REVIEW 4 major objections 4 minor 1 references
Evidence of Higher Order Topology in Multilayer WTe$_2$ from Josephson Coupling through Anisotropic Hinge States
T0 review · 4 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash
Pith's one-line read Multilayer WTe2 conducts supercurrent through one-dimensional hinge states that are localized along the a-axis but not the b-axis.
desk verdict A well-controlled Josephson-imaging experiment gives real evidence for edge-localized supercurrent in WTe2, but the leap from that to higher-order topology rests on an unproven and admittedly unprotected localization assumption. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The experimental probe is the magnetic-field dependence of the Josephson critical current, $I_c(B)$, which is the magnitude of the Fourier transform of the real-space supercurrent density $J(x)$. Reconstructing $J(x)$ by inverse Fourier transform converts the Fraunhofer pattern into a map of where supercurrent flows. The theoretical counterpart is the lattice Hamiltonian $H(k) = H_{\mathrm{MNL}}(k) + V_c + V_{so}$ (Eq. 1), built from a monopole nodal-line semimetal with a tilting mass and a spin-orbit term that gaps the nodal line, which produces the helical hinge states. The quantitative check is the ballistic short-junction bound $e I_{J,h} R_{N,h} = \pi \Delta$, with $R_{N,h} = h/e^2$ and $\Delta$ the BCS gap of the Nb electrode, giving the maximum supercurrent a single hinge channel can carry; the measured edge currents are comparable to and below that bound.
What would settle it
Measure the two-terminal normal-state resistance of a single a-axis edge contacted by normal leads: a single helical hinge should give a resistance near $h/e^2$, the value the paper assumes for one channel, whereas trivial edge conduction through many disordered channels would give a much lower resistance. Alternatively, separate the Weyl points by strain so the higher-order topological phase is theoretically absent; if the edge-enhanced supercurrent survives in that regime, the edge channels are not protected hinge states.
Extended reading notes
Core claim
The central claim is that the Josephson critical current of a Nb-WTe2-Nb junction, measured as a function of perpendicular magnetic field, encodes the real-space location of the conducting channels inside the WTe2 flake. In devices with transport along the a-axis, the inverse Fourier transform of the Fraunhofer pattern shows two sharp current peaks at the edges, sitting on a uniform bulk background; the edge contributions range from 22 to 143 nA across devices, at or below the single-channel bound $e I_{J,h} R_{N,h} = \pi \Delta_{Nb} \approx 140$ nA for Nb electrodes. Along the b-axis the pattern is a standard single-slit Fraunhofer shape and the reconstructed current density is flat. The same model Hamiltonian that produces higher-order topology in WTe2 gives helical hinge states localized at a-axis hinges and delocalized along b-axis hinges, matching the measured anisotropy. The paper concludes that these observations provide evidence for the higher-order topological phase of WTe2 and its anisotropic topological hinge states.
Load-bearing premise
The central assumption is that the one-dimensional edge channels seen in the junctions are true topological hinge states, which requires them to stay localized at the hinges of the real Td-WTe2 crystal even though that crystal lacks inversion symmetry, the symmetry that would otherwise protect their positions; the paper's supplementary discussion concedes that nothing in the structure forces this localization and that it is assumed to be inherited from the centrosymmetric 1T' phase.
Editorial extensions
If this is right
- Multilayer WTe2 becomes a transport-accessible higher-order topological insulator: its hinge channels carry a measurable supercurrent even though the bulk is a semimetal.
- The Fraunhofer-to-current inversion turns a standard Josephson measurement into a real-space imaging tool for boundary modes, applicable to other candidate hinge-state materials without scanning probes.
- The a-axis versus b-axis anisotropy is a concrete design rule: devices seeking hinge transport should be aligned so current flows along the a-axis, where hinge states stay localized.
- Because the hinge channels are helical and proximity-coupled to superconductors, these junctions are natural platforms for searching for Majorana bound states at the ends of hinge wires.
