{"id":"e864482b-7a67-4773-bd38-247e1b5712a8","arxiv_id":"1908.08221","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"STM shows an unpaired Star-of-David layer with an odd electron count is insulating, supporting Mott localisation as sufficient to gap 1T-TaS2 and linking insulator-metal transitions to interlayer stacking.","lead":"This paper uses scanning tunnelling microscopy to show that the insulating state in layered 1T-TaS2 can form even in a surface layer with an odd number of electrons per unit cell, which points to Mott localisation rather than unit-cell doubling as the driver. The result clarifies a long-running debate about this quantum spin liquid candidate and connects insulator-metal switching to interlayer stacking.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Odd-electron premise of the small-gap termination is unverified: ACAC stacking is assumed and a surface period-doubling reconstruction is not ruled out.","rationale":"The reader's verdict is CONDITIONAL and identifies the ACAC stacking plus no-reconstruction assumption as the weakest point. My stress-test agrees: this is exactly the load-bearing premise because the whole argument collapses if the small-gap surface does not have an odd number of electrons per unit cell. The paper's own text flags the assumption ('the bulk structure of the CDW should be preserved') without verifying it, and the counting of 18 large-gap versus 6 small-gap surfaces is interpreted under that model. I considered whether a more serious internal problem exists, such as the small gap being a hybridization gap between the 'unpaired' layer and the layer beneath, but the paper's own introduction notes that TC-only stacking was predicted metallic, so the unpaired TC-layer argument is at least internally consistent. I therefore do not escalate to REJECT or UNVERDICTED; rather, the existing CONDITIONAL verdict already correctly captures the need for independent structural confirmation and a surface-reconstruction check. The proposed diffraction plus Fourier-transform test would settle the concern directly and is feasible with the same crystal batch.","tokens_in":10276,"tokens_out":10821,"duration_ms":130119,"concrete_test":"On the same or identical-growth crystals, perform single-crystal X-ray or electron diffraction to determine the bulk CDW stacking sequence; then reanalyze atomically resolved STM topographs of the small-gap termination (e.g., the raw data behind SI Fig. S3) with a full 2D Fourier transform at multiple bias voltages, searching for any superlattice peak at fractional positions of the √13×√13 reciprocal lattice. If the stacking is confirmed as ACAC and no period-doubling superlattice peak is present, the odd-electron premise is supported; if the stacking is non-ACAC or an extra superlattice peak appears, the electron count per surface unit cell could be even and the Mott conclusion would not follow.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central inference is the textbook Mott criterion: a surface system with an odd number of electrons per unit cell that is nevertheless insulating cannot be explained by band theory, so Mott localization is required. The load-bearing premise is therefore the claim that the small-gap (Type 2) termination is a genuinely odd-electron layer. That premise rests on the assumed ACAC stacking of the bulk CDW and on the absence of surface reconstruction, but the paper explicitly states 'the bulk structure of the CDW should be preserved' (p. 2) as an assumption rather than a demonstrated fact, and no diffraction or bulk-sensitive measurement of the actual crystals is reported. The local phase-jump analysis at one or two step edges supports the stacking assignment only if the ACAC model is already in place. Moreover, even with ACAC stacking, the unpaired layer could undergo an in-plane reconstruction or period-doubling distortion that re-pairs the nominally unpaired SD orbital, making the electron count per surface unit cell even. The paper does not report a systematic Fourier-space search for superlattice peaks beyond the √13×√13 CDW on the small-gap surfaces, and no direct measurement of the surface electron count is provided. If either the stacking is not ACAC or a period-doubling reconstruction exists, the ~50 meV gap could be a band or CDW gap rather than a Mott gap, and 'Mott localisation alone' would not follow from the data. The secondary 'Mottness collapse' claim rests on a single observed TB-stacked metallic region, which is too isolated to independently support the mechanism.