{"id":"35d0407a-32cb-469b-b3b6-940d96ed118c","arxiv_id":"1909.00074","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"The direct STS measurement shows CrBr3 has a small electronic energy gap, around 0.57 eV peak-to-peak or 0.29 eV onset, instead of the optically quoted 1.68-2.1 eV.","lead":"Scanning tunneling spectroscopy on exfoliated CrBr3 flakes reports an electronic energy gap near 0.57 eV (peak-to-peak) or 0.29 eV (signal onset), far below the 1.68 to 2.1 eV range assumed from optical data. The paper argues this small gap resolves a long-standing controversy about this 2D magnetic material.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Apparent 0.29/0.57 eV gap may be a junction/transport artifact; set-point and thickness controls are missing.","rationale":"The reader's weakest assumption identifies the same load-bearing concern: the near-Fermi-level dI/dV suppression may reflect tip-induced band bending, low sample conductivity, or substrate effects rather than the intrinsic DOS gap. The paper's strongest evidence for intrinsicness is the spatial reproducibility and the matching of higher-energy peak pairs to optical transitions, but neither validates the zero-bias region: the optical transitions connect higher-energy peak pairs, while the small-gap claim rests entirely on peaks 1 and a and the conductance suppression between them. The DFT+U calculation with U=5, J=3 is a fitted parameter set, not an independent probe. The manuscript's own admission that poor conductivity and no degassing cause noisy measurements is an internal flag that junction/transport effects have not been ruled out. A set-point current sweep at fixed bias would directly test whether the apparent gap is intrinsic or set by the junction. Since the reader already recommended a conditional verdict and my concern does not move that verdict, no change is needed.","tokens_in":10936,"tokens_out":6025,"duration_ms":62466,"concrete_test":"On the same thick flake, acquire dI/dV at a fixed bias range (e.g., ±1 V) with set-point currents spanning at least an order of magnitude (e.g., 0.1 nA, 1 nA, and 2 nA at 1 V) while keeping all other settings fixed. If the onset gap (0.29 eV) and peak-to-peak gap (0.57 eV) remain constant within ±0.05 eV, the features are likely intrinsic; if they shift or wash out, the apparent gap is a tunneling/transport artifact.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that the dI/dV suppression between peak 1 (+0.25 ± 0.04 eV) and peak a (−0.32 ± 0.02 eV) in Fig. 2(c) is the intrinsic band gap of bulk CrBr3. The paper's controls rule out localized defects (spatially reproducible spectra, atomic-resolution images) but not junction-wide effects. The manuscript itself notes the measurements are noisy because of the poor conductivity of CrBr3 and the absence of degassing, an admission that sample conductivity is a concern. On a poor conductor on HOPG, the bias-dependent series resistance of the flake and/or tip-induced band bending can suppress dI/dV near zero bias even if the DOS gap differs from the reported 0.29/0.57 eV. No experiment varies tip–sample separation at fixed bias, compares flakes of different thickness (the monolayer spectrum is explicitly unusable for gap extraction, Fig. 4(c)), or measures temperature dependence across TC to separate a true DOS gap from a transport/charging artifact. The 50 meV lock-in modulation and Feenstra normalization also complicate the definition of the 0.29 eV onset. The qualitative conclusion of a small gap may be correct, but 'unambiguous determination' is not yet supported; the claim should remain conditional pending a junction-artifact control.