{"id":"5ed6f1d4-dd14-49bf-b906-b1c3a36b2078","arxiv_id":"2411.18205","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Bulk 1T-TaSe2 in its CDW phase is a moderately correlated metal, and its insulating domains and low-energy peaks stem from CDW stacking faults and quantum well states rather than from strong correlations.","lead":"Micro-focused photoemission on 1T-TaSe2 shows the bulk is a moderately correlated metal with insulating patches, and explains sharp low-energy peaks as quantum well states caused by stacking faults in the charge density wave. This offers a simpler, stacking-based account of phenomena previously blamed on strong correlations, such as a Kondo resonance.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Insulating-domain assignment to AA stacking faults is not uniquely established; a Mott-insulating surface layer could also explain the gap, so a quantitative comparison to a Mott surface calculation is needed.","rationale":"The paper makes a strong case that bulk 1T-TaSe2 is a moderately correlated metal: transport and specific-heat data, the DMFT Fermi surface and quasiparticle weight, and the kz-resolved comparison all support this. The QWS interpretation is also well supported by both the phase-accumulation model and the continued-fraction slab DMFT, which reproduce the number, energy, and flat dispersion of the observed states. The least secure element is the microscopic origin of the insulating domains. The authors infer AA stacking faults from the qualitative similarity between the insulating-domain ARPES and the AA-stacked slab calculation, but they do not directly measure the local stacking on the same regions, and they concede in the conclusion that the reason the surface layer is predominantly insulating remains an open question. Because the monolayer of the same compound is Mott insulating in their own DMFT, and because a Mott surface layer could in principle produce a gap of similar magnitude and valence-band dispersion, the data do not uniquely select the band-insulator stacking-fault scenario. This is the single most load-bearing assumption: if the insulating domains are Mott-like, the abstract's statement that correlation effects play a secondary role fails. The proposed computational test is concrete and feasible with existing methods, and it would clarify whether the ARPES data can distinguish the two mechanisms. The reader's verdict of CONDITIONAL remains appropriate; my read does not change it.","tokens_in":17799,"tokens_out":10279,"duration_ms":108795,"concrete_test":"Using the authors' existing DFT+DMFT framework, compute the surface spectral function of a semi-infinite AC-stacked slab in which the top layer is treated as a Mott insulator (e.g., with the monolayer U value or a U chosen to reproduce the monolayer gap), and quantitatively compare its gap size, valence-band dispersion, and number of backfolded bands to the ARPES insulating-domain spectra. If this Mott-surface calculation matches the data as well as or better than the AA-stacked slab of Fig. 2(f), the stacking-fault interpretation loses its uniqueness. As an independent check, co-register micro-ARPES and STM on the same regions to directly identify the local stacking order in metallic versus insulating domains.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that stacking faults, not correlations, dictate the insulating domains rests on identifying the insulating regions with AA-stacked CDW layers. The evidence is a qualitative match between ARPES and DFT+DMFT spectra of an AA-stacked slab (Fig. 2(f)) plus literature on TaS2. However, the same experimental data may also be compatible with a Mott-insulating surface layer: the monolayer of 1T-TaSe2 is Mott insulating in the authors' own DMFT at the same U, and the insulating-domain ARPES (gap of ~0.2 eV, valence bands between -0.4 and -0.2 eV) is reminiscent of spectra previously interpreted as a lower Hubbard band. The paper does not compare the insulating-domain data to a calculation of a Mott surface layer on a metallic bulk, so the uniqueness of the stacking-fault assignment is not established. If the insulating domains were Mott-like, correlation effects would be primary, directly contradicting the abstract. The QWS observations are consistent with the stacking scenario but do not by themselves identify the nature of the confining barrier.