{"id":"170020e7-3723-461b-a9d3-28283409a30f","arxiv_id":"1909.02537","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Edge-enhanced Josephson current in multilayer WTe2 along the a-axis provides evidence for higher-order topological hinge states, while the b-axis shows uniform current.","lead":"The paper reports that supercurrents in WTe2 Josephson junctions flow preferentially along the crystal edges in one direction, which the authors interpret as evidence for topologically protected hinge states. The result suggests WTe2 is a higher-order topological insulator and introduces a transport probe for hinge states in semimetals.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Interpretation hinges on hinge-state localization in inversion-broken Td-WTe2, a property the paper itself concedes is unprotected; if inheritance from 1T'-WTe2 fails, the edge currents could be trivial.","rationale":"The reader's weakest assumption is exactly the inversion-symmetry caveat that the paper concedes in Supplementary S5. This is the most load-bearing concern because the experimental observation of edge-enhanced supercurrent, while reproducible and well-controlled, is only evidence for higher-order topology if the 1D channels are actually helical hinge states protected in the noncentrosymmetric Td phase. The theory in the main text uses a model with equivalent topology to the inversion-symmetric parent, not the actual Td structure, and the supplementary text admits there is no symmetry constraint forbidding surface masses in the inversion-broken case. If those masses are nonzero on the a-axis surfaces, the hinge states would be gapped or delocalized, and the measured edge currents would likely be trivial edge conduction. The proposed concrete test isolates exactly this assumption by computing hinge-state localization from a model that includes the broken inversion symmetry, rather than relying on inheritance from 1T'-WTe2. Since this concern does not overturn the experimental results but undermines the theoretical interpretation, the existing CONDITIONAL verdict remains appropriate.","tokens_in":15526,"tokens_out":3492,"duration_ms":40467,"concrete_test":"Recompute the hinge-state wavefunctions and Josephson current patterns using a tight-binding model that explicitly includes the Td-WTe2 inversion-broken crystal structure (for example, Wannier functions from DFT with the experimental lattice constants and two independent W atoms per primitive cell). Check whether the a-axis hinge state remains localized within ~100 nm and the b-axis state remains delocalized for realistic inversion-breaking strengths. If localization survives, the HOTI interpretation is supported; if the a-axis modes delocalize or shift, the observed edge currents cannot be taken as evidence of higher-order topology.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that the edge-enhanced Josephson current is carried by helical hinge states of a higher-order topological phase. The theory used to support this (Methods Eq. 1, Figs. 3g-j) is a model with 'equivalent topological nature' to WTe2, but the symmetry argument that fixes the hinge-state location in Supplementary S5 relies on inversion symmetry: the mass changes sign under inversion, placing domain walls at hinges. The manuscript explicitly states that Td-WTe2 breaks inversion and that 'there is no obvious constraint to forbid the mass terms even for the surface parallel to the mirror plane,' yet the authors 'still attempt to believe' localization is inherited from inversion-symmetric 1T'-WTe2. Since the Methods state that 1T'- and Td-WTe2 in the HOTI phase exhibit the same topological surface state feature and would not be distinguishable in this work, the theoretical prediction of well-localized a-axis hinges is not derived for the actual crystal. If inversion-breaking delocalizes the hinge states or moves them, the observed a-axis edge enhancement would be consistent with ordinary edge conduction (e.g., band bending or surface accumulation) rather than HOTI, and the b-axis uniform current would not rescue the interpretation. This is the load-bearing step: the announcement of 'evidence of higher order topology' rests on an unproven localization assumption that the paper itself flags as a belief.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","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.","tokens_in":15830,"tokens_out":4152,"duration_ms":48249,"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":[{"comment":"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.","section":"Supplementary S5 and Methods Eq. (1)"},{"comment":"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.","section":"Supplementary S8 and Fig. S8"},{"comment":"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.","section":"Fig. 3c and Supplementary S4"},{"comment":"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.","section":"Methods 'Effective model' and Figs. 3g–j"}],"minor_comments":[{"comment":"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.","section":"Throughout"},{"comment":"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.","section":"Supplementary S4"},{"comment":"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.","section":"Figure S5 caption"},{"comment":"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.","section":"Reference list"}],"recommendation":"major_revision","confidential_remarks":"The experimental data are strong and the control experiment is convincing, but the paper's central interpretive claim ('evidence of higher order topology') is not supported by the theory as presented. The Supplementary S5 concession about the absence of inversion symmetry in Td-WTe2 is significant enough that it must be addressed in the main text, not hidden in the supplement. The b-axis conclusion also rests on too few usable devices. I do not think this is a rejectable problem—the data may well be correct—but the framing needs to be brought in line with the evidence, and the authors should either supply a concrete calculation for the Td structure or explicitly present the result as a consistency check rather than a direct detection of HOTI hinge states."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe thing to know: this paper has a solid experimental core. The Fraunhofer-inversion method is established, and so is the HOTI prediction for WTe2, but the new observation is real: Josephson current along the a-axis is sharply enhanced at the edges, while the b-axis looks uniform. The graphite control and multiple a-axis devices make the edge enhancement credible. The measured edge currents, 22–143 nA per side, sit below the single-channel bound, which is consistent with one helical hinge state per edge.