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports scanning tunneling microscopy and spectroscopy (STM/S) on exfoliated CrBr3 flakes on HOPG at 77 K, together with LDA+U calculations. The authors identify nine reproducible peaks in the dI/dV spectra and interpret the two closest to the Fermi level (peak 1 at +0.25 eV and peak a at -0.31 eV) as conduction and valence band edges, implying a peak-to-peak gap of 0.57 ± 0.04 eV and an onset gap of 0.29 ± 0.05 eV. This is claimed to settle a long-standing controversy about the bulk energy gap of CrBr3, which was previously believed to lie between 1.68 and 2.1 eV. The authors support their assignment by matching the remaining peak pairs to published optical transition energies and by a DFT+U calculation using U = 5 eV and J = 3 eV.","tokens_in":11158,"tokens_out":7153,"duration_ms":59257,"significance":"If the interpretation is correct, the result is significant because it revises the fundamental gap of CrBr3 downward by roughly 1 eV, with implications for the interpretation of optical and transport measurements and for the choice of DFT parameters for this family of 2D magnets. The strengths of the paper are that the STS data are spatially reproducible, measured on exfoliated crystals with atomic-resolution imaging, and the manuscript explicitly reports key limitations, including the absence of degassing and the poor conductivity of the flakes. The comparison with reported optical transitions is a useful consistency check. However, the central numerical claim is currently ambiguous (two different gap values are both called the energy gap), and the evidence does not yet eliminate junction/transport artifacts, so the paper overstates the certainty of an 'unambiguous determination.'","major_comments":[{"comment":"The manuscript reports two different quantities as 'the energy gap' of CrBr3: the peak-to-peak separation of peaks 1 and a (0.57 ± 0.04 eV) and the onset of the dI/dV signal near zero bias (0.29 ± 0.05 eV). These two numbers are not interchangeable: the peak-to-peak separation in STS is a measure of the energetic distance between two maxima in the LDOS and is generally not equal to the band gap, whereas the onset is the physically relevant gap. The abstract and conclusion emphasize the 0.57 eV value, but the body text gives no rigorous extraction criterion for the onset gap, and the Feenstra normalization with ΔV values of 0.15–0.9 V can shift the apparent onset. This ambiguity is load-bearing because the paper's central claim is the unambiguous determination of a small gap.","section":"Results and Discussions, Fig. 2(c) and Conclusions"},{"comment":"The inference that the suppression of dI/dV between peaks 1 and a is the intrinsic band gap is not controlled against junction/transport artifacts. The manuscript itself notes the measurements are noisy because of the poor conductivity of CrBr3 and the absence of degassing (p. 6-7), and the monolayer spectrum is explicitly unusable for gap extraction because of the HOPG contribution (Fig. 4(c)). On such a sample, tip-induced band bending, a bias-dependent series resistance of the flake, or the underlying HOPG layer can suppress dI/dV near zero bias even if the true DOS gap is different. No measurement at constant height with varied tip-sample separation, no flake-thickness series, and no temperature dependence across TC = 32 K are provided to rule out these effects. The paper's claim of an 'unambiguous' small gap therefore goes beyond the current evidence.","section":"Results and Discussions, Fig. 2(c)"},{"comment":"The LDA+U calculation is presented as confirming the small gap, but the parameters U = 5 eV and J = 3 eV are selected specifically because they reproduce the measured spectrum ('A combination of U = 5 and J = 3 was chosen as it fits the best to the experimental results'). This is circular and cannot serve as independent support for the measured gap. At most, it demonstrates that a plausible parameter set yields a similar gap magnitude; the DFT results should be presented as a model whose parameters are set by the STS data, not as a confirmation.","section":"Results and Discussions, Fig. 3(a)"},{"comment":"The optical-transition matching is post-hoc and combinatorially permissive. With nine measured peaks, there are numerous possible conduction-valence band pairs, and the table selects one pair for each reported transition without a statistical assessment of how likely a match is by chance. For example, the 1.35 eV PL feature is assigned to pair 2-a with a separation of 1.27 ± 0.05 eV, which deviates by 80 meV from the reported PL energy; this discrepancy is not discussed. The matching therefore does not provide strong independent evidence that peaks 1 and a are the band edges.","section":"Results and Discussions, Table 2"},{"comment":"The