Phase transformation kinetics in MoS2 governed by S-S repulsive interactions and defect-interface compatibility

Feng Ding; Pai Li; ZengFeng Di; Ziao Tian

arxiv: 2604.24209 · v1 · submitted 2026-04-27 · ❄️ cond-mat.mtrl-sci

Phase transformation kinetics in MoS2 governed by S-S repulsive interactions and defect-interface compatibility

Pai Li , Ziao Tian , ZengFeng Di , Feng Ding This is my paper

Pith reviewed 2026-05-08 02:52 UTC · model grok-4.3

classification ❄️ cond-mat.mtrl-sci

keywords MoS2phase transformation2D materialssulfur vacanciesgrain boundariesinterface compatibilitykinetic barriersmetastable phase

0 comments

The pith

Local compatibility between defects and interfaces, not global defect levels, controls phase transformations in MoS2

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper establishes that the metastable T' phase in monolayer MoS2 resists converting to the stable H phase because repulsive interactions between sulfur atoms create high energy barriers for both starting the change and moving the boundary forward. Sulfur vacancies, which lower barriers at some interfaces, become unstable at the lowest-energy interface and move away into the T' region instead, leaving the front free of defects. Nanostructure simulations show the H phase always begins at corners or edges and advances only along one specific interface type that is low in energy but slow in motion. This reframes control of structural changes in 2D materials around matching defects to particular interfaces rather than simply increasing overall defect numbers.

Core claim

Repulsive S-S interactions impose high energy barriers during both nucleation and grain boundary propagation in the T' to H phase transformation of monolayer MoS2. Sulfur vacancies alleviate barriers in certain interfaces but fail at the most stable ZZ-Mo|- interface due to thermodynamic instability, migrating instead into the T' phase and leaving the advancing front defect-free. Direct simulations confirm that H-phase nucleation initiates at corners or edges, with all observed growth fronts adopting the ZZ-Mo|- configuration consistent with its low interfacial energy but slow kinetics.

What carries the argument

The local compatibility between sulfur vacancies and specific advancing interfaces such as ZZ-Mo|-, which determines whether vacancies lower barriers or must migrate away

If this is right

H-phase nucleation begins at corners or edges of nanostructures.
All growth fronts adopt the ZZ-Mo|- interface due to its low energy but slow kinetics.
Vacancies migrate into the T' phase, leaving the ZZ-Mo|- front defect-free.
Transformation speed depends on local defect-interface matching rather than average defect concentration.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The same interface-compatibility rule may govern phase changes in other transition-metal dichalcogenides.
Nanostructure geometry or strain could be tuned to select interfaces that retain defects and speed transformations.
In-situ imaging of vacancy positions relative to moving fronts could directly test the migration mechanism.
Interface engineering might allow deliberate kinetic trapping of desired metastable phases in devices.

Load-bearing premise

The machine learning potential and first-principles calculations faithfully capture the S-S repulsive interactions and the relative stabilities of different interfaces and defects without significant systematic errors.

What would settle it

A simulation or observation in which sulfur vacancies remain stable at the ZZ-Mo|- interface and accelerate its propagation would falsify the claim that interface-specific compatibility governs the kinetics.

read the original abstract

The metastable T' phase in monolayer MoS2 exhibits remarkable persistence despite a strong thermodynamic driving force toward the stable H phase. Using machine learning-accelerated molecular dynamics and first-principles calculations, we reveal that this kinetic arrest originates from repulsive S-S interactions, which impose high energy barriers during both nucleation and grain boundary propagation. While sulfur vacancies can alleviate these barriers in certain interfaces, they fail to accelerate transformation at the most stable interface, ZZ-Mo|-, due to their thermodynamic instability there. Instead, vacancies migrate into the T' phase, leaving the advancing front defect-free. Direct simulations of nanostructures confirm that H-phase nucleation initiates at corners or edges, and all observed growth fronts adopt the ZZ-Mo|- configuration, consistent with its low interfacial energy but slow kinetics. Our work establishes that phase transformation in 2D materials is governed not by global defect concentration, but by the local compatibility between defects and moving interfaces, offering a new paradigm for controlling structural transitions through interface-specific design.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The simulations point to local interface-defect compatibility as the rate limiter in MoS2 T' to H transformation, but the ML potential's handling of vacancy-interface energies lacks the direct DFT checks needed to lock that in.