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents spatially resolved ARPES and laser ARPES measurements on bulk 1T-TaSe2 in the low-temperature CDW phase, showing coexisting insulating and metallic domains, a chiral Fermi surface, and series of low-energy sharp states. The authors combine DFT and DFT+DMFT slab calculations to argue that the insulating domains arise from AA-stacked CDW stacking faults forming band insulators, that the metallic regions are a moderately correlated Fermi liquid, and that the sharp low-energy states are quantum well states confined between an insulating fault and the vacuum. They conclude that the stacking arrangement and quantum size effects, rather than electron correlations, dominate the observed electronic phenomena.","tokens_in":17965,"tokens_out":4973,"duration_ms":45177,"significance":"If the main claims hold, this work resolves a long-standing inconsistency between transport, STM, and earlier ARPES studies of 1T-TaSe2 by identifying the spatially averaged 'Mott insulating surface' as a superposition of metallic and stacking-fault-insulating domains. The micro-focus ARPES data are of high quality, the bulk DMFT comparison with a cRPA-derived U is a reasonable first-principles approach, and the observation of quantum well states with near-flat dispersion is novel and intriguing. However, the central attribution of the insulating domains to AA stacking rather than to a Mott-insulating surface layer is not uniquely established, and several quantitative comparisons involve manually adjusted parameters. These issues need to be resolved before the paper can be accepted as a definitive resolution of the controversy.","major_comments":[{"comment":"The assignment of the insulating domains to AA-stacked CDW layers is not uniquely established. The DFT+DMFT AA-stacked slab calculation reproduces the insulating-domain ARPES data only qualitatively and underestimates the gap ('Albeit with smaller gap size...'), while the authors' own monolayer DMFT at the same U produces a Mott insulator. The experimental insulating-domain spectrum (gap of about 0.2 eV, valence bands between -0.4 and -0.2 eV) is also reminiscent of the lower Hubbard band previously interpreted as a Mott surface feature. To support the claim that stacking faults, not correlations, dictate the insulating domains, the authors should either perform a quantitative DMFT calculation of a Mott-insulating surface layer coupled to a metallic bulk and compare it directly with Fig. 2(b,d), or provide a direct spatial correlation between the ARPES domain pattern and the local stacking geometry (e.g., via STM/LEED on the same sample regions).","section":"Main text, 'Metallic and insulating band structures' and Fig. 2"},{"comment":"The phase accumulation model for the quantum well states uses a phase shift Phi = 0.9 chosen to reproduce a hypothetical monolayer calculation and quantum well thicknesses L matched to the observed energy spacing between QWSs. Because these parameters are calibrated to the data, the quantitative agreement between the model and the measured QWS energies is partly built in. The independent support from the DMFT slab calculations (Fig. 4d,e) mitigates this concern, but the manuscript should explicitly state that Phi and L are fitted quantities and demonstrate the sensitivity of the QWS assignment to variations in these parameters (e.g., a structure plot with realistic error bars rather than a single best fit).","section":"Supplemental Material, Section IX, Eq. (S1)"},{"comment":"The DMFT Fermi-surface comparison uses a doping of 0.8 electrons per SOD chosen to optimize agreement with the ARPES data. Although the measured Luttinger volume (0.84 +/- 0.2 electrons/CDW unit cell) is consistent with this value, the doping is still a free parameter. The authors should clarify how the uncertainty in the experimental filling propagates into the DMFT Fermi-surface comparison and whether the agreement with the measured Fermi surface persists at the upper or lower bounds of the Luttinger volume.","section":"Main text, Fig. 3 and Supplemental Fig. S3"}],"minor_comments":[{"comment":"The sentence 'Fig. 4(a) shows a cluster map of the metallic area displayed in Fig. 1(d)' should refer to Fig. 1(e) (the ARPES spatial maps), not Fig. 1(d) (the LEED pattern).","section":"Main text, after Fig. 3"},{"comment":"The phrase 'rotated by 13 .9' contains a spurious space; also, the space group 'P ¯1' is rendered inconsistently and should be typeset as P-1 (or with a proper overline).","section":"Main text, first paragraph"},{"comment":"The abbreviation CDW is used in the abstract but defined only in the main text; consider defining it at first use in the abstract.","section":"Abstract"},{"comment":"The sentence 'The dashed unit cell indicates the translation of the CDW between consecutive layers' would be clearer with a direct pointer to the relevant dashed cell in the figure.","section":"Fig. 1 caption"},{"comment":"The sentence 'It remains