Pith. sign in

REVIEW 4 major objections 5 minor 48 references

Kinetics of Vacancy-Assisted Reversible Phase Transition in Monolayer MoTe$_2$

T0 review · 4 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper claims that reversible 2H–1T' switching in monolayer MoTe2 is governed by tellurium vacancies, not by unit-cell barriers: Te divacancies nucleate triangular 1T' islands, while the reverse transition is diffusionless and needs no…

desk verdict A genuinely new vacancy-mediated mechanism for the MoTe2 phase switch, but the quantitative kinetics rest on a single unbenchmarked MLIP. read the letter →

arxiv 2507.12565 v1 pith:VJKMOWI4 submitted 2025-07-16 cond-mat.mtrl-sci cond-mat.mes-hall

Fei Shuang , Daniel Ocampo , Reza Namakian , Arman Ghasemi , Poulumi Dey , Wei Gao This is my paper
classification cond-mat.mtrl-scicond-mat.mes-hall
keywords MoTe2monolayer2Hto1T'phasetransitiontelluriumvacanciesdivacancynucleationmachine-learnedinteratomicpotentialkineticMonteCarlomean-fieldkineticsreversibleswitching
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper argues that the 2H-to-1T' transition in monolayer MoTe2 is not a concerted unit-cell distortion but a vacancy-mediated nucleation-and-growth process, and that the reverse transition is a fast, diffusionless unwinding. Using a machine-learned interatomic potential trained on SCAN-DFT data, with molecular dynamics, free-energy barrier calculations, kinetic Monte Carlo, and a mean-field kinetic model, the authors identify the load-bearing events: Te monovacancies coalesce into mobile divacancies, divacancies migrate and seed triangular 1T' islands, and islands grow row by row by absorbing vacancies at their alpha edges. They further find that when the external stimulus is removed, the 1T' island reverts to 2H from its corners inward, leaving three-fold spoke-like vacancy lines, and that the subsequent 2H-to-1T' cycles are rapid, do not require new vacancies, and can be driven by mild stimuli. This matters because reversible, low-energy phase switching in MoTe2 is a candidate mechanism for reconfigurable electronics and memory devices, and the paper turns it into a concrete two-stage recipe: strong-stimulus pre-device nucleation followed by mild-stimulus in-device cycling.

What carries the argument

The object that carries the argument is the tellurium divacancy: two adjacent Te vacancies behave as a mobile defect whose coalescence, migration, and encounter with a single Te vacancy nucleate triangular 1T' islands. Around that object the paper builds a phase-front kinetic machinery: forward growth is described by a mean-field model whose variables are the number of live phase fronts per row and the transformed fraction per row, with front birth proportional to vacancy density and front death by collision; the reverse transition is described by a row-by-row equation in which corner atoms hop first and interior atoms require one or two parent atoms to hop beforehand. The kinetic Monte Carlo simulations use barriers extracted from minimum free-energy paths, so the mean-field equations and the simulations are compared on equal footing.

What would settle it

A DFT nudged-elastic-band calculation of the first Te-atom hop at the alpha phase boundary under 4% armchair strain is the sharpest test: if that barrier is clearly above the roughly 0.4 eV used in the kinetic Monte Carlo, then the growth rates, the critical-size scaling, and the claimed ease of reversible switching lose quantitative support. An experimental counterpart is to image a pristine 2H monolayer under mild strain in a transmission electron microscope and look for the spoke-like vacancy lines that the model says must be created by a stronger first stimulus before mild reversible switching can occur.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that the kinetic bottleneck of the 2H-to-1T' transition in MoTe2 lives at the defect level, not in the unit-cell lattice distortion. Under 4% armchair strain with about 5% Te vacancies at 600 K, the first step is the coalescence of second-nearest-neighbor Te monovacancies into a divacancy, with a free-energy barrier of 0.76 to 0.86 eV, while isolated Te monovacancy migration stays expensive and is not the route. The divacancy is mobile, and when it meets another Te vacancy it nucleates a triangular 1T' island. Growth then proceeds by a site-by-site phase-front mechanism along the two alpha boundaries, with a maximum barrier around 0.4 eV, or by a vacancy-free mode once the island edge exceeds roughly 10 nm, at a higher barrier near 0.9 eV; a beta-boundary mode fed by migrating divacancies is a secondary pathway. Upon unloading, the island shrinks from its three corners toward the center, the reverse transition being about an order of magnitude faster than the forward one, and the Te vacancies that enabled growth are left as three-fold spoke-like lines in the 2H lattice. The paper concludes that the spatial distribution and mobility of Te vacancies, not unit-cell barriers, govern both whether and how fast the phase switch occurs.

Load-bearing premise

The load-bearing premise is that the machine-learned potential is quantitatively reliable for exactly the defect and phase-boundary events it was not directly benchmarked on; if its 0.4 eV growth barrier or its divacancy-coalescence barriers are wrong, the kinetic conclusions shift with them.