Reading between the lines
- An untested consequence is that the helical character of the edge channels should produce spin-polarized supercurrent, which could be looked for as a magnetic-field-dependent asymmetry in the critical current; the paper does not report such a measurement.
- The reconstructed edge-channel widths, roughly 100 to 600 nm, are much larger than the atomic-scale hinge width expected for a clean higher-order topological insulator, suggesting the channels may be broadened by disorder or by the proximity region under the superconducting electrodes.
- The same inverse-Fourier technique could be applied to devices with tunable junction length, testing whether the edge channels persist as the junction is shortened toward the ballistic limit.
- If the edge channels are truly single helical channels, their normal-state two-terminal resistance should be close to $h/e^2$; measuring this directly would distinguish topological hinge states from trivial metallic edge conduction.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper reports Nb-WTe2-Nb proximity Josephson junctions fabricated on multilayer Td-WTe2 flakes and uses magnetic-field interference patterns (Fraunhofer patterns) to spatially resolve the supercurrent distribution. Along the a-axis, the reconstructed current density J(x) exhibits sharp enhancements at both edges of the junction, with integrated edge currents of 8.8–143 nA per edge, comparable to the theoretical upper bound for a single 1D hinge channel. Along the b-axis, the current density is uniform and the interference pattern is single-slit-like. A graphite control junction shows no edge enhancement, and additional a-axis and b-axis devices show consistent behavior. The authors interpret the anisotropic edge-enhanced supercurrent as evidence of helical 1D hinge states of a higher-order topological insulator (HOTI) phase in WTe2, supported by a model Hamiltonian calculation that reproduces the qualitative features of the interference patterns.
Significance. If the interpretation holds, this work would provide a rare transport signature of higher-order topology in a type-II Weyl semimetal and demonstrate a useful technique—Josephson interferometry—for detecting hinge states in semimetallic systems where bulk conduction obscures conventional edge transport. The experimental strengths are real: the graphite control excludes a generic fabrication artifact, the multiple a-axis devices give reproducibility, and the comparison of the integrated edge current to the single-channel bound eIJ,hRN,h = πΔ is a quantitative, theory-independent check. The main weakness is that the identification of the edge currents with topologically protected hinge states rests on a theoretical model whose symmetry assumptions are not satisfied by the actual Td-WTe2 structure, a concession the manuscript itself makes in Supplementary S5.
major comments (4)
- [Supplementary S5 and Methods Eq. (1)] The central claim that the observed a-axis edge supercurrent is carried by topologically protected hinge states is not established for Td-WTe2. The model Hamiltonian in Methods Eq. (1) is introduced as a toy model with 'equivalent topological nature' to WTe2, but the bulk-boundary correspondence argument in S5 that places hinge states on a-axis hinges relies on inversion symmetry (the mass changes sign under inversion). Td-WTe2 breaks inversion, and S5 explicitly concedes: 'there is no obvious constraint to forbid the mass terms even for the surface parallel to the mirror plane' and that the authors 'still attempt to believe' localization is inherited from inversion-symmetric 1T'-WTe2. This is a load-bearing step: without a symmetry constraint or an explicit calculation for the inversion-broken Td structure, the edge-enhanced supercurrent could equally arise from trivial edge conduction (e.g., band bending or surface accumulation). Since the abstract claims 'evidence of the higher order topological phase,' this gap between the data and the interpretation must be addressed, either by providing a calculation that demonstrates hinge-state localization in the actual Td-WTe2 structure or by softening the claim to consistency with a HOTI model and explicitly discussing the trivial-edge alternative.