read the letter

The paper's core observation is that S-S repulsions raise barriers at the lowest-energy ZZ-Mo|- growth front, and vacancies there are unstable enough to migrate into the T' grain instead of staying to speed things up. Nucleation starts at corners or edges in the nanostructure runs, and every advancing front settles into that same configuration. That local-compatibility angle is the actual new piece; it moves beyond the usual global-defect-concentration story for these 2D transitions. The ML-accelerated MD on realistic flakes is the part that works well here, because it lets them watch the fronts evolve without assuming uniform conditions. The first-principles calculations are mentioned as support, which is the right direction. The soft spot is validation. The migration result and the claim that vacancies only help at non-stable interfaces both depend on the potential getting the relative formation energies and barriers right to within a few tens of meV. The stress-test note is fair: there is no reported direct DFT benchmark on a supercell that contains the moving ZZ-Mo|- interface plus a vacancy. If the potential under-ranks the repulsion at that front, the vacancies would stay put and the kinetics story reverts to a more conventional defect-assisted picture. The abstract is clear that first-principles work was done, but without the specific comparison numbers or system sizes it is hard to judge how tight the check was. This is for people who model or engineer phase changes in TMD monolayers. A reader who already works with ML potentials for MoS2 or similar systems will get the most out of the interface-specific design suggestion. The simulation evidence is concrete enough that a serious referee should see it, even if the validation section needs tightening.

Referee Report

1 major / 2 minor

Summary. The manuscript claims that the persistence of the metastable T' phase in monolayer MoS2, despite a thermodynamic drive to the H phase, arises from high energy barriers due to S-S repulsive interactions during nucleation and grain-boundary propagation. Sulfur vacancies lower barriers at some interfaces but are thermodynamically unstable at the lowest-energy ZZ-Mo|- front, migrating into the T' grain and leaving the advancing interface defect-free. Direct simulations of nanostructures show H-phase nucleation at corners/edges with all observed growth fronts adopting the ZZ-Mo|- configuration. The work concludes that phase transformation kinetics are controlled by local defect-interface compatibility rather than global defect concentration, proposing a new paradigm for interface-specific design in 2D materials.

Significance. If the underlying energetics hold, the result reframes phase transformation control in 2D materials around local interface-defect compatibility, with potential implications for designing stable or switchable structures via targeted interface engineering. The computational approach using ML-accelerated molecular dynamics for large-scale direct simulations of nanostructure evolution, combined with first-principles calculations, provides concrete mechanistic observations that go beyond mean-field defect models.

major comments (1)

[Results section on ZZ-Mo|- interface and vacancy migration] The central interpretive claim—that vacancies fail to accelerate transformation at the ZZ-Mo|- interface because they are thermodynamically unstable there and migrate away, leaving a defect-free front—depends on the ML potential correctly ranking vacancy formation energies and migration barriers relative to the interface. No direct DFT benchmark is reported for supercells containing the moving ZZ-Mo|- interface plus a vacancy (Methods or Results sections on interface energetics and vacancy behavior). If the potential underestimates the vacancy-interface repulsion by even ~0.05 eV, the migration direction reverses and the 'local compatibility' argument reduces to a conventional defect-assisted picture.

minor comments (2)

[Abstract] The abstract refers to 'first-principles calculations' without specifying which quantities (e.g., interface energies, vacancy formation energies) were recomputed with DFT or the size of the validation set against the ML potential.
[Simulation methods and results] Details on the system sizes, boundary conditions, and statistical sampling for the 'direct simulations of nanostructures' would improve reproducibility and allow assessment of finite-size effects on nucleation sites and front selection.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their careful and constructive review. The concern about direct validation of the ML potential for vacancy behavior at the dynamic ZZ-Mo|- interface is a substantive point that merits clarification. We address it below and indicate where the manuscript will be revised.

read point-by-point responses

Referee: [Results section on ZZ-Mo|- interface and vacancy migration] The central interpretive claim—that vacancies fail to accelerate transformation at the ZZ-Mo|- interface because they are thermodynamically unstable there and migrate away, leaving a defect-free front—depends on the ML potential correctly ranking vacancy formation energies and migration barriers relative to the interface. No direct DFT benchmark is reported for supercells containing the moving ZZ-Mo|- interface plus a vacancy (Methods or Results sections on interface energetics and vacancy behavior). If the potential underestimates the vacancy-interface repulsion by even ~0.05 eV, the migration direction reverses and the 'local compatibility' argument reduces to a conventional defect-assisted picture.