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports a low-temperature STM study of 1T-TaS2. The authors identify two distinct surface spectral types: a large-gap (~150 meV) type observed on 18 of 24 cleaved surfaces and a small-gap (~50 meV) type observed on the remaining 6. Using step-terrace topography and analysis of the in-plane phase of the √13×√13 CDW, they assign the large-gap type to termination at the TA-stacked bilayer and the small-gap type to a surface exposing an unpaired Star-of-David layer (TC stacking). They argue that this unpaired layer has an odd number of electrons per surface unit cell and, being insulating, demonstrates that Mott localization alone drives gap formation. They also observe a single metallic region which they attribute to TB stacking, interpreting it as evidence that interlayer stacking controls metal-insulator transitions.","tokens_in":10564,"tokens_out":6131,"duration_ms":55284,"significance":"The paper addresses a key open question in 1T-TaS2: whether the insulating state is a Mott insulator or a band insulator stabilized by unit-cell doubling. The experiment is well conceived: by studying different surface terminations of the same material, it attempts to isolate the layer whose electron count is odd. The step-terrace phase analysis is internally consistent, and the 18:6 distribution of the two spectral types is a large statistical sample at the level of surfaces. The electron-count argument connecting an odd number of electrons per unit cell to the necessity of interactions is textbook and is applied cleanly. If the assignments are correct, the observation of an insulating odd-electron layer would be a direct demonstration that Mott localization is sufficient to open the gap, and the metallic region attributed to TB stacking would be a striking example of stacking-controlled Mottness collapse. However, the strength of these conclusions is currently limited by the lack of independent verification of the ACAC stacking assumption and by the small number of side-by-side observations.","major_comments":[{"comment":"The central inference that the small-gap termination is an odd-electron Mott insulator depends on two unverified assumptions: (i) the bulk CDW stacking is ACAC, and (ii) the cleaved surface does not reconstruct. The paper states that 'the bulk structure of the CDW should be preserved' (p. 2), but no bulk-sensitive measurement (e.g., X-ray scattering) is presented, and no systematic Fourier-space search for superlattice peaks beyond the √13×√13 CDW is reported for the small-gap surfaces. If the actual stacking differs from ACAC, or if a surface period-doubling re-pairs the nominally unpaired SD orbital, the electron count per surface unit cell could be even, and the ~50 meV gap could be a band or CDW gap rather than a Mott gap. This would invalidate the paper's central claim.","section":"Fig. 2 and p. 2"},{"comment":"The assignment of the two spectral types to the two ACAC cleavage planes rests on a small number of step-terrace observations: one four-region terrace (Figs. 2–3) with a single partial replication (Supplementary Fig. S2), plus one buried-domain-wall case (Supplementary Fig. S1). The phase-jump analysis is internally consistent, but it presupposes the ACAC model to interpret the displacements; it does not independently establish the stacking. The authors do not report the total number of step edges examined or any instance that would be inconsistent with ACAC. With only two step-edge cases, the possibility of a fortuitous coincidence or of a different stacking sequence (e.g., ABAB or a stacking fault) cannot be excluded.","section":"Figs. 2–3 and Supplementary Figs. S1–S2"},{"comment":"The paper claims that the unpaired layer has 'an odd number of electrons per (surface) unit cell' (p. 4). This is true only if the surface unit cell equals the √13×√13 cell containing one SD and if the surface layer retains the bulk atomic structure. The STM data do not directly measure the surface electron count, and an in-plane distortion that doubles the surface unit cell (e.g., pairing of SD centers) would make the count even. No such reconstruction is ruled out; the atomic-resolution imaging in Supplementary Fig. S3 is limited to one region and does not establish the absence of long-range period-doubling. Please provide a quantitative check (e.g., analysis of STM topographs for superlattice peaks, or comparison of measured atomic positions to the bulk structure) or state this as a limitation that prevents the Mott conclusion.","section":"p. 4, 'The surface of unpaired SD clusters...'"