Editorial extensions

If this is right

  • A one-time, stronger-stimulus nucleation step is enough: once spoke-like vacancy lines are left behind, cyclic 2H-to-1T' switching is diffusionless, rapid, and achievable with mild strain, so MoTe2 devices could be conditioned before use.
  • Growth speed and final morphology are set by vacancy concentration: at low density the island stalls until its edge is long enough, with the critical length scaling roughly as the inverse of vacancy density, while higher density gives faster row-by-row expansion.
  • Both armchair and zigzag tensile strain can favor the 1T' phase, because the three symmetry-equivalent variants resolve applied strain differently, correcting the earlier conclusion that zigzag tension suppresses the transition.
  • Individual small 1T' islands at low vacancy density may sit still in a given realization, yet the mean-field growth rate is real because a small fraction of islands contains the needed vacancy.
  • Growth without vacancies becomes possible beyond a critical island size, around a 10 nm edge under 4% strain, but with a higher barrier, giving a second, slower route to phase patterning.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If vacancies govern nucleation generally, then switching statistics in nanoscale MoTe2 devices should be stochastic: write times should vary with the local Te-vacancy density, a signature that in situ transmission electron microscopy or transport experiments could look for.
  • The 0.4 eV alpha-boundary growth barrier is a sharp prediction; computing that exact hop with density functional theory at 4% armchair strain would confirm whether the machine-learned-potential kinetic picture holds or needs revision.
  • The two-stage pre-device and in-device picture offers an explanation for irreproducible mild-stimulus switching reports, but the authors flag that explanation as a cautious hypothesis without direct experimental evidence, so it needs a targeted conditioning experiment.
  • The same phase-front birth-death model could be transferred to other 2H-to-1T' transition metal dichalcogenide monolayers by rescaling the vacancy density, hopping barriers, and row geometry, once equivalent defect pathways are verified for those materials.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper develops a moment tensor potential for monolayer MoTe2 trained on SCAN-DFT data (10,966 configurations, active learning, 4.78/4.89 meV/atom energy MAE, 65.75/64.98 meV/Å force RMSE, 0.53/0.52 eV stress MAE) and uses MLIP-based MD, NEB/PAFI free-energy calculations, kinetic Monte Carlo, and mean-field kinetic equations to propose a two-stage picture of the 2H-to-1T' transition. The central claim is that the initial transition is nucleation-limited and vacancy-mediated: Te monovacancies coalesce into mobile divacancies (barrier 0.86 eV at 300 K), which seed triangular 1T' islands; growth proceeds by vacancy absorption at alpha boundaries (E_f ~ 0.4 eV), by divacancy migration, or, above a critical size, by a higher-barrier vacancy-free mechanism (0.97 eV). Upon unloading, the reverse transition is claimed to be diffusionless, rapid, and vacancy-insensitive, leaving three-fold spoke-like vacancy lines that enable reversible switching under mild stimuli. The paper also proposes a pre-device/in-device two-stage phase-engineering strategy and a qualitative explanation for irreproducibility of some phase-transition experiments.

Significance. If the proposed mechanism is correct, it is significant because it replaces the unit-cell concerted-barrier picture with an explicit vacancy-nucleation and island-growth picture, and it offers falsifiable predictions: a critical island size scaling inversely with vacancy density, a three-fold vacancy-line remnant after reverse transformation, and a distinction between hard-to-reproduce initial switching and easy subsequent cycling. Strengths of the manuscript include the systematic MLIP training with active learning, validation against phonons and elastic constants, use of PAFI for free-energy barriers, unbiased MD that discovers rather than prescribes pathways, and the public release of code, DFT dataset, and MLIP. The KMC/mean-field comparison is transparent and internally consistent. However, the quantitative conclusions—timescales, critical sizes, and reversibility under mild stimuli—depend on a small set of MLIP-computed barriers that are not benchmarked against DFT, and the agreement between KMC and mean-field theory is not an independent check because both use the same fitted input parameters.