manuscript repeatedly relies on the supplementary figures and table (Fig. S2(d), Fig. S3(b), Fig. S3(c), Table S1) for the raw dI/dV spectra, the Gaussian peak fitting, and the complete peak-pair list. None of these were included with the submitted manuscript, so the quantitative peak positions and uncertainties, and hence the central gap determination, cannot be independently verified. The authors should submit the supplementary material for review or present the essential data in the main text.","section":"Experiments and Methods; Fig. S2, Fig. S3, Table S1"}],"minor_comments":[{"comment":"Abstract: 'pyropytic graphite' should be 'pyrolytic graphite'; in the Methods section, 'PMDS' should be 'PDMS'.","section":"Abstract and Methods"},{"comment":"The header 'Valance band' should be 'Valence band'.","section":"Table 1"},{"comment":"Reference [48] is incomplete: 'Phys. Rev. B 50, (1994)' lacks the article/page information.","section":"References"},{"comment":"The onset gap of 0.29 ± 0.05 eV is reported without a description of the onset criterion (e.g., threshold crossing, linear extrapolation); this should be specified in the figure caption or Methods.","section":"Fig. 2(c) and Results"},{"comment":"The statement that 'the DOS is not sensitive to the magnetic phase transition' is not directly tested because all dI/dV spectra were taken at 77 K, well above TC = 32 K; this should be worded as an inference from the optical comparison.","section":"Conclusions"}],"recommendation":"major_revision","confidential_remarks":"The paper reports a striking result that will be of wide interest, but the current version overstates the certainty of the gap determination. I recommend that the editor require the supplementary material for review and consider whether the 'unambiguous' language should be tempered unless the junction-artifact controls are added or clearly acknowledged as beyond the present scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: the experimental data is a genuine new data point, but the exact gap is not established, and the paper overclaims by calling it unambiguous. The qualitative conclusion that the gap is much smaller than the old optical values is plausible and probably right.\n\nWhat is actually new: direct STS on exfoliated bulk flakes, with reproducible spectra and atomic-resolution images, and a peak structure that lines up with all the old optical transition energies. That matching is a solid consistency check. I also appreciate the candid admission that the monolayer spectrum can't be used because of substrate contribution. The edge degradation observations are a minor extra.\n\nWhere it gets soft:\n\n- They quote two different gaps: 0.57 eV peak-to-peak and 0.29 eV onset, and elsewhere call both 'the energy gap.' Peak-to-peak separation of DOS peaks is not a band gap. The onset value is more physical but harder to pin down given the 50 meV modulation and the Feenstra normalization with a chosen broadening parameter. This needs to be resolved.\n- The stress-test concern about junction artifacts is fair. CrBr3 is a poor conductor, they didn't degas, and they have no thickness, tip-separation, or temperature-dependent control. Any of those could suppress dI/dV near zero bias. This doesn't kill the small-gap conclusion, but it does kill the word 'unambiguously.'\n- The DFT+U confirmation is circular: U=5 and J=3 were chosen to fit the measured gap. They acknowledge DFT is mean-field, but then present it as support. It's really a consistency check at best.\n- Their argument that peaks 1 and a are intrinsic because spectra are reproducible and the atomic lattice is defect-free is reasonable but not airtight; subsurface defects or tip effects wouldn't show up in atomic resolution.\n\nWho this is for: anyone working on CrBr3 or 2D magnets; the data needs to be in the conversation. It deserves a serious referee, but the authors should be pushed to define the gap, add at least one junction-artifact control or publicly walk back the 'unambiguous' framing, and reframe the DFT as illustrative.