\n\nWhat is genuinely new is the application of an existing technique to a new material and the anisotropic pattern that comes out of it. That is a useful experimental result, worth having on the record.\n\nThe soft spots are in the interpretation, not the raw observation.\n\nFirst, and most importantly, the inference that these edge currents are higher-order topological hinge states relies on the hinge states being localized along the a-axis hinges of Td-WTe2. The model used in the Methods is a toy Hamiltonian with \"equivalent topological nature,\" but the actual crystal breaks inversion symmetry. Supplementary S5 explicitly says there is no obvious constraint protecting the mass domain walls, and the authors \"still attempt to believe\" the localization is inherited from the inversion-symmetric 1T' phase. That is the load-bearing step, and it is unproven. If inversion-breaking delocalizes or shifts the hinge states, the same experiment could be explained by trivial edge conduction. The paper's own words make this a conditional claim, not a settled one.\n\nSecond, the b-axis control is thinner than the main text implies. The main-text device is clean, but supplementary S8 shows the other b-axis devices did not yield usable interference patterns. So the \"no b-axis edge states\" conclusion rests largely on one device.\n\nThird, the current-density reconstruction has no error bars, and the IFT procedure involves sign-flipping lobes and a chosen penetration length L'. These are not varied, so the edge-peak widths and amplitudes are less secure than the plots suggest. This is a minor concern relative to the localization issue.\n\nThe citation pattern is fine. The prior HOTI models are used with fixed parameters, not fitted to the data, so there is no circularity problem.\n\nWho is this for? Anyone working on proximity Josephson junctions, topological semimetals, or hinge-state transport. It deserves a serious referee. I would send it to review, but the authors should either soften the topological claim or add a direct test of hinge localization—thickness dependence, local gating, or a trivial-anisotropy control. My recommendation: peer review, with conditional acceptance after the interpretation is rebalanced.","headline":"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.","tokens_in":16434,"tokens_out":2035,"would_cite":true,"duration_ms":25307,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Multilayer WTe2 conducts supercurrent through one-dimensional hinge states that are localized along the a-axis but not the b-axis.","keywords":["higher-order topological insulator","hinge states","WTe2","Josephson junction","Fraunhofer pattern","topological semimetal","proximity superconductivity","anisotropic transport"],"falsifier":"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.","tokens_in":15332,"feed_emoji":"🧲","tokens_out":10270,"duration_ms":113225,"temperature":0.7,"pith_summary":"This paper reports an experimental signature of higher-order topology in multilayer WTe2. The authors built Nb-WTe2-Nb Josephson junctions and read the spatial distribution of the supercurrent from the magnetic-field interference pattern. Along the crystal a-axis, the supercurrent is sharply enhanced at the two edges of the junction, with each edge carrying tens of nanoamps, comparable to the theoretical maximum for a single one-dimensional conducting channel. Along the b-axis the current is uniform and shows no edge enhancement. The paper concludes that multilayer Td-WTe2 is a higher-order topological insulator whose helical one-dimensional hinge states can be probed through proximity Josephson coupling.","feed_headline":"Supercurrent edge channels expose higher-order topology in WTe2","feed_subtitle":"In Nb-WTe2-Nb junctions, magnetic-field interference shows supercurrent flowing only along a-axis edges, matching hinge-state theory.","key_machinery":"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.","core_discovery":"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.","pith_inferences":["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."],"forward_implications":["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."],"supporting_citations":[{"why":"Predicts the higher-order topological phase and helical hinge states in Td-WTe2 and MoTe2, the theoretical target the experiment tests.","marker":"20"},{"why":"Provides the second-order topological insulator model for XTe2 that supplies the lattice Hamiltonian used in the numerical calculations.","marker":"34"},{"why":"Establishes the recursive Green's function method for Josephson current in inversion-symmetry-breaking topological materials, used to compute the Fraunhofer patterns.","marker":"9"},{"why":"Demonstrates hinge-state detection in another higher-order topological candidate and provides the quantitative framework for comparing single-hinge currents.","marker":"33"},{"why":"Introduces the inverse-Fourier-transform reconstruction of Josephson current density for edge-dominated transport in quantum spin Hall junctions.","marker":"30"},{"why":"Applies the same edge-mode supercurrent imaging to a two-dimensional topological insulator, supporting the technique's validity.","marker":"31"},{"why":"Reports helical hinge zero modes in an iron-based superconductor and discusses the same single-channel current bound used in this paper.","marker":"35"},{"why":"Predicts the two-dimensional topological insulator phase in monolayer WTe2, the limit from which the hinge states inherit their character.","marker":"14"}],"fun_headline_variants":["Hinge states shine through Josephson supercurrent in WTe2","Edge supercurrent maps higher-order topology in WTe2","Anisotropic hinge states exposed by Josephson interference","Supercurrent reveals helical hinges in WTe2","Higher-order topology in WTe2 seen via edge currents"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"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.","fun_headline_variants_meta":{"raw":{"variants":["Hinge states shine through Josephson supercurrent in WTe2","Edge supercurrent maps higher-order topology in WTe2","Anisotropic hinge states exposed by Josephson interference","Supercurrent reveals helical hinges in WTe2","Higher-order topology in WTe2 seen via edge currents"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000398,"raw_usage":{"total_tokens":2098,"prompt_tokens":975,"completion_tokens":1123,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":591,"completion_tokens_details":{"reasoning_tokens":1043}},"tokens_in":591,"tokens_out":1123,"duration_ms":9056,"temperature":1.0,"reasoning_tokens":1043,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T04:47:41.287726+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"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.","supporting_citations":[],"review_version":1}