Authors: We agree that a direct DFT benchmark on supercells that simultaneously contain a moving ZZ-Mo|- interface and a vacancy would be ideal. Such calculations remain prohibitive because of the system sizes needed to accommodate both the extended interface and sufficient distance for vacancy migration. The ML potential was trained on a broad DFT dataset that includes vacancy formation energies at static ZZ-Mo|- interfaces (smaller supercells) as well as migration barriers within the T' phase; these comparisons are reported in the Methods and Supplementary Information. The thermodynamic preference for vacancies to leave the interface is therefore anchored in those validated energy differences rather than in an untested extrapolation. To address the referee’s concern explicitly, we will add a dedicated paragraph and supplementary figure in the revised manuscript that tabulates ML versus DFT vacancy formation energies for several static positions near the ZZ-Mo|- boundary, confirming that the repulsion is reproduced within 0.03 eV. We will also note the computational limitations that preclude full dynamic DFT validation at present. revision: yes

Circularity Check

0 steps flagged

No circularity; conclusions drawn from independent simulation outputs

full rationale

The paper's central claims rest on direct outputs of ML-accelerated MD trajectories and first-principles calculations that rank interface energies, vacancy formation energies, and migration barriers. No derivation step equates a 'prediction' to a fitted input by construction, nor does any load-bearing premise reduce to a self-citation whose content is itself unverified. The interpretation that kinetics are controlled by local defect-interface compatibility rather than global vacancy concentration is presented as an inference from the computed behaviors (vacancies unstable at ZZ-Mo|- fronts, nucleation at corners), not as a tautology or renamed known result. The work is self-contained against external benchmarks in the sense that its numerical results are generated ab initio within the study.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claim depends on the accuracy of the computational models and the assumption that observed simulation behaviors translate to physical reality.

axioms (2)

domain assumption The interatomic potential used in ML-MD accurately describes S-S repulsive interactions in both H and T' phases.
Central to the energy barrier calculations.
domain assumption The simulated nanostructures and interface models are representative of experimental conditions in monolayer MoS2.
For extrapolating to real phase transformation behavior.

pith-pipeline@v0.9.0 · 5476 in / 1310 out tokens · 54968 ms · 2026-05-08T02:52:22.050982+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

44 extracted references · 44 canonical work pages

[1]

Li, W., Qian, X. & Li, J. Phase transitions in 2D materials. Nat. Rev. Mater. 6, 829–846 (2021)

work page 2021
[2]

Yu, Y. et al. High phase-purity 1T’-MoS2- and 1T’-MoSe2-layered crystals. Nat. Chem. 10, 638–643 (2018)

work page 2018
[3]

Voiry, D. et al. Conducting MoS2 nanosheets as catalysts for hydrogen evolution reaction. Nano Lett. 13, 6222–6227 (2013)

work page 2013
[4]

Liu, L. et al. Phase-selective synthesis of 1T’ MoS2 monolayers and heterophase bilayers. Nat. Mater. 17, 1108–1114 (2018)

work page 2018
[5]

Shang, C. et al. Superconductivity in the metastable 1 T’ and 1 T”’ phases of MoS2 crystals. Phys. Rev. B 98, 184513 (2018)

work page 2018
[6]

F., McGill, K

Mak, K. F., McGill, K. L., Park, J. & McEuen, P. L. The valley Hall effect in MoS2 transistors. Science 344, 1489–1492 (2014)

work page 2014
[7]

Cheng, P., Sun, K. & Hu, Y. H. Memristive behavior and ideal memristor of 1T phase MoS2 nanosheets. Nano Lett. 16, 572–576 (2016)

work page 2016
[8]

& Chhowalla, M

Acerce, M., Voiry, D. & Chhowalla, M. Metallic 1T phase MoS2 nanosheets as supercapacitor electrode materials. Nat. Nanotechnol. 10, 313–318 (2015)

work page 2015
[9]

N., Li, Y

Duerloo, K.-A. N., Li, Y. & Reed, E. J. Structural phase transitions in two-dimensional Mo- and W-dichalcogenide monolayers. Nat. Commun. 5, 4214 (2014)

work page 2014
[10]

Enyashin, A. N. et al. New route for stabilization of 1T-WS2 and MoS2 phases. J. Phys. Chem. C 115, 24586–24591 (2011)

work page 2011
[11]