major comments (4)
  1. [Machine learning interatomic potential; Fig. 1; Table S1] The rate-determining barriers that enter the KMC and mean-field models—MV coalescence (0.86 eV at 300 K, Fig. 3d), DV migration (0.75 eV, Table S2), vacancy-assisted alpha-edge growth (E_f=0.4 eV, Methods), and vacancy-free growth (0.97 eV, Table S2)—are computed only with the MLIP, whose force RMSE is 65.75 meV/Å and whose elastic constants deviate by up to 15.45% (1T' C22) and 17.995% (2H C12) in Table S1. No direct MTP-versus-DFT comparison is reported for these specific defect and phase-boundary pathways. Since rates depend on barriers as exp(-E/kT) with v0=1.5e13 s^-1, a 0.1 eV systematic error changes a room-temperature rate by roughly 50x and could reorder the G1/G2/G3 hierarchy and the inferred critical sizes. The authors should provide at least a direct DFT/MTP comparison for the key saddle points (or single-point DFT energies on the NEB paths) and a sensitivity analysis of the predicted timescales with respect to barrier uncertainties.
  2. [Eqs. (3)-(5); Figs. 4f-h, 5a] The agreement between the mean-field kinetic model and KMC is not an external validation of the physical mechanism, because both use the same E_f, E_b, and v0, and Eq. (5) introduces rho_mod specifically to emulate the KMC rule of at least one vacancy per row. This post hoc adjustment is acknowledged in the text, but the figures presenting the theory-KMC match should state clearly that the match verifies internal consistency of the two coarse-grained descriptions, not the accuracy of the underlying barrier set. To claim quantitative predictive power, the authors should compare KMC/mean-field predictions against MLIP-MD observations in a regime where both are applicable, or against experimental timescales, rather than only against each other.
  3. [Results: 'Key mechanisms revealed by MLIP-MD'; Methods: 'MD and tfMC simulations'; Supplementary Note 3] The central mechanism is inferred from MLIP-MD at 600 K with 5% vacancies and 4% armchair strain, and from tfMC at 800 K with a displacement amplitude of 0.25. The generalization to 'low vacancy density' and 'mild external stimuli' is not demonstrated: at dilute densities the coalescence step requires two second-nearest-neighbor monovacancies (Fig. 3a), whose probability is exponentially small, and the KMC simulations enforce at least one vacancy per row, so the early-stage growth behavior does not represent a truly random dilute distribution. Figures 7b and 7c rest on this assumed regime of validity. The paper should state explicitly the vacancy-density, strain, and temperature ranges for which the proposed mechanisms are established, and should either provide dilute-density simulations or quantitative estimates showing that the same pathways dominate.
  4. [Supplementary Note 3; Table S2] Supplementary Note 3 reports that the three key MD events occur on the order of 1 ns, while the transition-state-theory estimates in Table S2 give 770 ns for MV coalescence and 95 ns for DV migration at 600 K; the text attributes the discrepancy to 'the presence of other defects.' This is an unexplained order-of-magnitude discrepancy in the very rates the paper uses to support its timescale claims. The authors should either resolve this discrepancy (for example by verifying the event definitions in MD and the saddle-point assignment), or explicitly state that the tabulated TST timescales are not directly comparable with the MD observations and should not be used for quantitative predictions.
minor comments (5)
  1. [References] Reference [33] contains the phrase 'this is very important!!!!' in the visible reference text, which is unprofessional and should be removed; several references (e.g., [17], [19], [28], [33]) also contain duplicated URLs that should be cleaned.
  2. [Methods: DFT calculations] 'V asPkit' should be 'VASPKIT'.
  3. [Results: '1T' phase growth via absorption of random vacancies'] The phrase 'time evaluation of 1T' area' should read 'time evolution of the 1T' area'.
  4. [Alternative 1T' phase growth pathways] The text 'tfMC simulations (Figs. S8 and S9)' contains a typo ('tfMC' should be 'tfMC' or spelled out), and the tfMC method is not introduced until the Methods section; a forward reference to Methods would help the reader.
  5. [Abstract and Summary] The claim that reversible transitions 'can be driven by mild external stimuli' is not quantified anywhere in the manuscript; the simulations use 4% strain and temperatures of 600-800 K, so 'mild' should be defined or qualified to avoid overstating the result.

Circularity Check

1 steps flagged · score 4.0 of 10

Mostly self-contained vacancy-mediated mechanism; one localized circular step: the ρmod correction in Eq. (5) is constructed to emulate the KMC per-row vacancy rule, so the subsequent dashed-curve agreement is a forced consistency check, not an independent prediction.

  1. fitted input called prediction [Section '1T′ phase growth via absorption of random vacancies'; Eq. (5) and Fig. 4f-h.]
    "To make the mean field model emulate the KMC rule (at least one vacancy per row), we introduce a modified vacancy density ρmod(i)= ρ/(1−(1−ρ)^{N_i}) ... Replacing ρ in Eq. (3) with ρmod and re-solving the equations yields the results represented by dashed lines in Fig. 4f–h. At early times, the growth rate predicted with ρmod closely matches the KMC results across all three densities."