- [Supplementary S8 and Fig. S8] The b-axis anisotropy conclusion is weakly supported. The text states that for Dev. B2–B4 'we cannot observe non-vanishing or ordinary interference pattern' because the critical current is much smaller than in Dev. A, so the uniform current-density conclusion for the b-axis rests effectively on a single clean device, Dev. B (Fig. 3e–f). The other b-axis devices do not provide independent confirmation of the single-slit pattern. Given that the anisotropic hinge-state picture is central to the paper, the b-axis null result needs either more usable devices or a more guarded statement: the data show that a-axis junctions can carry edge-enhanced supercurrent while clean b-axis junctions do not, but the absence of edge enhancement along b is not as firmly established as the presence along a.
- [Fig. 3c and Supplementary S4] The extracted current density J(x) has no error bars or uncertainty quantification. The inverse Fourier transform procedure requires assumptions about the current-phase relation (sinusoidal) and about the odd/even decomposition of the Josephson current, and the finite magnetic-field range limits the real-space resolution. Without an estimate of how these assumptions and the experimental noise affect J(x), the visual difference between the a-axis edge peaks and the bulk baseline is not quantitatively demonstrated. I request that the authors estimate the resolution of the IFT, propagate the uncertainty in the measured Ic(B), and confirm that the edge enhancements are statistically significant relative to the uniform-background hypothesis.
- [Methods 'Effective model' and Figs. 3g–j] The theory comparison is qualitative and uses hand-picked model parameters (m1 = -3t, m2 = 0.3t, m3 = 0.2t, etc.). The model is not fitted to the data, and the model Hamiltonian is a generic HOTI model rather than a first-principles description of Td-WTe2. The abstract's phrase 'in good agreement with theoretical calculations' therefore overstates the level of support. I recommend either quantifying the agreement (e.g., comparing the lobe decay, peak positions, and edge-peak widths between theory and experiment) or explicitly describing the comparison as qualitative support for the localization pattern, not a quantitative validation of the Hamiltonian.
minor comments (4)
- [Throughout] There are several typos and inconsistencies: 'Plank's constant' should be 'Planck's constant'; in Supplementary S1, 'WTe2 was observes large magnetoresistance' should be 'WTe2 was observed to show large magnetoresistance'; 'Polari sations' in the Methods has an errant space; the device labels Dev. A/Dev. B and Dev. A1/Dev. B1 are used interchangeably and should be defined consistently.
- [Supplementary S4] The IFT description should state explicitly that the reconstruction is band-limited by the maximum applied field, which sets the real-space resolution, and that the assumed sinusoidal current-phase relation may not hold for highly transmissive contacts; a short sentence on the expected resolution and its effect on the extracted edge widths would be helpful.
- [Figure S5 caption] The caption does not indicate which crystal axes correspond to the plotted surfaces; adding axis labels would make the comparison with the main-text schematics (Figs. 3a and 3d) clearer.
- [Reference list] Reference 20 is cited as a preprint; if the published version (Phys. Rev. B 100, 125101 (2019)) is now available, it should be cited instead of or in addition to the arXiv preprint.
Circularity Check
No significant circularity: the experiment independently tests a prior, fixed-parameter HOTI model; the limited S5 localization assumption is a stated caveat, not a circular reduction.
full rationale
The paper's central inference—that a-axis edge-enhanced Josephson current is carried by helical hinge states—is not forced by construction. The theoretical Hamiltonian (Methods Eq. 1) is adopted from refs 20 and 34 with explicit fixed parameters, not fitted to the measured Ic(B); the calculated hinge-state wave functions and Fraunhofer patterns are forward-model predictions compared qualitatively with Dev. A and Dev. B. The Josephson current bound IJ,h = pi*Delta_Nb*R_N,h/e uses the standard short ballistic junction formula with R_N,h = h/e^2 and the BCS gap of Nb, so no fitted quantity is renamed as a prediction. The inverse Fourier transform reconstruction of J(x) is a mathematical inversion of the measured interference pattern, and the graphite control rules out an artefact. The only notable caveat is Supplementary S5, which concedes that Td-WTe2 breaks inversion symmetry and that 'there is no obvious constraint to forbid the mass terms even for the surface parallel to the mirror plane,' with the authors stating they 'still attempt to believe' localization is inherited from 1T'-WTe2. That is an honest limitation on model applicability, but it does not make the derivation circular: the quoted assumption is not used to define the measured edge currents or to set the model parameters. Self-citations (e.g., ref. 9 for the recursive Green's function method) are methodological and not load-bearing for the HOTI interpretation. On this evidence the circularity score is 0.
