{"id":"67e00aa2-3133-49c1-a1cb-24fac7c5b018","arxiv_id":"2511.14570","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Area-selective ARPES shows interlayer Umklapp scattering and a Kondo-like peak in 4Hb-TaS2, tilting the balance from a pure Mott-Hubbard to a hybridized Kondo-lattice description.","lead":"Using a tightly focused photoemission beam, the authors mapped electrons on both exposed faces of the layered superconductor 4Hb-TaS2 and found that metallic states from one layer are folded by the charge-order pattern of the neighboring layer, coupling near the Fermi level. The result favors a picture in which a barely filled flat band and Kondo-like hybridization—not a simple Mott insulator—shape the exotic superconductivity.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Incipient flat band and Kondo peak hinge on charge-transfer estimate and unverified sub-resolution peak width.","rationale":"I agree with the reader that the weakest assumption is the charge-transfer/incipient-FB chain. This is the point where the argument is most vulnerable: the ARPES-observed windmill Fermi surfaces (Umklapp replicas) are supported by dispersion matching and simulation, but the Kondo-like peak and the 'reconciliation' of Kondo vs. Mott physics hinge on a barely filled flat band. The reader's concern is mitigated by STM observations of ~12% filled SoDs, but those topographic images do not directly locate the unoccupied flat band in energy. The alternative concern—the sub-resolution peak width—is related: if the peak is not intrinsic, the Kondo claim fails regardless of the fillings. Both concerns are testable with STS and high-resolution ARPES. The verdict should remain CONDITIONAL, pending these checks.","tokens_in":12705,"tokens_out":12882,"duration_ms":132548,"concrete_test":"Perform STS at 4.5 K on the 1T termination of 4Hb-TaS2, acquiring dI/dV spectra on empty SoD centers. If the unoccupied flat band lies within ~50 meV above EF (a peak at small positive bias), the incipient-FB scenario and the 0.92 e/SoD transfer are supported. If the flat band appears at significantly higher energy or is absent, the rigid-band estimate is wrong and the Kondo-like peak in ARPES requires another explanation.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The conclusion's second half—that the surface 1T layer hosts an incipient flat band just above EF and that the EDC peak at EF is a Kondo resonance—depends on the rigid-band charge transfer of 0.92±0.04 e/SoD (§II.B) and on a 6.5 meV HWHM peak measured with a stated energy resolution below 20 meV (Methods). The charge transfer is derived from a ~30 meV VHS shift between surface and subsurface 1H layers, converted via an unstated rigid-band model; a systematic error in the conversion factor of >10% would push the flat band filling outside the 0.08±0.04 range, eliminating the dilute-moment Kondo picture. The peak width, if taken at face value, is narrower than the resolution, so the feature may be a resolution-limited Fermi-edge artifact rather than an intrinsic Kondo resonance. The alternative explanation—that the peak is the tail of the 1H conduction band—is not excluded by the presented integrated EDCs. These issues are load-bearing because they are the direct evidence for the hybridization claim that 'reconciles the Kondo and Mott-Hubbard models.'","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports area-selective ARPES measurements on the natural van der Waals superlattice 4Hb-TaS2, addressing the long-standing question of the interlayer coupling between superconducting 1H-TaS2 and Mott-insulating 1T-TaS2 layers. The authors identify termination-dependent electronic structures: on the 1T termination they observe chiral 'windmill' Fermi surfaces around the zone center, which they assign not to intrinsic 1T-layer states but to Umklapp replicas of subsurface 1H bands folded by the sqrt(13)×sqrt(13) superstructure of the surface 1T layer. They further report a Kondo-like peak at the Fermi level, attributed to hybridization between 1H conduction states and an incipient flat band of the surface 1T layer, with a filling of ~0.08 electrons per Star-of-David cluster. The paper additionally attributes distinct 3×3 and 2×2 charge orders on the surface and subsurface 1H layers to interlayer charge transfer, leading to segmented Fermi surfaces and opposite shifts of the van Hove singularities. The conclusions aim to reconcile the Kondo and Mott-Hubbard descriptions of this material.","tokens_in":13024,"tokens_out":3270,"duration_ms":36329,"significance":"If the main claims hold, this work provides a substantial reassessment of the low-energy electronic structure of 4Hb-TaS2: the 'windmill' states are reinterpreted as Umklapp-scattered 1H states rather than coherent 1T states, and the Fermi-level spectral weight is interpreted as evidence for interlayer hybridization with an incipient flat band. The strengths include the use of area-selective ARPES to separate the two terminations, the dispersion matching with calculated 1H bands, the simulation of Umklapp folding, and the consistency with prior STM observations of SoD charge