\n\nRecommendation: send to peer review, conditional on revision.","headline":"A plausible direct STS measurement of a small gap in CrBr3 that overclaims certainty; the qualitative finding is likely right but the exact value and the 'unambiguous' framing need more controls.","tokens_in":11800,"tokens_out":2563,"would_cite":true,"duration_ms":24442,"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":"CrBr3's real band gap is 0.57 eV, not the 1.68–2.1 eV value assumed for decades.","keywords":["CrBr3","scanning tunneling spectroscopy","band gap","van der Waals magnet","dI/dV","DFT+U","2D ferromagnetism","layered semiconductor"],"falsifier":"Measure STS on the same material under deliberately varied junction conditions and on an insulating substrate: if the zero-bias suppression and the 0.57 eV peak separation disappear or shift when tip-sample distance, bias set point, or substrate is changed, the small gap is a tunneling artifact. Independently, infrared absorption or photoemission that finds no optical or filled-state onset near 0.3–0.6 eV would contradict the claim.","tokens_in":10711,"feed_emoji":"🧲","tokens_out":4338,"duration_ms":37245,"temperature":0.7,"pith_summary":"This paper claims that the layered magnetic semiconductor CrBr3 has a far smaller electronic energy gap than the 1.68–2.1 eV range assumed from optical absorption for decades. Using scanning tunneling spectroscopy on exfoliated flakes at 77 K, the authors find two intrinsic peaks flanking the Fermi level, one in the conduction band at 0.25 eV and one in the valence band at −0.31 eV, giving a peak-to-peak gap of 0.57 ± 0.04 eV and an onset gap of 0.29 ± 0.05 eV. If correct, the result settles a controversy opened by a 1.35 eV photoluminescence feature and changes how the material's optical, magnetic, and transport data are interpreted. The paper also proposes a DFT+U parameter set (U = 5 eV, J = 3 eV) that reproduces the small gap.","feed_headline":"CrBr3's real band gap is 0.57 eV, not 1.68–2.1 eV","feed_subtitle":"STS data at 77 K put the gap below 0.6 eV and match every known optical transition.","key_machinery":"The load-bearing measurement is the normalized differential conductance spectrum dI/dV, which is proportional to the local density of states, taken with a lock-in amplifier on thick exfoliated CrBr3 flakes on HOPG at 77 K. The normalization follows a broadening formula (Eq. 1) that suppresses noise without shifting peak positions. The argument is carried by the multi-peak spectrum: nine reproducible peaks, four on the conduction side and five on the valence side, whose pairwise energy differences line up with decades of optical absorption, Kerr rotation, reflection, and photoluminescence data. The near-Fermi pair, peak 1 at 0.25 eV and peak a at −0.31 eV, fixes the claimed small gap, while DFT+U calculations with U = 5 eV and J = 3 eV open a gap of similar size and leave CrBr3 gapless when U = J = 0.","core_discovery":"The central claim is that bulk CrBr3 is a small-gap semiconductor with an electronic band gap of 0.57 ± 0.04 eV, determined as the separation between the two dI/dV peaks nearest the Fermi level measured at 77 K. A more conservative reading of the same spectra puts the onset of tunneling conductance at 0.29 ± 0.05 eV. The authors argue that the two near-zero peaks are intrinsic band edges, not defect states, because the spectra are reproducible across locations and the atomic-resolution topography is defect-free. They further show that the energy differences between all other conduction-band and valence-band peaks in the dI/dV spectrum match every reported optical transition from 1.35 eV to 3.8 eV, which they take as evidence that the measured spectrum faithfully represents the density of states.","pith_inferences":["If the small gap is intrinsic, CrBr3 may behave more like a narrow-gap semiconductor in transport and optical devices than previously assumed, with thermal activation across 0.3–0.6 eV accessible in ordinary experiments; the paper does not pursue this consequence.","A testable extension is to measure STS on CrBr3 flakes on an insulating substrate and with varied tip-sample distances to see whether the zero-bias suppression and 0.57 eV peak separation persist; the paper does not model junction effects such as tip-induced band bending.","The assignment of the 1.35 eV photoluminescence to a peak-pair difference rather than the