    The modified density ρmod is defined by conditioning the bulk density on the event that at least one vacancy is present in row i, which is exactly the rule enforced in the KMC setup ('we ensure at least one vacancy per row'). Re-solving Eq. (3) with this parameter is therefore not an independent confirmation of the KMC curves; the dashed curves are constructed to encode the KMC rule. The paper presents the outcome as 'the growth rate predicted with ρmod closely matches the KMC results', which is a consistency check rather than a prediction. This step is disclosed and local, and the paper's central mechanism claim does not depend on this agreement, but the specific comparison in Fig. 4f-h is circular by construction.

full rationale

The central claims—nucleation by Te-monovacancy coalescence, mobile divacancies, triangular-island growth, and the diffusionless reverse transition—are grounded in MLIP-MD trajectories (Fig. S3) and in MEP/PAFI barrier calculations from the trained MLIP, not in the mean-field fit. The MLIP is trained on SCAN-DFT and validated on held-out energies, forces, elastic constants, and phonon spectra; no self-citation chain or imported uniqueness theorem is load-bearing. The main weakness is quantitative verification: the barrier set that drives KMC and mean-field rates comes from the MLIP, whose training data include NEB transition states for vacancy migration and 2H-to-1T' transitions, and the paper reports no direct MTP-versus-DFT comparison for the specific rate-limiting barriers; that is an accuracy risk, not a constructional circularity. Similarly, the agreement between mean-field theory and KMC in Figs. 4f-h and 5a is an internal consistency check because both calculations use the same barrier values (Ef=0.4 eV, Eb=0.42 eV) and attempt frequency v0=1.5e13 s^-1; it does not externally validate those barriers. The one explicitly constructional element is Eq. (5), where ρmod is chosen to emulate the KMC one-vacancy-per-row rule, making the dashed-curve match expected rather than predictive. Overall, the derivation chain is mostly self-contained, with one localized, self-disclosed circular step; hence a moderate score rather than a high one.

Assumptions & free parameters 5 free parameters · 7 assumptions · 0 invented entities

The central kinetics rest on (i) an MLIP whose defect-path accuracy is not independently benchmarked, (ii) Arrhenius rate assumptions and uniform-vacancy statistics, (iii) an ad hoc KMC one-vacancy-per-row rule patched into the mean-field model via Eq. (5), and (iv) an extrapolation from high-strain, high-vacancy driven MD to dilute mild-stimulus conditions. No invented physical entities are added; the spoke-like vacancy lines and triangular islands are defect configurations observed in the simulations.

free parameters (5)
  • Forward Te-hop barrier E_f for 2H to 1T' growth = 0.4 eV
    Adopted as representative of the highest barrier among sequential alpha-edge hops from MLIP-NEB; used to set every forward hop rate in KMC and the mean-field model (Methods, KMC section).
  • Backward Te-hop barrier E_b = 0.42 eV
    Set as E_f + Delta_E_cell with Delta_E_cell = 0.02 eV phase free-energy difference at 4% strain; controls reverse rates and hence the predicted growth asymmetry.
  • Attempt frequency v0 = 1.5e13 s^-1
    Assumed standard atomic attempt frequency; enters every hopping rate and all timescales in Table S2.
  • MTP cutoff radius and maximum moment level = Rcut=7 Angstrom, levmax=20
    Chosen MLIP hyperparameters that determine descriptor accuracy; all barriers inherit their errors.
  • Critical growth probability threshold = 90%
    Used to define Lc, the edge length beyond which growth is statistically uninterrupted; the threshold is a practical choice, not derived from the physics.
assumptions (7)
  • domain assumption Arrhenius hopping rates with a single attempt frequency v0 describe all Te hops.
    KMC rate v_i = v0 exp(-E_i/kBT) and Eq. (1) net rate are the backbone of all kinetic predictions (Methods, KMC section).
  • domain assumption Te vacancies are initially uniformly and independently distributed with density rho.
    Eq. (3) uses birth rate 2 rho v_n F_{i-1}, and rho is later replaced by rho_mod (Eq. 5) because the KMC setup violates uniform placement.
  • domain assumption Phase fronts within a row are evenly spaced.
    Eq. (2) mean travel distance and the death-rate term v_n/l_avg require this mean-field ordering assumption.
  • domain assumption The MLIP trained on SCAN-DFT is accurate enough for the specific defect and phase-boundary pathways studied, beyond its training/test metrics.
    All NEB/PAFI barriers and MD snapshot mechanisms come from the MTP; independent DFT checks for the new divacancy and phase-boundary paths are not provided (Fig. 1, Table S1).
  • ad hoc to paper The KMC rule of at least one vacancy per row is physically representative of growth at dilute vacancy densities.
    The rule is imposed to keep growth continuous; authors acknowledge it biases early growth and compensate with Eq. (5).
  • ad hoc to paper Mechanisms found under 4% strain with 5% vacancies generalize to other stimuli and to dilute, near-ambient conditions.
    The abstract and Summary assert generalizability; no non-strain or experimental evidence is provided.
  • domain assumption Reverse transition sequence is governed only by one- and two-parent dependencies, with all vacancies assumed to lie exactly on spoke lines.
    Eq. (6) uses f_{i-1} and f^2_{i-1}; authors attribute the theory's overestimate to this idealization (Fig. 5a).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Kinetics of Vacancy-Assisted Reversible Phase Transition in Monolayer MoTe$_2$." pith.science (2026). https://pith.science/paper/VJKMOWI4

@misc{pith2026250712565,
  author       = {Pith},
  title        = {Pith review of: Kinetics of Vacancy-Assisted Reversible Phase Transition in Monolayer MoTe$_2$},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VJKMOWI4}},
  note         = {Machine review of arXiv:2507.12565}
}
abstract