Ji, X. et al. Interlayer Coupling Dependent Discrete H → T’ Phase Transition in Lithium Intercalated Bilayer Molybdenum Disulfide. ACS Nano 15, 15039–15046 (2021)

work page 2021
[12]

O., Huang, Y.-S

Lin, Y.-C., Dumcenco, D. O., Huang, Y.-S. & Suenaga, K. Atomic mechanism of the semiconducting-to-metallic phase transition in single-layered MoS2. Nat. Nanotechnol. 9, 391–396 (2014)

work page 2014
[13]

Zhu, J. et al. Argon plasma induced phase transition in monolayer MoS2. J. Am. Chem. Soc. 139, 10216–10219 (2017)

work page 2017
[14]

Nam, D.-H. et al. Anion extraction-induced polymorph control of transition metal dichalcogenides. Nano Lett. 19, 8644–8652 (2019)

work page 2019
[15]

1T’-transition metal dichalcogenide monolayers stabilized on 4H-Au nanowires for ultrasensitive SERS detection

Li, Z. 1T’-transition metal dichalcogenide monolayers stabilized on 4H-Au nanowires for ultrasensitive SERS detection. Nat. Mater

work page
[16]

L., Johannes, M

Zhuang, H. L., Johannes, M. D., Singh, A. K. & Hennig, R. G. Doping-controlled phase transitions in single-layer MoS2. Phys. Rev. B 96, 165305 (2017)

work page 2017
[17]

Geng, X. et al. Pure and stable metallic phase molybdenum disulfide nanosheets for hydrogen evolution reaction. Nat. Commun. 7, 10672 (2016)

work page 2016
[18]

Kan, M. et al. Structures and phase transition of a MoS2 monolayer. J. Phys. Chem. C 118, 1515–1522 (2014)

work page 2014
[19]

& Ding, F

Zhao, W. & Ding, F. Energetics and kinetics of phase transition between a 2H and a 1T MoS2 monolayer—a theoretical study. Nanoscale 9, 2301–2309 (2017). 14

work page 2017
[20]

& Yakobson, B

Zou, X., Zhang, Z., Chen, X. & Yakobson, B. I. Structure and dynamics of the electronic heterointerfaces in MoS2 by first-principles simulations. J. Phys. Chem. Lett. 11, 1644–1649 (2020)

work page 2020
[21]

& Mei, D

Jin, Q., Liu, N., Chen, B. & Mei, D. Mechanisms of semiconducting 2H to metallic 1T phase transition in two-dimensional MoS2 nanosheets. J. Phys. Chem. C 122, 28215–28224 (2018)

work page 2018
[22]

L., Tchougréeff, A

Deringer, V. L., Tchougréeff, A. L. & Dronskowski, R. Crystal Orbital Hamilton Population (COHP) Analysis As Projected from Plane-Wave Basis Sets. J. Phys. Chem. A 115, 5461–5466 (2011)

work page 2011
[23]

Zeng, J. et al. DeePMD-kit v2: A software package for deep potential models. J. Chem. Phys. 159, 054801 (2023)

work page 2023
[24]

Xiao, Y., Zhou, M., Liu, J., Xu, J. & Fu, L. Phase engineering of two-dimensional transition metal dichalco- genides. Sci. China Mater. 62, 759–775 (2019)

work page 2019
[25]

& Medhekar, N

Bourgeois, L., Zhang, Y., Zhang, Z., Chen, Y. & Medhekar, N. V. Transforming solid-state precipitates via excess vacancies. Nat. Commun. 11, 1248 (2020)

work page 2020
[26]

Zhang, X. et al. Defect-characterized phase transition kinetics. Appl. Phys. Rev. 9, (2022)

work page 2022
[27]

Sekerka, R. F. Equilibrium and growth shapes of crystals: how do they differ and why should we care? Cryst. Res. Technol. 40, 291–306 (2005)

work page 2005
[28]

Marks, L. D. & Peng, L. Nanoparticle shape, thermodynamics and kinetics. J. Phys. Condens. Matter 28, 53001 (2016)

work page 2016
[29]

Tuning the phase stability of Mo-based TMD monolayers through coupled vacancy defects and lattice strain

Tang, Q. Tuning the phase stability of Mo-based TMD monolayers through coupled vacancy defects and lattice strain. J. Mater. Chem. C 6, 9561–9568 (2018)

work page 2018
[30]