disproportionation. However, the most novel claims—the Kondo-like peak and the incipient flat-band filling—rest on energy scales comparable to or smaller than the stated energy resolution, and on a charge-transfer estimate whose systematic uncertainty is not quantified. These issues are load-bearing and require additional evidence before the paper can fully settle the debate.","major_comments":[{"comment":"The Kondo-like peak is a central claim, but the fitted half width at half maximum of 6.5 meV at 16 K is substantially smaller than the stated energy resolution of below 20 meV (Methods V.B). This means the feature could be a resolution-limited Fermi-edge artifact or the tail of the 1H conduction band rather than an intrinsic Kondo resonance. The authors should provide a quantitative line-shape analysis that accounts for the instrumental resolution, a temperature-dependence study that shows proper Kondo scaling, or a comparison with a model EDC including a Fermi-Dirac function and background. Without this, the identification of the peak as Kondo-like is not established.","section":"§II.B, Fig. 3(h), Fig. S8"},{"comment":"The interlayer charge transfer of 0.92±0.04 electrons per SoD is obtained 'based on a rigid band model' by comparing the doping levels of surface and subsurface 1H layers. The conversion from the observed VHS shifts to a charge transfer is not explicitly presented, and the quoted uncertainty does not include the 20 meV energy resolution or the systematic uncertainty of the rigid-band approximation. Because the incipient-flat-band filling of 0.08±0.04 e/SoD is the foundation of the Kondo-like interpretation, the authors must state the rigid-band model, its conversion factor, and a realistic error budget including resolution effects. A shift of order 0.1 e/SoD would move the flat band away from EF and invalidate the dilute-moment Kondo scenario.","section":"§II.B, Fig. 2"},{"comment":"The dichotomous VHS shifts on the two 1H layers are quoted as about 20 meV (surface: from EF-10 meV to EF+10 meV; subsurface: from EF-40 meV to EF-60 meV). These shifts are comparable to the stated energy resolution of 20 meV. The claim that 3×3 and 2×2 CDWs shift the VHS in opposite directions is therefore not convincingly resolved above experimental uncertainty. The authors should quantify the uncertainty on each VHS energy, possibly using fits with a resolution-convolved spectral function, and demonstrate that the measured shifts are statistically significant.","section":"§II.C, Fig. 4"}],"minor_comments":[{"comment":"The symbol for sqrt(13) appears garbled (e.g., '√𝟏𝟏𝟏𝟏 × √𝟏𝟏𝟏𝟏') in the abstract. Please ensure correct mathematical rendering throughout.","section":"Abstract and captions"},{"comment":"The letter 'c' is used without parentheses in the Fig. 1 caption ('c, Core level spectra'). Please unify the figure-reference style.","section":"§II.A, Fig. 1(b)"},{"comment":"The statement that 'additional investigations are required to exclude the possibility of intrinsic surface band splitting arising from unknown mechanisms' is an important caveat. It would be helpful to see a brief discussion of what such investigations would entail and whether any existing data bear on this.","section":"§II.A"},{"comment":"The integrated EDC on the 1H termination is described as an 'unperturbed Fermi-Dirac line profile,' but no fit is shown. Adding a fit with residual statistics would strengthen the comparison with the Kondo-like peak on the 1T termination.","section":"§II.B, Fig. 3"},{"comment":"The ab initio calculation supporting the charge-transfer estimate is only referenced in the Supplemental Material. Please summarize the computational setup and the key result in the main text or provide an explicit comparison with the rigid-band value.","section":"§II.B, Fig. S5"}],"recommendation":"major_revision","confidential_remarks":"The paper addresses a timely and contested problem, and the Umklapp-scattering interpretation of the windmill Fermi surfaces appears to be well supported by dispersion matching and simulation. The main risk is the Kondo-like peak and the incipient-flat-band filling, which rest on energy scales at or below the stated resolution and on a charge-transfer estimate with unquantified systematic error. The authors should be asked to provide a rigorous uncertainty analysis and, if possible, higher-resolution measurements or a deconvolution procedure before this can be accepted as the definitive resolution of the Kondo versus Mott debate."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Punchline: This is a strong ARPES paper that convincingly reinterprets the chiral \"windmill\" Fermi surfaces on the 1T termination as Umklapp replicas of subsurface 1H bands. The Kondo-like peak, however, is a much softer claim — the measured width sits below the stated resolution, and the incipient flat band position rests on a rigid-band charge-transfer estimate that is not fully disclosed.