Pith. sign in

REVIEW 4 major objections 4 minor 6 references

Melting point depression of charge density wave in 1T-TiSe$_2$ due to size effects

T0 review · 4 major / 4 minor · reviewed 2026-08-04 · deepseek-v4-flash

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

Pith's one-line read Charge density wave melting in 1T-TiSe2 depends on flake size: below an effective radius of ~100 nm the melting point drops, and flakes approaching the 10–50 nm correlation length fail to order.

desk verdict A plausible and novel observation of lateral-size CDW melting depression in TiSe2, with the quantitative conclusion resting on diffraction measurements that may be biased by flake bending/tilting. read the letter →

arxiv 2509.16730 v1 pith:2OPII3AQ submitted 2025-09-20 cond-mat.mtrl-sci

Saif Siddique , Mehrdad T. Kiani , Omri Lesser , Stephen D. Funni , Nishkarsh Agarwal , Maya Gates , Miti Shah , William Millsaps
show 8 more authors
Suk Hyun Sung Noah Schnitzer Lopa Bhatt David A. Muller Robert Hovden Ismail El Baggari Eun-Ah Kim Judy J. Cha
This is my paper
classification cond-mat.mtrl-sci PACS 71.45.Lr
keywords chargedensitywave1T-TiSe2meltingpointdepressionfinite-sizescalingGinzburg-Landaumodelcorrelationlengthcryo-TEMnanoflake
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

The paper reports that the charge density wave (CDW) in 1T-TiSe2 nanoflakes loses stability as the flake's lateral size shrinks: flakes with effective radius below ~100 nm melt at temperatures well below the bulk 210–230 K transition, and the smallest flakes show no CDW order at all at 20 K. By tracking 2x2 superlattice peaks in nanobeam electron diffraction during in-situ cryo-TEM warming experiments, the authors map the melting temperature as a function of flake area. The data are fitted with a Ginzburg-Landau model in which the CDW order parameter vanishes at the flake edge, yielding a zero-temperature correlation length of 10–50 nm, consistent with the reported CDW domain size and with the measured spacing of Ti self-intercalant clusters that act as nucleation sites. The authors conclude that a correlated electronic phase transition follows classical nucleation theory, with finite size cutting off the divergence of the correlation length and setting a minimum size for CDW stability.

What carries the argument

The carrying mechanism is a scalar Ginzburg-Landau order parameter for the commensurate CDW, with a free energy containing quadratic, quartic, and gradient terms. Enforcing the order parameter to vanish at the flake boundary and minimising yields the closed-form melting-point relation T_melt(R) = T_c (1 − (ξ0/R)^2), in which the zero-temperature correlation length ξ0 sets the size threshold. The paper also uses atomic-resolution STEM to identify Ti self-intercalants in the van der Waals gap as the microscopic nucleation sites, with measured spacings of tens of nanometres consistent with ξ0. That length-scale agreement connects the macroscopic Ginzburg-Landau fit to a concrete defect structur

What would settle it

Measure the CDW transition temperature in a set of lithographically patterned circular flakes of 1T-TiSe2 with radii between 20 and 150 nm, using the same nanobeam diffraction protocol with sample-calibrated temperatures. A melting-temperature trend that deviates from T_c (1 − (ξ0/R)^2), or CDW order appearing in sub-50 nm flakes containing no Ti intercalants, would falsify the claimed finite-size scaling and defect-starvation mechanism.

Watch

Extended reading notes

Core claim

In 1T-TiSe2 nanoflakes, the CDW melting temperature falls as the flake shrinks below an effective radius of ~100 nm, and the smallest flakes show no CDW order at 20 K. The size dependence follows the Ginzburg-Landau relation T_melt(R) = T_c (1 − (ξ0/R)^2), derived by requiring the order parameter to vanish at the boundary. Fitting 19 flakes yields ξ0 = 10–50 nm (10–20 nm for near-circular flakes), matching reported CDW domain sizes and the 28–58 nm spacings between Ti self-intercalant clusters seen by atomic-resolution STEM. The suppression of CDW in the smallest flakes is attributed to defect starvation: below the mean intercluster spacing, some flakes contain no nucleation sites.