Eda, G. et al. Coherent atomic and electronic heterostructures of single-layer MoS2. ACS Nano 6, 7311–7317 (2012)

work page 2012
[31]

& Furthmüller, J

Kresse, G. & Furthmüller, J. Eﬀiciency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set. Comput. Mater. Sci. 6, 15–50 (1996)

work page 1996
[32]

& Furthmüller, J

Kresse, G. & Furthmüller, J. Eﬀicient iterative schemes for ab initio total-energy calculations using a plane-wave basis set. Phys. Rev. B 54, 11169–11186 (1996)

work page 1996
[33]

P., Burke, K

Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized Gradient Approximation Made Simple. Phys. Rev. Lett. 77, 3865–3868 (1996)

work page 1996
[34]

Blöchl, P. E. Projector augmented-wave method. Phys. Rev. B 50, 17953–17979 (1994)

work page 1994
[35]

Monkhorst, H. J. & Pack, J. D. Special points for Brillouin-zone integrations. Phys. Rev. B 13, 5188–5192 (1976)

work page 1976
[36]

Li, P., Zeng, X. & Li, Z. Understanding High-Temperature Chemical Reactions on Metal Surfaces: A Case Study on Equilibrium Concentration and Diffusivity of CxHy on a Cu(111) Surface. JACS Au 2, 443–452 (2022)

work page 2022
[37]

& Yakobson, B

Zou, X., Zhang, Z., Chen, X. & Yakobson, B. I. Structure and Dynamics of the Electronic Heterointerfaces in MoS2 by First-Principles Simulations. J. Phys. Chem. Lett. 11, 1644–1649 (2020)

work page 2020
[38]

& Williams, L

Abdi, H. & Williams, L. J. Principal component analysis. WIREs Comput. Stat. 2, 433–459 (2010)

work page 2010
[39]

P., Kondor, R

Bartók, A. P., Kondor, R. & Csányi, G. On representing chemical environments. Phys. Rev. B 87, 184115 (2013). 15

work page 2013
[40]

Himanen, L. et al. DScribe: Library of descriptors for machine learning in materials science. Comput. Phys. Commun. 247, 106949 (2020)

work page 2020
[41]

Hjorth Larsen, A. et al. The atomic simulation environment—a Python library for working with atoms. J. Phys. Condens. Matter 29, 273002 (2017)

work page 2017
[42]

Fast Parallel Algorithms for Short-Range Molecular Dynamics

Plimpton, S. Fast Parallel Algorithms for Short-Range Molecular Dynamics. J. Comput. Phys. 117, 1–19 (1995)

work page 1995
[43]

Hoover, W. G. Canonical dynamics: Equilibrium phase-space distributions. Phys. Rev. A 31, 1695–1697 (1985)

work page 1985
[44]

& Yanilkin, A

Korotaev, P., Belov, M. & Yanilkin, A. Reproducibility of vibrational free energy by different methods. Comput. Mater. Sci. 150, 47–53 (2018). Acknowledgements We acknowledge funding supports from the Strategic Priority Research Program of the Chinese Academy of Sciences (Grant Nos. XDB0670000), the National Natural Science Foundation of China (Grant No. ...

work page 2018

[1] [1]

Li, W., Qian, X. & Li, J. Phase transitions in 2D materials. Nat. Rev. Mater. 6, 829–846 (2021)

work page 2021

[2] [2]

Yu, Y. et al. High phase-purity 1T’-MoS2- and 1T’-MoSe2-layered crystals. Nat. Chem. 10, 638–643 (2018)

work page 2018

[3] [3]

Voiry, D. et al. Conducting MoS2 nanosheets as catalysts for hydrogen evolution reaction. Nano Lett. 13, 6222–6227 (2013)

work page 2013

[4] [4]

Liu, L. et al. Phase-selective synthesis of 1T’ MoS2 monolayers and heterophase bilayers. Nat. Mater. 17, 1108–1114 (2018)

work page 2018

[5] [5]

Shang, C. et al. Superconductivity in the metastable 1 T’ and 1 T”’ phases of MoS2 crystals. Phys. Rev. B 98, 184513 (2018)

work page 2018

[6] [6]

F., McGill, K

Mak, K. F., McGill, K. L., Park, J. & McEuen, P. L. The valley Hall effect in MoS2 transistors. Science 344, 1489–1492 (2014)

work page 2014

[7] [7]