We investigate the kinetics of phase transition between the 2H and 1T$^\prime$ phases in monolayer MoTe$_2$ using atomistic simulations based on a machine learning interatomic potential trained on SCAN-DFT data, combined with mean field kinetic theory to interpret the underlying mechanisms. The transition is found to involve both diffusive and diffusionless mechanisms. Nucleation of 1T$^\prime$ phase is initiated by the coalescence of neighboring Te monovacancies into divacancies, which are found to be mobile and can interact with other Te vacancies to form small triangular 1T$^\prime$ islands. Growth of these islands proceeds either by incorporating pre-existing vacancies at the phase boundaries or, in their absence, by absorbing divacancies that migrate from the surrounding lattice. Once a critical island size is reached, vacancy-free growth becomes possible although with a higher activation barrier. Upon removal of external stimuli, the system reverts to 2H phase, during which Te vacancies reorganize into three-fold spoke-like vacancy lines at the island center. This reverse process and the subsequent 1T$^\prime$$\leftrightarrow$2H reversible transitions are diffusionless, rapid, do not require additional vacancies and can be driven by mild external stimuli. Although our analysis focuses on strain-induced transitions, the kinetic mechanisms are expected to be generalizable to other types of stimuli.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

48 extracted references · 34 canonical work pages

  1. [33]

    J. H. Kim, H. Sung, G. H. Lee, Phase Engineering of Two-Dimensional Transition Metal Dichalcogenides. Small Science 4, 2300093 (2024), this is very important!!!!, doi:10. 1002/SMSC.202300093, https://onlinelibrary.wiley.com/doi/full/10.1002/ smsc.202300093https://onlinelibrary.wiley.com/doi/abs/10.1002/smsc. 202300093https://onlinelibrary.wiley.com/doi/10...

    work page doi:10.1002/smsc 2024

  2. [1]

    Wuttig, N

    M. Wuttig, N. Yamada, Phase-change materials for rewriteable data storage.Nature materials 6 (11), 824 (2007)

    work page 2007

  3. [2]

    Wang, et al., Optically reconfigurable metasurfaces and photonic devices based on phase change materials

    Q. Wang, et al., Optically reconfigurable metasurfaces and photonic devices based on phase change materials. Nature Photonics 10 (1), 60 (2016)

    work page 2016

  4. [3]

    X. Qian, J. Liu, L. Fu, J. Li, Quantum spin Hall effect in two-dimensional transition metal dichalcogenides. Science 346 (6215), 1344–1347 (2014)

    work page 2014

  5. [4]

    K.-A. N. Duerloo, Y. Li, E. J. Reed, Structural phase transitions in two-dimensional Mo- and W-dichalcogenide monolayers. Nature Communications 5, 4214 (2014), doi:10.1038/ ncomms5214, https://www.nature.com/articles/ncomms5214

    work page 2014

  6. [5]

    W. Li, X. Qian, J. Li, Phase transitions in 2D materials. Nature Reviews Materials 6, 829– 846 (2021), doi:10.1038/s41578-021-00304-0, https://www.nature.com/articles/ s41578-021-00304-0

    doi:10.1038/s41578-021-00304-0 2021

  7. [6]

    Y. Yoo, Z. P. DeGregorio, Y. Su, S. J. Koester, J. E. Johns, In-Plane 2H-1T ′ MoTe2 Ho- mojunctions Synthesized by Flux-Controlled Phase Engineering. Advanced Materials 29, 1605461 (2017), doi:10.1002/adma.201605461, https://onlinelibrary.wiley.com/ doi/10.1002/adma.201605461

    work page doi:10.1002/adma.201605461 2017

  8. [7]

    Zhou, et al., Intrinsic structural defects in monolayer molybdenum disulfide.Nano Letters 13, 2615–2622 (2013), doi:10.1021/NL4007479

    W. Zhou, et al., Intrinsic structural defects in monolayer molybdenum disulfide.Nano Letters 13, 2615–2622 (2013), doi:10.1021/NL4007479

    doi:10.1021/nl4007479 2013

Show all 48 references
  1. [8]

    D. J. Hynek, et al., cm2-Scale Synthesis of MoTe2 Thin Films with Large Grains and Layer Control. ACS Nano 15, 410–418 (2021), doi:10.1021/acsnano.0c08069, https://pubs. acs.org/doi/10.1021/acsnano.0c08069

    2021 doi

  2. [9]

    D. H. Keum, et al. , Bandgap opening in few-layered monoclinic MoTe 2. Nature Physics 11, 482–486 (2015), doi:10.1038/nphys3314, https://www.nature.com/articles/ nphys3314. 26

    2015 doi

  3. [10]