Load-bearing premise

The Ginzburg-Landau analysis assumes that every irregularly shaped flake behaves like a circular disk of the same area, with the CDW order parameter forced to zero at the boundary; if flakes pin or enhance the order at their edges, or if shape matters more than area, the fitted correlation length and the predicted suppression size become inaccurate.

Editorial extensions

If this is right

  • CDW-based devices made from 1T-TiSe2 need a lateral size above roughly 100 nm radius to retain bulk-like switching temperatures; sub-50 nm devices may not support CDW order at all.
  • The extracted zero-temperature correlation length of 10–50 nm provides a quantitative intrinsic length scale for CDW order in 1T-TiSe2, consistent with domain sizes seen in bulk.
  • Controlling the density and spacing of Ti self-intercalants should allow deliberate tuning of CDW nucleation and stability at the nanoscale.
  • The same Ginzburg-Landau finite-size scaling should apply to other layered CDW compounds, giving a direct experimental route to their correlation lengths.
  • Flakes with no intercalant cluster in the interior are predicted to stay CDW-free even at 20 K, a testable consequence of the defect-starvation argument.

Reading between the lines

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

  • If the boundary condition is relaxed to allow partial pinning at the oxidized edge, the fitted ξ0 could shift; testing clean, lithographically defined circular flakes would isolate this effect.
  • A direct real-space imaging of the order-parameter profile near the flake edge (by STM or cryogenic STEM) would test the assumed vanishing boundary condition.
  • If defect starvation is the mechanism, deliberately seeding sub-50 nm flakes with intercalants at controlled positions should restore CDW order.
  • The reported ~15 K offset between holder and sample temperature is a constant shift; sample-calibrated temperatures would strengthen quantitative comparison with the Ginzburg-Landau prediction.
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 / 4 minor

Summary. The paper reports an in-situ cryo-TEM study of 1T-TiSe2 nanoflakes with effective radii up to 250 nm. Using nanobeam electron diffraction, the authors track the 2×2 CDW superlattice peaks while heating from 20 K or 100 K, and find that the CDW melting temperature decreases with decreasing flake size below an effective radius of about 100 nm and is completely suppressed for the smallest flakes. The melting data are fit to the Ginzburg-Landau expression T_melt(R) = Tc(1 − (ξ0/R)^2), yielding ξ0 = 10–50 nm, and this length scale is compared with measured spacings of Ti self-intercalant clusters (28–58 nm). The paper concludes that CDW stability in 1T-TiSe2 exhibits finite-size scaling and that the CDW transition follows classical nucleation theory.

Significance. If the central observation is correct, this would be an important demonstration that a 2D electronic phase transition obeys finite-size scaling, with a quantitative link between the thermodynamic melting point, the zero-temperature correlation length, and the density of intrinsic nucleation sites. The paper has several concrete strengths: it uses direct diffraction evidence from 19 nanoflakes, spans a wide size range, applies a physically motivated Ginzburg-Landau boundary-value model, and provides independent atomic-resolution measurements of intercalant spacings. The extracted correlation-length range is also compared with reported CDW domain sizes. However, the quantitative conclusions rest on an operational definition of T_melt that is vulnerable to sample-tilt artifacts, and the model fit uses the same data to set ξ0 and then to validate the suppression boundary. These issues require additional control experiments or analyses before the claim can be fully accepted.