Cheng, P., Sun, K. & Hu, Y. H. Memristive behavior and ideal memristor of 1T phase MoS2 nanosheets. Nano Lett. 16, 572–576 (2016)

work page 2016

[8] [8]

& Chhowalla, M

Acerce, M., Voiry, D. & Chhowalla, M. Metallic 1T phase MoS2 nanosheets as supercapacitor electrode materials. Nat. Nanotechnol. 10, 313–318 (2015)

work page 2015

[9] [9]

N., Li, Y

Duerloo, K.-A. N., Li, Y. & Reed, E. J. Structural phase transitions in two-dimensional Mo- and W-dichalcogenide monolayers. Nat. Commun. 5, 4214 (2014)

work page 2014

[10] [10]

Enyashin, A. N. et al. New route for stabilization of 1T-WS2 and MoS2 phases. J. Phys. Chem. C 115, 24586–24591 (2011)

work page 2011

[11] [11]

Ji, X. et al. Interlayer Coupling Dependent Discrete H → T’ Phase Transition in Lithium Intercalated Bilayer Molybdenum Disulfide. ACS Nano 15, 15039–15046 (2021)

work page 2021

[12] [12]

O., Huang, Y.-S

Lin, Y.-C., Dumcenco, D. O., Huang, Y.-S. & Suenaga, K. Atomic mechanism of the semiconducting-to-metallic phase transition in single-layered MoS2. Nat. Nanotechnol. 9, 391–396 (2014)

work page 2014

[13] [13]

Zhu, J. et al. Argon plasma induced phase transition in monolayer MoS2. J. Am. Chem. Soc. 139, 10216–10219 (2017)

work page 2017

[14] [14]

Nam, D.-H. et al. Anion extraction-induced polymorph control of transition metal dichalcogenides. Nano Lett. 19, 8644–8652 (2019)

work page 2019

[15] [15]

1T’-transition metal dichalcogenide monolayers stabilized on 4H-Au nanowires for ultrasensitive SERS detection

Li, Z. 1T’-transition metal dichalcogenide monolayers stabilized on 4H-Au nanowires for ultrasensitive SERS detection. Nat. Mater

work page

[16] [16]

L., Johannes, M

Zhuang, H. L., Johannes, M. D., Singh, A. K. & Hennig, R. G. Doping-controlled phase transitions in single-layer MoS2. Phys. Rev. B 96, 165305 (2017)

work page 2017

[17] [17]

Geng, X. et al. Pure and stable metallic phase molybdenum disulfide nanosheets for hydrogen evolution reaction. Nat. Commun. 7, 10672 (2016)

work page 2016

[18] [18]

Kan, M. et al. Structures and phase transition of a MoS2 monolayer. J. Phys. Chem. C 118, 1515–1522 (2014)

work page 2014

[19] [19]

& Ding, F

Zhao, W. & Ding, F. Energetics and kinetics of phase transition between a 2H and a 1T MoS2 monolayer—a theoretical study. Nanoscale 9, 2301–2309 (2017). 14

work page 2017

[20] [20]

& Yakobson, B

Zou, X., Zhang, Z., Chen, X. & Yakobson, B. I. Structure and dynamics of the electronic heterointerfaces in MoS2 by first-principles simulations. J. Phys. Chem. Lett. 11, 1644–1649 (2020)

work page 2020

[21] [21]

& Mei, D

Jin, Q., Liu, N., Chen, B. & Mei, D. Mechanisms of semiconducting 2H to metallic 1T phase transition in two-dimensional MoS2 nanosheets. J. Phys. Chem. C 122, 28215–28224 (2018)

work page 2018

[22] [22]

L., Tchougréeff, A

Deringer, V. L., Tchougréeff, A. L. & Dronskowski, R. Crystal Orbital Hamilton Population (COHP) Analysis As Projected from Plane-Wave Basis Sets. J. Phys. Chem. A 115, 5461–5466 (2011)

work page 2011

[23] [23]

Zeng, J. et al. DeePMD-kit v2: A software package for deep potential models. J. Chem. Phys. 159, 054801 (2023)

work page 2023

[24] [24]

Xiao, Y., Zhou, M., Liu, J., Xu, J. & Fu, L. Phase engineering of two-dimensional transition metal dichalco- genides. Sci. China Mater. 62, 759–775 (2019)

work page 2019

[25] [25]