    J. H. Sung, et al. , Coplanar semiconductor–metal circuitry defined on few-layer MoTe 2 via polymorphic heteroepitaxy. Nature Nanotechnology 12, 1064–1070 (2017), doi:10.1038/ nnano.2017.161, https://www.nature.com/articles/nnano.2017.161

    2017

  4. [11]

    K.-A. N. Duerloo, E. J. Reed, Structural Phase Transitions by Design in Monolayer Alloys. ACS Nano 10, 289–297 (2016), doi:10.1021/acsnano.5b04359, https://pubs.acs.org/ doi/10.1021/acsnano.5b04359

    2016 doi

  5. [12]

    Zou, et al., Atomic Insights into Phase Evolution in Ternary Transition-Metal Dichalco- genides Nanostructures

    Y. Zou, et al., Atomic Insights into Phase Evolution in Ternary Transition-Metal Dichalco- genides Nanostructures. Small 14, 1800780 (2018), doi:10.1002/smll.201800780, https: //onlinelibrary.wiley.com/doi/10.1002/smll.201800780

    2018 doi

  6. [13]

    Deng, et al., Controlled Synthesis of Mo𝑥W1–𝑥Te2 Atomic Layers with Emergent Quantum States

    Y. Deng, et al., Controlled Synthesis of Mo𝑥W1–𝑥Te2 Atomic Layers with Emergent Quantum States. ACS Nano 15, 11526–11534 (2021), doi:10.1021/acsnano.1c01441, https://pubs. acs.org/doi/10.1021/acsnano.1c01441

    2021 doi

  7. [15]

    Y.-C. Lin, D. O. Dumcenco, Y.-S. Huang, K. Suenaga, Atomic mechanism of the semiconducting-to-metallic phase transition in single-layered MoS2. Nature Nanotechnology 9, 391–396 (2014), doi:10.1038/nnano.2014.64, https://www.nature.com/articles/ nnano.2014.64

    2014 doi

  8. [16]

    Tan, et al

    Y. Tan, et al. , Controllable 2H-to-1T’ phase transition in few-layer MoTe 2. Nanoscale 10, 19964–19971 (2018), doi:10.1039/C8NR06115G, https://pubs.rsc.org/en/ content/articlehtml/2018/nr/c8nr06115ghttps://pubs.rsc.org/en/content/ articlelanding/2018/nr/c8nr06115g

    2018 doi

  9. [17]

    Si, et al., Photoinduced Vacancy Ordering and Phase Transition in MoTe 2

    C. Si, et al., Photoinduced Vacancy Ordering and Phase Transition in MoTe 2. Nano Letters 19, 3612–3617 (2019), doi:10.1021/acs.nanolett.9b00613, https://pubs.acs.org/doi/ 10.1021/acs.nanolett.9b00613. 27

    2019 doi

  10. [18]

    Ryu, et al., Laser-Induced Phase Transition and Patterning of hBN-Encapsulated MoTe2

    H. Ryu, et al., Laser-Induced Phase Transition and Patterning of hBN-Encapsulated MoTe2. Small 19, 2205224 (2023), doi:10.1002/smll.202205224, https://onlinelibrary. wiley.com/doi/10.1002/smll.202205224

    2023 doi

  11. [19]

    K ¨oster, et al., Phase transformations in single-layer MoTe2 stimulated by electron irradiation and annealing

    J. K ¨oster, et al., Phase transformations in single-layer MoTe2 stimulated by electron irradiation and annealing. Nanotechnology 35, 145301 (2024), doi:10.1088/1361-6528/ad15bb, https: //iopscience.iop.org/article/10.1088/1361-6528/ad15bb

    2024 doi

  12. [20]

    Cho, et al., Phase patterning for ohmic homojunction contact in MoTe2

    S. Cho, et al., Phase patterning for ohmic homojunction contact in MoTe2. Science 349, 625– 628 (2015), doi:10.1126/science.aab3175, https://www.science.org/doi/10.1126/ science.aab3175

    2015 doi

  13. [21]

    Sun, et al., Elastic Properties and Fracture Behaviors of Biaxially Deformed, Polymor- phic MoTe2

    Y. Sun, et al., Elastic Properties and Fracture Behaviors of Biaxially Deformed, Polymor- phic MoTe2. Nano Letters 19, 761–769 (2019), doi:10.1021/acs.nanolett.8b03833, https: //pubs.acs.org/doi/10.1021/acs.nanolett.8b03833

    2019 doi

  14. [22]

    Akinwande, et al

    D. Akinwande, et al. , A review on mechanics and mechanical properties of 2D materials—Graphene and beyond. Extreme Mechanics Letters 13, 42–77 (2017), doi:10.1016/j.eml.2017.01.008, https://linkinghub.elsevier.com/retrieve/pii/ S235243161630236X

    2017 doi

  15. [23]

    Song, et al., Room Temperature Semiconductor–Metal Transition of MoTe 2 Thin Films Engineered by Strain