major comments (4)
  1. [Results, Fig. 2 and 'CDW melting temperature vs size' (Fig. 3)] The melting temperature is defined by the disappearance of 2×2 superlattice peaks in single nanobeam diffraction patterns, using a fixed peak-to-background threshold of 5. The authors themselves note that 'the intensity changes in Bragg peaks in Fig. 2 are due to sample motions, such as bending and tilting, due to the thermal drifts of the stage during temperature increase.' The same motions will suppress superlattice peak intensity irrespective of CDW melting. Since smaller, less rigid flakes are more likely to bend or tilt during thermal cycling, the observed size trend could be at least partly an artifact of orientation loss rather than genuine melting. The manuscript does not describe zone-axis re-tracking at each temperature, rocking-curve integration, or use of a size-normalized intensity metric. This concern is load-bearing because it threatens the primary observation, not just th
  2. [Ginzburg-Landau model, Eq. (1) and Fig. 3] The GL expression T_melt(R) = Tc(1 − (ξ0/R)^2) is derived for a circular system with vanishing order parameter at the boundary, but the experimental 'radius' is obtained from the projected area of irregular flakes, as stated in the text ('using the area to get the equivalent circular radius'). The authors acknowledge the shape issue and color-code by circularity, but they do not correct for it quantitatively. The fit range ξ0 = 10–50 nm (or 10–20 nm for near-circular flakes) is therefore not a precise, assumption-free measurement. Moreover, the claim that the absence of CDW in small flakes is 'as predicted by the model' is circular insofar as ξ0 is fit to the same melting data that are used to draw the suppression boundary. Please provide a sensitivity analysis of ξ0 under different shape treatments and boundary conditions, and separate the fit of ξ0 from the independent test of the supp
  3. [Results, 'If the TiSe2 flakes undergo the irreversible irradiation damage...'] The manuscript excludes flakes that show superlattice peaks at room temperature because they cannot be distinguished from irradiation-induced superstructure. This exclusion is reasonable, but it creates a potential selection bias: if small flakes are more or less likely to exhibit room-temperature superlattice peaks, the reported size dependence could be skewed. The paper does not report how many flakes were excluded, their sizes, or whether the exclusion rate is size-dependent. Please quantify the excluded fraction and verify that the melting-temperature trend is robust to this selection.
  4. [Results, final paragraph on temperature calibration] The liquid helium holder has a measured sample-temperature offset of about +15 K relative to the controller reading, and the liquid nitrogen holder is accurate to 10–20 K. Since the small flakes are measured predominantly with the helium holder, a systematic offset of this magnitude could shift the absolute T_melt values for small flakes. The authors state that this does not change the central observation, but the data in Fig. 3 are presented without applying or subtracting this offset. Please provide the holder used for each flake and show the melting curves after applying the calibration, or otherwise demonstrate that the size trend is unaffected by the +15 K offset.
minor comments (4)
  1. [Fig. 4 caption] The title 'Ti self-intercalants in TiSe' should read 'TiSe2'.
  2. [Fig. 3 and 'circularity metric'] The definition of circularity as the ratio of smallest to largest Feret diameters is nonstandard; Feret diameter is usually the maximum caliper dimension. Please clarify the exact metric and its numerical range.
  3. [Abstract and Introduction] The abstract invokes 'classical nucleation theory,' but the model used is equilibrium Ginzburg-Landau theory. The connection between the two is not developed. Please either revise the wording or briefly explain how the GL finite-size depression relates to nucleation theory.
  4. [Overall] Consider providing a table with all 19 flakes: effective radius, circularity, measured melting-range bounds, holder used, and whether CDW was observed. This would increase transparency and allow independent re-analysis.

Circularity Check

1 steps flagged · score 4.0 of 10

Partial circularity: the small-flake CDW suppression is a restatement of the fitted G-L curve, though the raw size trend and intercalant-spacing measurement provide independent support.

  1. fitted input called prediction [Results, Ginzburg-Landau fit (Fig. 3) and Abstract; Eq. T_melt(R)=Tc(1-(ξ0/R)^2)]
    "Next, we tune ξ0 to capture the spread in CDW melting temperatures in TiSe2 flakes and find that ξ0 falls within 10 nm and 50 nm. ... As the nanoflake dimensions approach ξ0, the G-L theory predicts CDW will be unstable with the transition temperature essentially approaching zero. ... Indeed, we observed that some nanoflakes with small surface areas did not transition into a CDW phase at the base temperature of ~ 100 K ... two of the tested flakes did not show CDW ordering at the base temperature of 20 K."