},{"comment":"The 'Mottness collapse' claim is based on a single observation of a metallic region (Region 2) attributed to TB stacking. The paper explicitly notes that this region was 'observed nowhere else throughout measurements on twenty-three other samples' (p. 3), yet it still makes a general statement that 'the microscopic mechanism of insulator-metal transitions lies in degrees of freedom of interlayer stacking' (Abstract). A single event, with a stacking assignment inferred only from the in-plane displacement, does not support such a broad conclusion. The manuscript should either present additional instances or temper the claim.","section":"Fig. 2 and p. 4"}],"minor_comments":[{"comment":"The phrase 'twenty-three other samples' appears to be a typo; the paper reports 24 surfaces from 8 crystals, so 'other surfaces' is likely intended.","section":"p. 3"},{"comment":"The normalization of dI/dV data is described, but the resulting uncertainty in the reported gap values (~150 meV and ~50 meV) is not given. A sentence on how gap sizes vary across regions would be useful.","section":"Methods"},{"comment":"The schematic in Fig. 3e labels the regions but the caption does not describe the legend, particularly the grey layer for Region 2; please clarify the correspondence.","section":"Fig. 3e"},{"comment":"The line cuts in Supplementary Fig. S3(e) are not labeled with the direction convention; adding axis labels would improve interpretability.","section":"Supplementary Fig. S3"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is well-written and the STM data appear to be of high quality. However, the paper's title and abstract make a strong claim ('Mott localisation alone is enough') that exceeds what the presented data can establish without independent verification of the stacking pattern and surface structure. The authors may wish to consider a revised form that clearly states the conditional nature of the conclusion. The referee report is based solely on the manuscript as submitted."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's the thing: the paper reports two distinct STM terminations on cleaved 1T-TaS2 and argues that the one with a small gap (~50 meV) is an unpaired Star-of-David layer that is insulating despite having an odd electron count per surface unit cell. If that assignment holds, it settles the recent debate about whether the insulating state is a Mott insulator or a band insulator from unit-cell doubling, and it also shores up the quantum spin liquid story. That's the headline.\n\nWhat's genuinely new: the two-termination observation itself, and the direct correlation of a specific stacking configuration with the small-gap spectrum. The step-terrace phase-jump analysis is careful, and the electron-count argument is the standard textbook logic. The paper also cites and builds on the prior ab initio predictions (Ritschel, Lee) and the doublon dynamics evidence, so the citation pattern is responsible.\n\nThe soft spots are real but not fatal. The load-bearing assumption is that the bulk CDW stacking is ACAC. The paper states \"the bulk structure of the CDW should be preserved\" rather than demonstrating it with diffraction or any bulk-sensitive probe on these particular crystals. The phase-jump analysis is consistent with ACAC, but it only gets you there if you already assume the model. On top of that, the key side-by-side data come from one four-region terrace plus one partial replication in the SI; that's a thin base for a strong claim. The small-gap surface being a true insulator is inferred from dI/dV, and no raw data or error analysis is deposited, so it's hard to independently assess the gap size or check for a period-doubling reconstruction that would re-pair the nominally unpaired orbital. The 'Mottness collapse' in Region 2 rests on a single observed metallic region, which could be an extrinsic artifact, as the authors acknowledge.\n\nNone of this sinks the paper. The authors are appropriately tentative about the assignment in places, and the observation is genuinely new. But the conclusion 'Mott localisation alone is enough' should be read as 'consistent with Mott localisation under the ACAC premise,' not as a direct proof.\n\nWho would get value: anyone working on 1T-TaS2, charge density waves, Mott physics, or quantum spin liquid candidates. It deserves a serious referee. My recommendation: send it to peer review, but ask for the raw data, a diffraction measurement or at least a more systematic check of the stacking, and ideally a second independent step-edge example. With those additions, the claim would be much stronger.","headline":"A genuinely new two-termination STM observation that supports Mott localization in 1T-TaS2, but the conclusion leans on an assumed ACAC stacking and a small number of step-edge examples.","tokens_in":11085,"tokens_out":2949,"would_cite":true,"duration_ms":27736,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["71.30.+h","71.27.