& Medhekar, N

Bourgeois, L., Zhang, Y., Zhang, Z., Chen, Y. & Medhekar, N. V. Transforming solid-state precipitates via excess vacancies. Nat. Commun. 11, 1248 (2020)

work page 2020

[26] [26]

Zhang, X. et al. Defect-characterized phase transition kinetics. Appl. Phys. Rev. 9, (2022)

work page 2022

[27] [27]

Sekerka, R. F. Equilibrium and growth shapes of crystals: how do they differ and why should we care? Cryst. Res. Technol. 40, 291–306 (2005)

work page 2005

[28] [28]

Marks, L. D. & Peng, L. Nanoparticle shape, thermodynamics and kinetics. J. Phys. Condens. Matter 28, 53001 (2016)

work page 2016

[29] [29]

Tuning the phase stability of Mo-based TMD monolayers through coupled vacancy defects and lattice strain

Tang, Q. Tuning the phase stability of Mo-based TMD monolayers through coupled vacancy defects and lattice strain. J. Mater. Chem. C 6, 9561–9568 (2018)

work page 2018

[30] [30]

Eda, G. et al. Coherent atomic and electronic heterostructures of single-layer MoS2. ACS Nano 6, 7311–7317 (2012)

work page 2012

[31] [31]

& Furthmüller, J

Kresse, G. & Furthmüller, J. Eﬀiciency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set. Comput. Mater. Sci. 6, 15–50 (1996)

work page 1996

[32] [32]

& Furthmüller, J

Kresse, G. & Furthmüller, J. Eﬀicient iterative schemes for ab initio total-energy calculations using a plane-wave basis set. Phys. Rev. B 54, 11169–11186 (1996)

work page 1996

[33] [33]

P., Burke, K

Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized Gradient Approximation Made Simple. Phys. Rev. Lett. 77, 3865–3868 (1996)

work page 1996

[34] [34]

Blöchl, P. E. Projector augmented-wave method. Phys. Rev. B 50, 17953–17979 (1994)

work page 1994

[35] [35]

Monkhorst, H. J. & Pack, J. D. Special points for Brillouin-zone integrations. Phys. Rev. B 13, 5188–5192 (1976)

work page 1976

[36] [36]

Li, P., Zeng, X. & Li, Z. Understanding High-Temperature Chemical Reactions on Metal Surfaces: A Case Study on Equilibrium Concentration and Diffusivity of CxHy on a Cu(111) Surface. JACS Au 2, 443–452 (2022)

work page 2022

[37] [37]

& Yakobson, B

Zou, X., Zhang, Z., Chen, X. & Yakobson, B. I. Structure and Dynamics of the Electronic Heterointerfaces in MoS2 by First-Principles Simulations. J. Phys. Chem. Lett. 11, 1644–1649 (2020)

work page 2020

[38] [38]

& Williams, L

Abdi, H. & Williams, L. J. Principal component analysis. WIREs Comput. Stat. 2, 433–459 (2010)

work page 2010

[39] [39]

P., Kondor, R

Bartók, A. P., Kondor, R. & Csányi, G. On representing chemical environments. Phys. Rev. B 87, 184115 (2013). 15

work page 2013

[40] [40]

Himanen, L. et al. DScribe: Library of descriptors for machine learning in materials science. Comput. Phys. Commun. 247, 106949 (2020)

work page 2020

[41] [41]

Hjorth Larsen, A. et al. The atomic simulation environment—a Python library for working with atoms. J. Phys. Condens. Matter 29, 273002 (2017)

work page 2017

[42] [42]

Fast Parallel Algorithms for Short-Range Molecular Dynamics

Plimpton, S. Fast Parallel Algorithms for Short-Range Molecular Dynamics. J. Comput. Phys. 117, 1–19 (1995)

work page 1995

[43] [43]

Hoover, W. G. Canonical dynamics: Equilibrium phase-space distributions. Phys. Rev. A 31, 1695–1697 (1985)

work page 1985

[44] [44]

& Yanilkin, A

Korotaev, P., Belov, M. & Yanilkin, A. Reproducibility of vibrational free energy by different methods. Comput. Mater. Sci. 150, 47–53 (2018). Acknowledgements We acknowledge funding supports from the Strategic Priority Research Program of the Chinese Academy of Sciences (Grant Nos. XDB0670000), the National Natural Science Foundation of China (Grant No. ...

work page 2018