    The claim that CDW is suppressed for flakes with R < ξ0 is not an independent first-principles prediction: it follows directly from the fitted formula T_melt(R)=Tc(1-(ξ0/R)^2) by setting T_melt=0. Since ξ0 was tuned to reproduce the observed size-dependent melting temperatures in the same Figure 3 dataset, the 'prediction' of suppression near ξ0 is a restatement of the fitted curve extrapolated to zero temperature. The absence observations are independent data points, so the confirmation is real, but the wording 'as predicted by the model' overstates the independence: the model's threshold is calibrated on the same phenomenon it is used to predict.

full rationale

The central experimental observation—size-dependent CDW melting-point depression—is a direct, model-free measurement based on nanobeam diffraction and a fixed peak/background threshold. The Ginzburg-Landau fit provides a quantitative correlation length ξ0 = 10–50 nm, and the intercalant-spacing measurements in Fig. 4 give an independent, non-fitted length scale of tens of nanometers that corroborates ξ0. The main circularity is in the presentation of the small-flake CDW suppression as 'predicted by the model': the suppression threshold R < ξ0 is obtained by inserting the fitted ξ0 into the same G-L relation used to fit the melting data, so it is a curve extrapolation rather than a separate prediction. However, the absence of CDW in the two smallest flakes was not used to define ξ0, and the intercalant-spacing comparison provides external grounding. No load-bearing self-citations, imported uniqueness theorems, or ansatz-smuggling via citation were found. Overall, the paper contains one partially circular 'prediction' but the core size-dependent trend and the independent microscopic length-scale measurement give the central claim substantial content beyond the fit.

Assumptions & free parameters 2 free parameters · 6 assumptions · 0 invented entities

The central claim rests on a standard phenomenological model with one fitted parameter (xi0), plus several domain assumptions about material behavior, boundary conditions, and temperature calibration. No new entities are introduced. The paper's quantitative conclusion (correlation length 10-50 nm) is therefore only as strong as the GL boundary condition and the single-length-scale representation of irregular flakes.

free parameters (2)
  • xi0 (zero-temperature correlation length) = 10-50 nm (10-20 nm for circular flakes)
    Tuned to capture the spread in CDW melting temperature vs radius data; not independently measured in this paper.
  • CDW melting criterion (peak intensity / background ratio) = 5
    Chosen threshold to define melting from diffraction data; no sensitivity analysis, affects reported Tmelt values.
assumptions (6)
  • domain assumption Ginzburg-Landau theory with a scalar order parameter describes the commensurate CDW transition in 1T-TiSe2.
    Invoked in Results and Supplementary Note 2; phenomenological, not derived from microscopic theory.
  • ad hoc to paper The CDW order parameter vanishes at the flake boundary.
    Assumed to model finite-size effects; no experimental verification of the boundary condition.
  • domain assumption CDW transition temperature in TiSe2 is independent of thickness down to monolayer.
    Used to argue that thickness variations across the 19 flakes do not significantly affect Tmelt; cited from ref 17.
  • domain assumption The 2x2 superlattice peaks observed by nanobeam diffraction correspond to CDW order, and flakes showing such peaks at room temperature must be excluded.
    Relies on prior identification of CDW signatures and on the ambiguity with irradiation-induced superstructure; exclusion criterion could bias the data.
  • domain assumption Ti self-intercalant clusters act as CDW nucleation sites and their in-plane spacing sets the relevant length scale.
    Based on prior STM work (ref 16) and the authors' HAADF-STEM images; used to connect xi0 to intercalant spacing.
  • domain assumption The sample temperature is close to the holder sensor reading within 10-20 K, except for a known ~15 K offset for the helium holder.
    Calibrated on large flakes only; systematic offsets affect absolute Tmelt values but are claimed not to change the central trend.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Melting point depression of charge density wave in 1T-TiSe$_2$ due to size effects." pith.science (2026). https://pith.science/paper/2OPII3AQ

@misc{pith2026250916730,
  author       = {Pith},
  title        = {Pith review of: Melting point depression of charge density wave in 1T-TiSe$_2$ due to size effects},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2OPII3AQ}},
  note         = {Machine review of arXiv:2509.16730}
}
abstract

Classical nucleation theory predicts size-dependent nucleation and melting due to surface and confinement effects at the nanoscale. In correlated electronic states, observation of size-dependent nucleation and melting is rarely reported, likely due to the extremely small length scales necessary to observe such effects for electronic states. Here, using 1T-TiSe$_2$ nanoflakes as a prototypical two-dimensional (2D) charge density wave (CDW) system, we perform in-situ cryogenic electron microscopy with temperature down to 20 K and observe size-dependent nucleation and melting of CDWs. Specifically, we observe a melting point depression of CDW for 1T-TiSe$_2$ flakes with lateral sizes less than 100 nm. By fitting experimental data to a Ginzburg-Landau model, we estimate a zero-temperature correlation length of 10--50 nm, which matches the reported CDW domain size for 1T-TiSe$_2$. As the flake size approaches the correlation length, the divergence of the CDW correlation length near the transition is cut off by the finite flake size, limiting long-range order and thereby lowering the transition temperature. For very small flakes whose size is close to the correlation length, we also observe absence of CDWs, as predicted by the model. We thus show that an electronic phase transition follows classical nucleation theory.

