and β group V monolayer, such as phosphorene, arsenene and antimonene, are very stable phases among their own allotropes. For example, the counterpart bulk materials of β-arsenene, β-antimonene, and β-bismuthene are rhombohedral layered gray arsenic, gray antimony, and β-type bismuth under normal conditions with the same space group. To our knowledge, till now two types of layered phases experimentally exist in the group V phases, namely, the bulk α-P, α-As, β-As, β-Sb, β-Bi. So, for monolayer it seems that α and β are more stable than other phases (γ and δ phases). Of course, our previous work and Pandey's work together confirmed this point.
In the present study, the calculated dissociation energy of O2 on phosphorene is high. Did the authors use triplet or singlet O2 to get the dissociation energy?

Answer)
We appreciate the reviewer's comments. Our results for oxygen in phosphorene have been bench-marked with the results obtained previously (Wang et al. 2D mater. 3, 025011, 2016). The value reference has been cited in the revised manuscript. All references are corrected in the text in sequential order throughout the whole manuscript.
In our old manuscript, we use singlet O 2 to get the dissociation energy, and the calculated dissociation energy of O 2 on phosphorene is high. According to your kinder suggestion, we take into account the triplet O 2 to further study the dissociation energy. The v) The dissociation energy of O2 on antimonene is calculated to be -1.62 eV/molecule, which implies that O2 prefers to dissociate on antimonene. XPS measurement could confirm whether O2 is dissociated or not. Can the authors address this discrepancy?
Answer) We appreciate the reviewer's comments.
Calculations indicates the tendency that O2 prefers to dissociate on antimonene, To verify the stability of antimonene experimentally, XPS was employed to measure sample stored for one week after sample synthesis. In Figure S2a, peaks at 528.4 eV and 538 eV correspond to Sb 3d5 and Sb 3d3, respectively, which is attributed to nonvalent antimony and consistent with reference 34. Its stability was further confirmed. Also, TEM-mapping in Figure 2 indicates the similar conclusion.
We think that the oxidation may happen according to the calculation, but the reaction is quite (iv) Text: DFT calculations are not 'ab-initio' calculations. ab initio -> first principles Answer) We appreciate the reviewer's comments.. In the new manuscript, we have changed "ab-initio" into "first-principles".

Reviewers' comments:
Reviewer #1 (Remarks to the Author): The authors have addressed my comments and suggestions properly, therefore, I would like to recommend the publication of this manuscript in Nature Communications.
Small questions:. 1) The authors should give solid characterization results to prove that the 1-nm sample shown in Figure 1h is indeed antimonene. In addition, no scale bar was given in Figure 1h.
2) In Figure 2a, no unit was given for the crystal parameters: a = b= c = 4.418.
3) Please check the labels of XPS Sb 3d peaks shown in Figure S2. 4) The representation of zone axis in Figure S6 should be changed from (001) (2010)). Even though lots of studies only utilize STM for analysis, they still can publish the only STM results in high impact factor journals because of the accuracy of STM.
Answer. We report the PVT synthesis of antimonene while The JAP paper reported the synthesis of antimonene via MBE.
First significant breakthrough is our method. As we konw, PVT, together with the generally similar CVD, wins much popularization and broad recognition for combining crystal quality, large scale production and low cost while expensive and complex instruments usually results in hindered advance of both scientific research and real application. A typical example is stanene fabricated via MBE (Nature Materials 14, 1020-1025 (2015)). Few experimental research follows the famous NM paper, illustrating how a complex and expensive method hindered the advance of stanene research. Anyway, a report of highly-popularized method will surely raise the great attention and promote the both scientific research and real application.
Then, samples, large enough for normal device fabrication, are achieved, which dramatically breaks through the JAP paper. By contrast, the JAP paper only fabricated domaind with a few hundred nm scale, which is far from most research and incurs comparatively low evaluation.
Moreover, abundant original data are listed detailedly in the following comparison Answer. We appciate the comment. This information is mentioned in Reference 23 and the crystal structure of mica is illustrated in Figure 3c in Reference 23. Answer. We appreciate the comments. In our manuscript, sample scales of monocrystalline polygons and ribbons are up to ca.8 micrometers and 50 micrometers, respectively, which can be treated as a great breakthrough in 2D material.
Of course, sample scale satisfying the needs of real applications is important during our research. Actually, even now, few 2D materials can strictly meet the need of real application, which is common during our research. And, from my point of view, real application isn"t the only purpose of scientific research and the sole criterion to 5. The size of the 30-nm-thick antimonene is ca. 5 um, while the bending radius is ca. 5 mm, which is significantly larger than the sample. I doubt that the bending effect can hardly be applied to the sample.
Answer. We appreciate the the reviewer's comment. The bending radius is truly significantly too large to for us to confirm the bending effect, if we only take the size difference into consideration.
Firstly, during the test, mica was attached onto the surface of iron rod to ensure the bend effect was equally applied on mica, as shown in Figure 5d. Since the macroscopic bending exists, the bending effect is regarded to be applied on mica due to the smooth surface. Figure R2. Results and experimental details of bending test (Figure 5d).
Then, the electric conductivity decrease (see pink arrow) after bending test was observed, indicating that the bending effect was truly applied to the sample and caused the change of electric conductivity. This phenomenon we observed is highly consistent with Reference 23, where authors also carried out a bending test. The Te flake, in Reference 23, possess a similar scale with the Sb sheet, which further confirms the feasibility of our test.  To address the potential doubts, we add the following description in the revised manuscript.