Assumptions & free parameters
free parameters (3)
- Flux penetration length L' =
170 to 210 nm across devices
- HOTI model parameters (m1, m2, m3, va, vb, vc, lambda_b, lambda_c, gamma_1, gamma_z, beta_a) =
t=1; m1=-3t; m2=0.3t; m3=0.2t; va=2t; vb=1.6t; vc=t; lambda_b=0.1t; lambda_c=t; gamma_1=0.4t; gamma_z=-0.4t; beta_a=1.5t
- Gaussian widths of edge current peaks =
about 100 nm for Dev. A; 144 to 575 nm for other devices
assumptions (5)
- domain assumption Td-WTe2 is expected to realize a higher-order topological phase with helical hinge states (Wang et al., ref 20; Ezawa, ref 34).
- domain assumption The simplified Hamiltonian H(k) (Eq. 1) has the same topological nature as multilayer WTe2.
- domain assumption Ic(B) equals the magnitude of a Fourier transform of a 1D Josephson current density J(x), assuming a sinusoidal current-phase relation.
- domain assumption Andreev reflection at the superconducting interface is suppressed for bulk states and enhanced for boundary states in Weyl semimetals.
- domain assumption The 18 to 38 layer WTe2 flakes are in the bulk limit, not the monolayer QSHI regime.
Cite this review
Pith. "Pith review of Evidence of Higher Order Topology in Multilayer WTe$_2$ from Josephson Coupling through Anisotropic Hinge States." pith.science (2026). https://pith.science/paper/H2BIKUDH
@misc{pith2026190902537,
author = {Pith},
title = {Pith review of: Evidence of Higher Order Topology in Multilayer WTe$_2$ from Josephson Coupling through Anisotropic Hinge States},
year = {2026},
howpublished = {\url{https://pith.science/paper/H2BIKUDH}},
note = {Machine review of arXiv:1909.02537}
}
abstract
The noncentrosymmetric Td-WTe$_2$, previously known as a type-II Weyl semimetal, is expected to have higher order topological phases with topologically protected, helical one-dimensional (1D) hinge states when their scarcely separated Weyl points get annihilated. However, the detection of these hinge states is difficult in the presence of the semimetallic behaviour of the bulk. Here, we spatially resolved the hinge states by analysing the magnetic field interference of supercurrent in Nb-WTe$_2$-Nb proximity Josephson junctions. The Josephson current along the a-axis of the WTe$_2$ crystal, but not along the b-axis, showed sharp enhancements at the edges of the junction; the amount of enhanced Josephson current was comparable to the upper limits of a single 1D conduction channel. Our experimental observations provide evidence of the higher order topological phase in WTe$_2$ and its corresponding anisotropic topological hinge states, in good agreement with theoretical calculations. Our work paves the way for hinge transport studies on topological semimetals in superconducting heterostructures, including their topological superconductivity.
Reference graph
Works this paper leans on
-
[1]
1 Ali, M. N. et al. Large, non-saturating magnetoresistance in WTe 2. Nature 514, 205- 208 (2014). 2 Woods, J. M. et al. Suppression of Magnetoresistance in Thin WTe2 Flakes by Surface Oxidation. ACS Appl. Mater. Interfaces 9, 23175-23180 (2017). 3 Zhu, Z. et al. Quantum Oscillations, Thermoelectric Coefficients, and the Fermi Surface of Semimetallic WTe2...
work page Pith review arXiv 2014
Reviewed August 14, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.