+a","68.37.Ef"],"model":"deepseek-v4-flash","headline":"A single unpaired layer of Star-of-David clusters in 1T-TaS2 holds an odd number of electrons per unit cell yet is insulating, proving Mott localisation alone drives the gap.","keywords":["1T-TaS2","Mott insulator","charge density wave","Star-of-David cluster","interlayer stacking","scanning tunnelling microscopy","unit-cell doubling","quantum spin liquid"],"falsifier":"Image the small-gap termination with a structure-sensitive technique: if it shows a reconstruction in which neighbouring Star-of-David clusters pair their remaining orbitals, the surface would have an even number of electrons per unit cell and the ~50 meV gap could be a band gap, not a Mott gap. Alternatively, a magnetic probe that finds no localised $S=1/2$ moments on the unpaired layer would contradict the cluster-Mott interpretation.","tokens_in":10112,"feed_emoji":"🔬","tokens_out":10497,"duration_ms":98331,"temperature":0.7,"pith_summary":"1T-TaS2 becomes insulating below about 180 K, but why has been disputed: the standard answer is Mott localisation on the half-filled orbital at the heart of each Star-of-David cluster, while a rival explanation says that the alternating ACAC interlayer stacking doubles the unit cell, making the electron count even and the gap a band gap. The paper uses a scanning tunnelling microscope to inspect both cleavage surfaces predicted by the ACAC pattern. The surface that exposes a single unpaired Star-of-David layer contains an odd number of electrons per surface unit cell, and it is still insulating, with a gap of about 50 meV. Since an odd electron count cannot be gapped by band filling alone, the gap must come from strong electron-electron interactions — Mott localisation is sufficient on its own. A metallic termination produced by an accidental restacking shows the same physics in reverse: interlayer stacking controls the collapse of Mottness, placing stack engineering at the heart of insulator-metal switching.","feed_headline":"Odd-electron surface proves 1T-TaS2 is a Mott insulator","feed_subtitle":"An odd-electron surface layer shows a gap, ruling out unit-cell doubling as the driver.","key_machinery":"The Star-of-David (SD) cluster is the working object: in the commensurate $\\sqrt{13}\\times\\sqrt{13}\\,R13.9^\\circ$ charge-density-wave state, each cluster of 13 Ta ions leaves one half-filled orbital, the candidate for Mott localisation. The argument is carried by the two cleavage planes of the ACAC stacking pattern: one plane separates intact $T_A$-stacked bilayers (even electron count per supercell, ~150 meV gap), the other cuts through a bilayer and exposes a single unpaired SD layer (odd electron count, ~50 meV gap). The authors determine which surface is which by measuring the in-plane phase jump of the CDW lattice across single-layer steps — a zero or non-zero shift identifies whether the step ends on the same bilayer or on an unpaired layer. The odd electron count of the unpaired layer is the load-bearing fact that separates the Mott mechanism from unit-cell doubling.","core_discovery":"The central claim is that Mott localisation alone is enough to drive the insulating state of 1T-TaS2, and that unit-cell doubling is not needed. From step-terrace spectroscopy combined with an analysis of the in-plane phase of the charge-density-wave lattice across steps, the authors identify two types of surface termination expected for the ACAC stacking pattern: an intact bilayer termination showing a gap of about 150 meV, and an unpaired Star-of-David layer showing a smaller gap of about 50 meV. The unpaired layer has 13 orbitals per surface unit cell, i.e. an odd number of electrons, and it is insulating; this is precisely the situation in which only Mott localisation can produce an insulating state. The authors therefore conclude that the insulating phase of bulk 1T-TaS2 is a cluster Mott insulator, and they interpret the metallic termination, which corresponds to a metastable $T_B$ stacking, as a Mottness-collapsed state in which the effective $U/t$ ratio has been reduced by the interlayer registry.","pith_inferences":["A monolayer of 1T-TaS2 is by definition an unpaired SD layer, so the data predict that exfoliated or epitaxial monolayers should be insulating with a gap near 50 meV; this is directly testable.","The difference in gap size between the bilayer (≈150 meV) and unpaired layer (≈50 meV) suggests the effective $U/t$ depends on the stacking environment, so measuring the gap as a function of film thickness or stacking vector would map how correlations respond to dimensionality.","Controlled tip-induced re-stacking of the top layer onto different stacking vectors ($T_A$, $T_B$, $T_C$) could turn the