Figures

Figures reproduced from arXiv: 2509.16730 by the authors.

Figure 4
Figure 4. Ti self-intercalants in TiSe. a. HAADF-STEM image of 1T-TiSe2 in the [100] projection. White dotted circles show intensity in the vdW gap from the intercalants. The atomic columns get distorted and appear blurred in the presence of intercalants. b. Intensity line profiles of atomic columns indicated by (i) and (j) in panel (a). Line profile (i) shows a peak (red arrow) in the vdW gap indicating an intercalant. In co… view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

6 extracted references · 4 canonical work pages

  1. [3]

    Finally, we note that all the temperature readings reported in this work are taken from the sensors on TEM holders, and do not necessarily represent the actual sample temperature. Testing on larger flakes (effective radii > 200 nm), which should have CDW melting temperatures comparable to bulk, shows that the temperature controller for the liquid nitrogen...

    2005

  2. [850]

    (8) Chen, H.; Wang, N.; Liu, H.; Wang, H.; Ravichandran, J

    https://doi.org/10.1038/nnano.2016.108. (8) Chen, H.; Wang, N.; Liu, H.; Wang, H.; Ravichandran, J. Charge-Density-Wave Resistive Switching and Voltage Oscillations in Ternary Chalcogenide BaTiS3. Adv. Electron. Mater. 2023, 9 (11), 2300461. https://doi.org/10.1002/aelm.202300461. (9) Vaskivskyi, I.; Mihailovic, I. A.; Brazovskii, S.; Gospodaric, J.; Mert...

    work page doi:10.1038/nnano.2016.108 2016

  3. [2018]

    (6) Rossnagel, K

    https://doi.org/10.1201/9780429501012. (6) Rossnagel, K. On the Origin of Charge-Density Waves in Select Layered Transition-Metal Dichalcogenides. J. Phys. Condens. Matter 2011, 23 (21), 213001. https://doi.org/10.1088/0953-8984/23/21/213001. (7) Liu, G.; Debnath, B.; Pope, T. R.; Salguero, T. T.; Lake, R. K.; Balandin, A. A. A Charge- Density-Wave Oscill...

    work page doi:10.1201/9780429501012 2011

  4. [2024]

    Ultra-Cold Cryogenic TEM with Liquid Helium and High Stability

    https://doi.org/10.48550/arXiv.2402.00636. (31) Kobayashi, K.; Yasuda, H. Formation of a Superstructure in 1T-TiSe2 Induced at Room Temperature by Electron Beam Irradiation. Mater. Res. Express 2018, 5 (8), 085006. https://doi.org/10.1088/2053-1591/aad222. (32) Guo, X.; Kogar, A.; Henke, J.; Flicker, F.; Juan, F. de; Sun, S. X.-L.; Khayr, I.; Peng, Y.; Le...

    work page Pith review arXiv doi:10.48550/arxiv.2402.00636 2018

  5. [2025]

    https://doi.org/10.48550/arXiv.2501.09968

    work page doi:10.48550/arxiv.2501.09968

  6. [2796]

    (21) Borodin, D

    https://doi.org/10.1038/s41467-018-05153-0. (21) Borodin, D. V.; Zaitsev-Zotov, S. V.; Nad’, Y. Coherence of a Charge Density Wave and Phase Slip in Small Samples of a Quasi-One- Dimensional Conductor TaS3. J. Exp. Theor. Phys. 1987, 1394–1409. (22) Huang, Z.; Song, X.; Chen, Y.; Yang, H.; Yuan, P.; Ma, H.; Qiao, J.; Zhang, Y.; Sun, J.; Zhang, T.; Huang, ...

    doi:10.1038/s41467-018-05153-0 1987

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.