an open question why the surface layer is predominantly insulating' is an important caveat, but it appears only at the end of the Letter; given that the uniqueness of the stacking-fault mechanism is a central issue, this limitation should be discussed more explicitly in the main text where the insulating-domain assignment is first introduced.","section":"Conclusion"}],"recommendation":"major_revision","confidential_remarks":"The micro-ARPES data and the QWS observation are strong assets, and the overall scenario is plausible. However, the central claim that stacking faults, not correlations, dictate the insulating domains requires either a quantitative exclusion of the Mott-surface alternative or a direct structural measurement correlating stacking with the ARPES domain pattern. The manuscript is likely suitable for publication after these load-bearing points are addressed."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe paper is worth reading. It settles a long-running puzzle: bulk 1T-TaSe2 in the CDW phase is a moderately correlated metal, and the earlier Mott-surface ARPES results came from spatial averaging over a mostly insulating surface with small metallic regions. The micro-spot ARPES directly images both kinds of domains, gives a clean chiral Fermi surface, and the DMFT bulk comparison is convincing. That alone is a real step forward.\n\nThe new piece I like best is the identification of the low-energy peak series as quantum well states. The phase accumulation model is simple, and although it uses a fitted phase shift and well thickness, the authors check it against DMFT slab calculations and the flat dispersion follows naturally from the out-of-plane band structure. That is a concrete, falsifiable picture.\n\nThe soft spots are real but not fatal. The assignment of the insulating domains to AA stacking faults is inferred from DFT+DMFT slabs and prior TaS2 literature; there is no direct structural measurement on the same regions probed by ARPES. The stress-test asks whether a Mott-insulating surface layer on a metallic bulk could also match the insulating-domain spectra. The paper does not compare against that alternative calculation, so the uniqueness of the stacking-fault story is not proven. I think the stacking-fault picture is more likely right—the slabs produce a gap without invoking correlations, and the same U gives a Mott insulator in the monolayer—but more likely is not established. The doping of 0.8 e/SOD in DMFT is a lesser concern; it is a reasonable vacancy proxy, but it is chosen to match the Fermi surface.\n\nWho gets value: anyone working on TMDs, CDW stacking effects, or spatial-resolved ARPES methodology. It is a strong experimental paper with a reasonable but not airtight theoretical framework.\n\nI would send it to referees. They should ask for a direct comparison of the insulating-domain ARPES to a Mott surface calculation, and ideally local structural correlation (STM or nc-AFM) on the same regions. With that, the paper would be much stronger.","headline":"Micro-ARPES finally shows the metallic bulk of 1T-TaSe2 and ties insulating patches to CDW stacking faults, with a nice quantum-well-state story; the stacking-fault assignment is not yet uniquely nailed against a Mott-surface alternative.","tokens_in":18622,"tokens_out":2371,"would_cite":true,"duration_ms":22477,"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":"Bulk 1T-TaSe2 is a moderately correlated metal, and the insulating domains and flat low-energy bands seen on its surface come from charge-density-wave stacking faults and quantum confinement, not from Mott physics.","keywords":["1T-TaSe2","charge density wave","ARPES","stacking faults","quantum well states","dynamical mean field theory","Mott insulator vs band insulator","correlated metal"],"falsifier":"Measure the stacking registry of the same domains with a local structural probe, for example cross-correlating STM topography with micro-ARPES on one terrace: an insulating region showing AC stacking, or a metallic region showing AA stacking, would break the central claim. Alternatively, showing that the low-energy peak series has a dispersion that no particle-in-a-box quantization with any well thickness can reproduce would falsify the quantum-well assignment.","tokens_in":17565,"feed_emoji":"🔬","tokens_out":5758,"duration_ms":50076,"temperature":0.7,"pith_summary":"This paper tries to settle what bulk 1T-TaSe2 actually is in its low-temperature charge-density-wave phase. Using micrometer-resolved photoemission, the authors find that the surface is a patchwork: most areas look insulating, but the metallic regions show a chiral Fermi surface and quasiparticle