\n\nThe genuinely new thing here is the area-selective, termination-resolved comparison. That lets the authors show the windmill features track the dispersion of the 1H bands folded by the sqrt(13) modulation of the surface 1T layer. The dispersion matching is quite convincing, and it is backed by prior STM and by the absence of such replicas on the 1H termination. The CDW results are also well executed: the 3x3 vs 2x2 reconstruction on surface versus subsurface 1H layers, the segmented Fermi surfaces, and the opposite VHS shifts are all supported by first-principles calculations. This part of the paper is genuinely useful and advances the discussion.\n\nThe soft spot is the second half of the conclusion. The Kondo-like peak has a fitted HWHM of 6.5 meV at 16 K, but the stated energy resolution is \"below 20 meV.\" That means the feature is likely resolution-limited or at least too narrow to characterize reliably. The integrated EDC alone does not exclude the tail of the 1H conduction band as an alternative explanation. And the incipient flat band that anchors the Kondo scenario is inferred from a charge transfer of 0.92±0.04 e/SoD estimated via a rigid-band conversion of a ~30 meV VHS shift. The conversion factor is not stated, and a 10% error would push the filling well outside the claimed 0.08±0.04 range. The simulation of the Umklapp Fermi surface also lacks disclosed parameters, which makes the spectral weight modulation hard to assess. These are real concerns, but they are localized: they affect the Kondo interpretation, not the Umklapp assignment.\n\nWho should read it: anyone working on 4Hb-TaS2, Kondo/Mott physics in vdW heterostructures, or layer-resolved ARPES. The paper deserves a serious referee — it has a clear, important central claim and enough corroboration to warrant the field's attention. It needs careful revision on the Kondo peak and the charge-transfer derivation before acceptance, but it is not a desk reject.","headline":"Careful ARPES paper that convincingly reinterprets the windmill Fermi surfaces as Umklapp replicas of 1H bands, but the Kondo peak claim sits on sub-resolution data and a rigid-band charge-transfer estimate.","tokens_in":13543,"tokens_out":2767,"would_cite":true,"duration_ms":28036,"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":"In 4Hb-TaS2, the chiral “windmill” Fermi surfaces are not 1T-layer states: they are Umklapp replicas of subsurface 1H bands, and a Kondo-like peak marks their hybridization with a barely filled flat band.","keywords":["4Hb-TaS2","interlayer coupling","Umklapp scattering","flat band","Kondo lattice","charge density wave","angle-resolved photoemission","van der Waals superlattice"],"falsifier":"Cut along the √13×√13 mini-Brillouin-zone directions on the 1T termination and compare the Fermi velocities of the windmill arms with the subsurface 1H band velocities measured on the 1H termination; if the arms do not retrace those bands under the superstructure vectors, the Umklapp-replica claim fails. Separately, tunneling into the 1T layer should reveal the incipient flat band just above the Fermi level; if the flat band is found fully occupied or far from the Fermi level, the Kondo-like peak cannot arise from 1H/1T hybridization.","tokens_in":12633,"feed_emoji":"🌀","tokens_out":7742,"duration_ms":66243,"temperature":0.7,"pith_summary":"This paper argues that the so-called chiral “windmill” Fermi surfaces seen on the 1T-terminated surface of 4Hb-TaS2 do not come from the 1T layer at all. Instead, they are metallic bands from the adjacent 1H layer, folded into the Brillouin-zone center by the √13×√13 “Star-of-David” superstructure of the 1T layer via Umklapp scattering, and then hybridized with a nearly empty flat band on that 1T layer, producing a Kondo-like peak at the Fermi level. The same interlayer charge transfer leaves the 1T flat band only about 8% filled and drives different charge-density-wave orders on the two kinds of 1H layer, shifting their van Hove singularities in opposite directions. If correct, this resolves the long-standing debate between Mott-Hubbard and Kondo-lattice descriptions of 4Hb-TaS2 and points to interlayer coupling as the engine of its exotic superconductivity.","feed_headline":"Chiral windmills in 4Hb-TaS2 are folded bands from hidden 1H layers","feed_subtitle":"The windmill pockets are 1H-layer states folded by the 1T layer's Star-of-David order—settling Kondo vs. Mott.","key_machinery":"The load-bearing mechanism is Umklapp scattering: the √13×√13 “Star-of-David” superstructure of the 1T layer acts as a periodic potential that folds the subsurface 1H bands into the mini-Brillouin zone, generating the chiral windmill pattern; the same superstructure hosts an incipient flat band whose hybridization with those folded 1H bands produces the Kondo-like resonance. Area-selective ARPES separates the two terminations, and the rigid-band charge-transfer estimate (0.92±0.04 electrons per Star of David, hence roughly 8% filling of the flat band) sets the energy window for the hybridization.","core_discovery":"Using area-selective angle-resolved photoemission with a micron-sized beam spot, the authors show that on the 1T termination the “windmill” pockets around Γ match the dispersion of subsurface 1H bands repeated under the √13×√13 reciprocal vectors; the