    S. Song, et al., Room Temperature Semiconductor–Metal Transition of MoTe 2 Thin Films Engineered by Strain. Nano Letters 16, 188–193 (2016), doi:10.1021/acs.nanolett.5b03481, https://pubs.acs.org/doi/10.1021/acs.nanolett.5b03481

    2016 doi

  16. [24]

    Kang, et al., Plasmonic Hot Electron Induced Structural Phase Transition in a MoS2 Mono- layer

    Y. Kang, et al., Plasmonic Hot Electron Induced Structural Phase Transition in a MoS2 Mono- layer. Advanced Materials 26, 6467–6471 (2014), doi:10.1002/adma.201401802, https: //onlinelibrary.wiley.com/doi/10.1002/adma.201401802

    2014 doi

  17. [25]

    Kim, et al., Long-Range Lattice Engineering of MoTe2 by a 2D Electride

    S. Kim, et al., Long-Range Lattice Engineering of MoTe2 by a 2D Electride. Nano Letters 17, 3363–3368 (2017), doi:10.1021/acs.nanolett.6b05199, https://pubs.acs.org/doi/10. 1021/acs.nanolett.6b05199

    2017 doi

  18. [26]

    Li, K.-A

    Y. Li, K.-A. N. Duerloo, K. Wauson, E. J. Reed, Structural semiconductor-to-semimetal phase transition in two-dimensional materials induced by electrostatic gating. Nature Communica- 28 tions 7, 10671 (2016), doi:10.1038/ncomms10671,https://www.nature.com/articles/ ncomms10671

    2016 doi

  19. [27]

    Wang, et al., Structural phase transition in monolayer MoTe2 driven by electrostatic dop- ing

    Y. Wang, et al., Structural phase transition in monolayer MoTe2 driven by electrostatic dop- ing. Nature 550, 487–491 (2017), doi:10.1038/nature24043, https://www.nature.com/ articles/nature24043

    2017 doi

  20. [28]

    S. S. Awate, et al., Strain-Induced 2H to 1T′ Phase Transition in Suspended MoTe 2 Using Electric Double Layer Gating. ACS Nano 17, 22388–22398 (2023), doi:10.1021/acsnano. 3c04701, https://pubs.acs.org/doi/10.1021/acsnano.3c04701

    2023 doi

  21. [29]

    Lin, et al., Defect engineering of two-dimensional transition metal dichalcogenides

    Z. Lin, et al., Defect engineering of two-dimensional transition metal dichalcogenides. 2D Materials 3, 022002 (2016), doi:10.1088/2053-1583/3/2/022002, https://iopscience. iop.org/article/10.1088/2053-1583/3/2/022002

    2016 doi

  22. [30]

    Zhou, et al., Defect Etching of Phase-Transition-Assisted CVD-Grown 2H-MoTe2

    Y. Zhou, et al., Defect Etching of Phase-Transition-Assisted CVD-Grown 2H-MoTe2. Small 17, 2102146 (2021), doi:10.1002/smll.202102146,https://onlinelibrary.wiley.com/ doi/10.1002/smll.202102146

    2021 doi

  23. [31]

    K ¨oster, et al

    J. K ¨oster, et al. , Defect Agglomeration and Electron-Beam-Induced Local-Phase Trans- formations in Single-Layer MoTe 2. The Journal of Physical Chemistry C 125, 13601– 13609 (2021), doi:10.1021/acs.jpcc.1c02202, https://pubs.acs.org/doi/10.1021/ acs.jpcc.1c02202

    2021 doi

  24. [32]

    Wang, et al., Vacancy Defects in 2D Transition Metal Dichalcogenide Electrocatalysts: From Aggregated to Atomic Configuration

    X. Wang, et al., Vacancy Defects in 2D Transition Metal Dichalcogenide Electrocatalysts: From Aggregated to Atomic Configuration. Advanced Materials 35, 2206576 (2023), doi: 10.1002/adma.202206576, https://onlinelibrary.wiley.com/doi/10.1002/adma. 202206576

    2023 doi

  25. [34]

    O. F. N. Okello, et al., Atomistic Probing of Defect-Engineered 2H-MoTe2 Monolayers. ACS Nano 18, 6927–6935 (2024), doi:10.1021/acsnano.3c08606,https://pubs.acs.org/doi/ 10.1021/acsnano.3c08606

    2024 doi

  26. [35]

    Ghasemi, W

    A. Ghasemi, W. Gao, Atomistic mechanism of stress modulated phase transition in mono- layer MoTe2. Extreme Mechanics Letters40, 100946 (2020), doi:10.1016/j.eml.2020.100946, https://linkinghub.elsevier.com/retrieve/pii/S2352431620301905

    2020

  27. [36]

    Krishnamoorthy, et al., Semiconductor–metal structural phase transformation in MoTe 2 monolayers by electronic excitation.Nanoscale 10 (6), 2742–2747 (2018)

    A. Krishnamoorthy, et al., Semiconductor–metal structural phase transformation in MoTe 2 monolayers by electronic excitation.Nanoscale 10 (6), 2742–2747 (2018)