bands with moderate correlation strength. They argue that the insulating patches are not a Mott state but band insulators caused by occasional AA stacking of the star-of-David CDW layers, and that the sharp low-energy peak series in metallic regions are quantum well states confined between such an insulating stacking fault and the vacuum. If correct, this reconciles transport, STM, and previous ARPES results and makes correlation physics a secondary player in the bulk, while leaving stacking order and quantum size effects as the controlling degrees of freedom.","feed_headline":"Stacking faults, not Mottness, explain 1T-TaSe2's insulating patches","feed_subtitle":"Micro-ARPES finds a correlated metal; 'insulating' patches and flat bands trace to CDW stacking defects and quantum wells.","key_machinery":"The load-bearing machinery is the CDW stacking arrangement of the layered crystal. In the bulk, layers stack with AC registry, where the central Ta atom of one star-of-David (SOD) cluster aligns with an outer Ta atom of the next, yielding a half-filled $d_{z^2}$ band that DMFT finds in a Fermi-liquid regime with $Z\\sim0.66$. A stacking fault that puts two SOD units directly on top of each other (AA registry) hybridizes these orbitals into a band insulator. The authors reproduce the insulating surface and the quantum-well peak series with DFT+DMFT slab calculations, extended to effectively 100 layers by a continued-fraction embedding, and they match the energies with a phase-accumulation quantization condition $2k_z(E)L+\\Phi=2\\pi n$.","core_discovery":"The central claim is that 1T-TaSe2 in the CDW phase is a moderately correlated metal, with a DMFT quasiparticle weight $Z \\sim 0.66$, and that the insulating domains observed by micro-ARPES are band insulators caused by stacking faults in the CDW, specifically AA stacking in which star-of-David units sit directly on top of each other. The low-energy series of sharp, almost flat states near the Fermi level are quantum well states: conduction electrons confined between an insulating stacking-fault layer and the vacuum, with energies, number, and dispersion set by the thickness of the metallic slab. The near-flat dispersion follows from quantizing the out-of-plane bulk band that crosses the Fermi level, so the flatness is a band-structure effect rather than a signature of strong correlations. Earlier reports of a Mott gap or charge-transfer gap are attributed to spatially averaging over insulating and metallic domains.","pith_inferences":["If stacking faults are the controlling variable, then exfoliated few-layer flakes with deliberately engineered stacking should show tunable quantum-well spectra, turning sample thickness into a control knob for the effective correlation strength.","The same stacking-fault picture may explain apparent Mott gaps in 1T-TaS2 and other van der Waals CDW compounds, where spatially averaged probes could similarly mix band-insulating and metallic domains.","A quantitative prediction that goes beyond the paper: the phase shift $\\Phi$ in the quantization condition should depend on the barrier height and thus on the stacking-fault depth; measuring QWS energies versus well thickness on many domains could extract this dependence and test the toy model.","The paper's 'moderately correlated metal' conclusion implies that transport should show Fermi-liquid signatures such as $T^2$ resistivity below the CDW transition, providing an independent experimental check."],"forward_implications":["Bulk 1T-TaSe2 is a Fermi-liquid-like metal with moderate renormalization, so earlier Mott-insulator interpretations based on spatially averaged photoemission need to be revisited.","Insulating domains are band insulators produced by CDW stacking faults, not by electron correlation alone.","The sharp low-energy states are quantum well states whose energies and near-flat dispersion are controlled by the thickness of the metallic slab between a stacking-fault barrier and the vacuum.","The chiral Fermi surface observed in metallic domains matches bulk AC-stacked CDW layers with $k_z$ integration, tying the surface signal to the bulk electronic structure.","Controlling sample thickness, for example by exfoliation, should provide a direct way to engineer the conduction-band width and hence the effective correlation strength."],"supporting_citations":[{"why":"Provides the transport evidence that bulk 1T-TaSe2 stays metallic to low temperature, which the paper's metallic-band picture reconciles with ARPES.","marker":"[18]"},{"why":"A recent ARPES study reporting metallic behavior that the present work extends with spatial resolution.","marker":"[20]"},{"why":"The prior