accidental metallic termination into a deliberate, reversible switch of Mottness — a candidate route for atom-scale memristive devices."],"forward_implications":["The long-standing cluster-Mott description of 1T-TaS2's low-temperature phase is restored; the insulating state can be understood without invoking unit-cell doubling.","Proposals of a quantum spin liquid state in 1T-TaS2 remain viable at surfaces and stacking faults, where an unpaired odd-electron layer can host a triangular lattice of localised $S=1/2$ moments.","Interlayer stacking becomes the microscopic control parameter for insulator-metal transitions in this material, explaining the metallic mosaic, hidden states, and voltage-switching phenomena reported in thin flakes.","The two measured gaps (~150 meV for the bilayer termination and ~50 meV for the unpaired layer) give concrete targets for cluster-Hubbard-model calculations of the effective $U$ and $t$ parameters."],"supporting_citations":[{"why":"Establishes the Star-of-David cluster motif and the 13-orbital count per cluster that yields an odd electron number per unit cell.","marker":"[3]"},{"why":"Proposes the cluster-Mott localisation picture for the insulating state, the interpretation the paper's result upholds.","marker":"[4]"},{"why":"Predicts that alternating TA/TC stacking doubles the unit cell and can produce an insulator without interactions, the rival explanation to be ruled out.","marker":"[6]"},{"why":"Argues that the insulating phase and metal-insulator transition arise from stacking-driven band effects, the alternative the paper challenges.","marker":"[7]"},{"why":"Reports nanoscale manipulation of the Mott state coupled to charge order, providing the backdrop for the metallic termination observed here.","marker":"[13]"},{"why":"Documents a metallic mosaic phase and suggests interlayer stacking as its origin, directly motivating the 'Mottness-collapse' interpretation.","marker":"[14]"},{"why":"Ab initio prediction that different inter-layer stacking vectors yield metallic versus insulating behaviour, the mechanism the paper verifies at the surface.","marker":"[19]"},{"why":"Observation of doublon excitations characteristic of a Mott state in 1T-TaS2, independent evidence favouring strong electron correlations.","marker":"[25]"}],"fun_headline_variants":["Mottness alone drives insulating state in 1T-TaS2","Unit-cell doubling ruled out for TaS2 insulator","Odd-electron layer proves Mott gap in 1T-TaS2","Mottness not stacking: key to TaS2 gap"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The conclusion depends on cleaved surfaces preserving the bulk ACAC stacking without reconstruction, so that the small-gap surface really is a single unpaired Star-of-David layer with an odd number of electrons per surface unit cell; if a surface reconstruction re-paired the remaining orbitals, the electron count would become even and a band gap could explain the spectrum.","fun_headline_variants_meta":{"raw":{"variants":["Mottness alone drives insulating state in 1T-TaS2","Unit-cell doubling ruled out for TaS2 insulator","Odd-electron layer proves Mott gap in 1T-TaS2","Mottness not stacking: key to TaS2 gap"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001075,"raw_usage":{"total_tokens":4491,"prompt_tokens":925,"completion_tokens":3566,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":541,"completion_tokens_details":{"reasoning_tokens":3494}},"tokens_in":541,"tokens_out":3566,"duration_ms":26157,"temperature":1.0,"reasoning_tokens":3494,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:45:59.020668+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Image the small-gap termination with a structure-sensitive technique: if it shows a reconstruction in which neighbouring Star-of-David clusters pair their remaining orbitals, the surface would have an even number of electrons per unit cell and the ~50 meV gap could be a band gap, not a Mott gap. Alternatively, a magnetic probe that finds no localised $S=1/2$ moments on the unpaired layer would contradict the cluster-Mott interpretation.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Argues that the insulating phase and metal-insulator transition arise from stacking-driven band effects, the alternative the paper challenges."},{"cited_title":"A., Mertelj, T","cited_arxiv_id":null,"evidence_quote":"Reports nanoscale manipulation of the Mott state coupled to charge order, providing the backdrop for the metallic termination observed here."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Ab initio prediction that different inter-layer stacking vectors yield metallic versus insulating behaviour, the mechanism the paper verifies at the surface."}],"review_version":1}