    2018

  28. [37]

    S. Yu, Y. Dai, B. Huang, W. Wei, Charge-Transfer-Driven Phase Transition of Two- Dimensional MoTe2 in Donor–Acceptor Heterostructures.The Journal of Physical Chemistry Letters 14, 7946–7952 (2023), doi:10.1021/acs.jpclett.3c02082, https://pubs.acs.org/ doi/10.1021/acs.jpclett.3c02082

    2023 doi

  29. [38]

    Jiang, Y.-P

    J.-W. Jiang, Y.-P. Zhou,Parameterization of Stillinger-Weber Potential for Two- Dimensional Atomic Crystals (InTech) (2017), doi:10.5772/intechopen.71929

    2017 doi

  30. [39]

    Yin, et al

    S. Yin, et al. , Atomistic simulations of dislocation mobility in refractory high- entropy alloys and the effect of chemical short-range order. Nature Communications 12, 4873 (2021), doi:10.1038/s41467-021-25134-0, https://www.nature.com/articles/ s41467-021-25134-0

    2021 doi

  31. [40]

    Mortazavi, et al

    B. Mortazavi, et al. , First-Principles Multiscale Modeling of Mechanical Properties in Graphene/Borophene Heterostructures Empowered by Machine-Learning Interatomic Po- tentials. Advanced Materials 33, 2102807 (2021), doi:10.1002/adma.202102807, https: //onlinelibrary.wiley.c...

    2021 doi

  32. [41]

    Jinnouchi, J

    R. Jinnouchi, J. Lahnsteiner, F. Karsai, G. Kresse, M. Bokdam, Phase Transitions of Hybrid Perovskites Simulated by Machine-Learning Force Fields Trained on the Fly with Bayesian In- ference. Physical Review Letters 122, 225701 (2019), doi:10.1103/PhysRevLett.122.225701, https...

    2019 doi

  33. [42]

    Y. Zhou, W. Zhang, E. Ma, V. L. Deringer, Device-scale atomistic modelling of phase-change memory materials. Nature Electronics6, 746–754 (2023), doi:10.1038/s41928-023-01030-x, https://www.nature.com/articles/s41928-023-01030-x

    2023 doi

  34. [43]

    Liu, et al., Layer-by-layer phase transformation in Ti3O5 revealed by machine-learning molecular dynamics simulations

    M. Liu, et al., Layer-by-layer phase transformation in Ti3O5 revealed by machine-learning molecular dynamics simulations. Nature Communications 15, 3079 (2024), doi:10.1038/ s41467-024-47422-1, https://www.nature.com/articles/s41467-024-47422-1

    2024

  35. [44]

    Buda, et al., Characterization of thin film materials using SCAN meta-GGA, an accurate nonempirical density functional

    I.-G. Buda, et al., Characterization of thin film materials using SCAN meta-GGA, an accurate nonempirical density functional. Scientific reports 7 (1), 44766 (2017)

    2017

  36. [45]

    I. S. Novikov, K. Gubaev, E. V. Podryabinkin, A. V. Shapeev, The MLIP package: moment tensor potentials with MPI and active learning. Machine Learning: Science and Technol- ogy 2, 025002 (2021), doi:10.1088/2632-2153/abc9fe, https://iopscience.iop.org/ article/10.1088/2632-2153/abc9fe

    2021 doi

  37. [47]

    J. Sun, A. Ruzsinszky, J. Perdew, Strongly Constrained and Appropriately Normed Semilocal Density Functional. Physical Review Letters 115, 036402 (2015), doi:10.1103/PhysRevLett. 115.036402, https://link.aps.org/doi/10.1103/PhysRevLett.115.036402

    2015 doi

  38. [48]

    V. Wang, N. Xu, J.-C. Liu, G. Tang, W.-T. Geng, V ASPKIT: A user-friendly interface facil- itating high-throughput computing and analysis using V ASP code. Computer Physics Com- munications 267, 108033 (2021), doi:10.1016/j.cpc.2021.108033, https://linkinghub. elsevier.com/ret...

    2021

  39. [49]

    A. P. Thompson, et al., LAMMPS - a flexible simulation tool for particle-based materials modeling at the atomic, meso, and continuum scales. Computer Physics Communications 271, 108171 (2022), doi:10.1016/j.cpc.2021.108171, https://linkinghub.elsevier. com/retrieve/pii/S001046...

    2022

  40. [50]

    Lehnert, et al

    T. Lehnert, et al. , Electron-Beam-Driven Structure Evolution of Single-Layer MoTe 2 for Quantum Devices. ACS Applied Nano Materials 2, 3262–3270 (2019), doi:10.1021/acsanm. 9b00616, https://pubs.acs.org/doi/10.1021/acsanm.9b00616. 32 Acknowledgments W.G. gratefully acknowledg...

    2019 doi

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.