interpretation of a broad photoemission peak as the lower Hubbard band of a Mott insulator, which the paper reinterprets as a spatial average over domains.","marker":"[21]"},{"why":"The charge-transfer insulator claim for bulk 1T-TaSe2 that the metallic spectra and DMFT results directly counter.","marker":"[22]"},{"why":"STM observation of multiple correlated-looking states and a zero-bias peak that the paper explains as quantum well states rather than Kondo resonances.","marker":"[23]"},{"why":"STM evidence that both AC and AA stacking configurations occur at the surface of 1T-TaSe2, supporting the stacking-fault mechanism.","marker":"[24]"},{"why":"First-principles work on the CDW instability and stacking energetics that underpins the small energy difference between stacking configurations.","marker":"[33]"},{"why":"Calculations showing that different CDW stacking orders in 1T-TaSe2 produce multiple quantum states, providing prior theoretical support for stacking-controlled electronic behavior.","marker":"[34]"},{"why":"The continued-fraction embedding method used to extend DMFT slab calculations toward a semi-infinite bulk.","marker":"[39]"}],"fun_headline_variants":["Stacking faults, not electron correlations, create 1T-TaSe2's insulating gaps","1T-TaSe2's flat bands are quantum well states, not signs of strong correlation","CDW stacking faults create 1T-TaSe2's metallic and insulating domains","Mottness out, stacking faults in: 1T-TaSe2's heterogeneity explained"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument assumes that the specific micrometer-scale regions where ARPES sees an insulating gap are the same regions where the CDW layers are AA-stacked, even though no experiment in this paper images the stacking order of the exact domains probed by photoemission.","fun_headline_variants_meta":{"raw":{"variants":["Stacking faults, not electron correlations, create 1T-TaSe2's insulating gaps","1T-TaSe2's flat bands are quantum well states, not signs of strong correlation","CDW stacking faults create 1T-TaSe2's metallic and insulating domains","Mottness out, stacking faults in: 1T-TaSe2's heterogeneity explained"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001068,"raw_usage":{"total_tokens":4448,"prompt_tokens":890,"completion_tokens":3558,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":506,"completion_tokens_details":{"reasoning_tokens":3461}},"tokens_in":506,"tokens_out":3558,"duration_ms":23206,"temperature":1.0,"reasoning_tokens":3461,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T11:24:25.569091+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the stacking registry of the same domains with a local structural probe, for example cross-correlating STM topography with micro-ARPES on one terrace: an insulating region showing AC stacking, or a metallic region showing AA stacking, would break the central claim. Alternatively, showing that the low-energy peak series has a dispersion that no particle-in-a-box quantization with any well thickness can reproduce would falsify the quantum-well assignment.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"A recent ARPES study reporting metallic behavior that the present work extends with spatial resolution."},{"cited_title":"Perfetti, A","cited_arxiv_id":null,"evidence_quote":"The prior interpretation of a broad photoemission peak as the lower Hubbard band of a Mott insulator, which the paper reinterprets as a spatial average over domains."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The charge-transfer insulator claim for bulk 1T-TaSe2 that the metallic spectra and DMFT results directly counter."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"STM observation of multiple correlated-looking states and a zero-bias peak that the paper explains as quantum well states rather than Kondo resonances."},{"cited_title":"Zhang, Z","cited_arxiv_id":null,"evidence_quote":"STM evidence that both AC and AA stacking configurations occur at the surface of 1T-TaSe2, supporting the stacking-fault mechanism."},{"cited_title":"Ge and A","cited_arxiv_id":null,"evidence_quote":"First-principles work on the CDW instability and stacking energetics that underpins the small energy difference between stacking configurations."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Calculations showing that different CDW stacking orders in 1T-TaSe2 produce multiple quantum states, providing prior theoretical support for stacking-controlled electronic behavior."},{"cited_title":"Petocchi, C","cited_arxiv_id":null,"evidence_quote":"The continued-fraction embedding method used to extend DMFT slab calculations toward a semi-infinite bulk."}],"review_version":1}