replicas’ spectral weight is enhanced near the Fermi level, with an integrated energy-distribution curve showing a Kondo-like peak (half width about 6.5 meV at 16 K) that fades with temperature. They attribute this peak to hybridization of the 1H conduction bands with the incipient, barely filled flat band of the surface 1T layer. They further show that interlayer charge transfer places that flat band just above the Fermi level (filling 0.08±0.","pith_inferences":["A quantitative test of the flat-band-filling premise: measure the 1T-layer doping by a method that does not assume rigid bands (for example, core-level shifts benchmarked to many-body calculations or momentum-resolved electron energy-loss spectroscopy); if the inferred electron count deviates by more than about 0.1 per Star of David, the Kondo-like peak would need a different explanation.","The Umklapp-replica interpretation predicts that the windmill arms’ Fermi velocities should exactly retrace the subsurface 1H bands; this can be checked with higher-momentum-resolution photoemission or scanning-tunneling quasiparticle interference, and a mismatch would rule out the mechanism.","If the filling of the 1T flat band could be tuned by gating or intercalation in thin flakes, the system should cross from a Kondo-like regime to an empty-band limit, changing the Kondo peak width and superconducting transition temperature in a predictable way—an experiment that follows directly from this picture.","The same folding logic should apply to other natural alternating-stack polytypes such as 6R-TaS2, where one layer’s charge-density-wave superstructure can act as a folding potential for the neighboring metallic layer; searching for windmill-like replicas there would test the generalizability."],"forward_implications":["Previously reported “1T-layer” Fermi-surface features, including the chiral windmill pockets, must be reinterpreted as 1H-band replicas; any model of the superconducting and vortex phases that assumes intrinsic coherent 1T states needs revision.","The Kondo-like peak at the Fermi level is direct evidence for a heavy-fermion-like coupling between itinerant 1H electrons and the incipient 1T flat band, supporting the Kondo scenario over a simple Mott-Hubbard picture.","Bulk 1T layers are over-doped Mott insulators with essentially no density of states at the Fermi level, so the superconducting condensate lives on the 1H layers; chiral p-wave pairing can emerge from the spin-polarized van Hove singularities there.","Interlayer charge transfer preselects different charge orders on surface (3×3) and subsurface (2×2) 1H layers, which segment the Fermi surfaces and move the van Hove singularities in opposite directions, one of them crossing the Fermi level and triggering a Lifshitz transition.","Natural van der Waals superlattices of transition-metal dichalcogenide layers become a tunable platform where interlayer coupling controls flat-band filling and correlated superconductivity."],"fun_headline_variants":["Direct ARPES shows 4Hb-TaS2 windmills are folded 1H bands","Kondo-like peak in 4Hb-TaS2 from flat-band hybridization","Charge order splits Fermi surfaces in 4Hb-TaS2","Hidden layers fold to form windmill states in 4Hb-TaS2","Interlayer coupling settles Kondo vs Mott in 4Hb-TaS2"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The Kondo-like peak’s attribution depends on the surface 1T layer hosting an incipient flat band just above the Fermi level, which is derived from a rigid-band charge-transfer estimate of 0.92±0.04 electrons per Star of David; if the true charge transfer is off by more than about 0.1 electrons per Star of David, the flat band would be either empty or appreciably occupied and the peak would need another explanation.","fun_headline_variants_meta":{"raw":{"variants":["Direct ARPES shows 4Hb-TaS2 windmills are folded 1H bands","Kondo-like peak in 4Hb-TaS2 from flat-band hybridization","Charge order splits Fermi surfaces in 4Hb-TaS2","Hidden layers fold to form windmill states in 4Hb-TaS2","Interlayer coupling settles Kondo vs Mott in 4Hb-TaS2"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.002477,"raw_usage":{"total_tokens":9397,"prompt_tokens":851,"completion_tokens":8546,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":595,"completion_tokens_details":{"reasoning_tokens":8443}},"tokens_in":595,"tokens_out":8546,"duration_ms":50066,"temperature":1.0,"reasoning_tokens":8443,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T21:34:02.217416+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Cut along the √13×√13 mini-Brillouin-zone directions on the 1T termination and compare the Fermi velocities of the windmill arms with the subsurface 1H band velocities measured on the 1H termination; if the arms do not retrace those bands under the superstructure vectors, the Umklapp-replica claim fails. Separately, tunneling into the 1T layer should reveal the incipient flat band just above the Fermi level; if the flat band is found fully occupied or far from the Fermi level, the Kondo-like peak cannot arise from 1H/1T hybridization.","supporting_citations":[],"review_version":1}