band edge suggests that PL-based gap estimates for other chromium trihalides may similarly need to be re-examined.","If confirmed by infrared absorption or photoemission, the 0.29 eV onset would place CrBr3 alongside other narrow-gap layered magnets and could change predictions for magnetotransport and spintronic applications."],"forward_implications":["The long-quoted 1.68 eV absorption feature becomes an inter-band transition between deeper peaks rather than the fundamental band gap, so optical assignments for CrBr3 need revision.","DFT calculations that force a gap near 1.7–2.1 eV require recalibration, and the U = 5 eV, J = 3 eV combination offers a concrete starting point for CrBr3.","A 0.29 eV onset gap should be observable in infrared absorption or electrical transport, giving independent experimental checks.","Because all reported optical transitions match the measured peak pairs, the electronic density of states appears essentially unchanged across the magnetic transition at 32 K, so spin ordering does not strongly reorganize the band structure.","Magnetic and optical interpretations that assume a large gap, such as exciton binding energies or d-d versus charge-transfer assignments, need re-evaluation."],"supporting_citations":[{"why":"Supplies the 1.68 eV absorption feature long quoted as the CrBr3 energy gap, which this paper reinterprets as a deeper inter-band transition.","marker":"[24]"},{"why":"Reports the 1.35 eV photoluminescence peak that opened the controversy over whether the true gap is much smaller than 1.68 eV.","marker":"[26]"},{"why":"Provides the 2.1 eV optical gap and transport context that the new 0.57 eV result directly contradicts.","marker":"[17]"},{"why":"Supplies the normalization formula used to convert raw dI/dV spectra into a quantity proportional to the local density of states.","marker":"[47]"},{"why":"Gives comparable STM/S measurements on MBE-grown CrBr3, providing the contrast in sample preparation and tip-sample conditions discussed for the imaging differences.","marker":"[9]"},{"why":"Reports the Kerr rotation and absorption transitions at 2.92–3.67 eV that are matched to the measured conduction-valence peak pairs.","marker":"[14,15,31]"},{"why":"Documents the wide range of DFT+U parameters and the importance of the on-site Coulomb repulsion U for CrBr3, motivating the chosen U = 5 eV, J = 3 eV set.","marker":"[20,40]"}],"fun_headline_variants":["STM reveals CrBr3's true band gap: 0.57 eV","CrBr3's gap shrinks from ~2 eV to 0.57 eV","CrBr3's gap isn't 2 eV, it's 0.57 eV","Tunneling spectroscopy finds CrBr3 gap of 0.57 eV"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The claim stands or falls on the assumption that the two dI/dV peaks near zero bias are the intrinsic band edges of bulk CrBr3 and that the suppressed conductance between them is the true band gap, not an artifact of tip-induced band bending, the low sample conductivity, or the HOPG substrate.","fun_headline_variants_meta":{"raw":{"variants":["STM reveals CrBr3's true band gap: 0.57 eV","CrBr3's gap shrinks from ~2 eV to 0.57 eV","CrBr3's gap isn't 2 eV, it's 0.57 eV","Tunneling spectroscopy finds CrBr3 gap of 0.57 eV"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001547,"raw_usage":{"total_tokens":6225,"prompt_tokens":1025,"completion_tokens":5200,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":641,"completion_tokens_details":{"reasoning_tokens":5112}},"tokens_in":641,"tokens_out":5200,"duration_ms":33299,"temperature":1.0,"reasoning_tokens":5112,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T10:02:14.636229+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure STS on the same material under deliberately varied junction conditions and on an insulating substrate: if the zero-bias suppression and the 0.57 eV peak separation disappear or shift when tip-sample distance, bias set point, or substrate is changed, the small gap is a tunneling artifact. Independently, infrared absorption or photoemission that finds no optical or filled-state onset near 0.3–0.6 eV would contradict the claim.","supporting_citations":[{"cited_title":"F.Dillon, H.Kamimura, andJ","cited_arxiv_id":null,"evidence_quote":"Supplies the 1.68 eV absorption feature long quoted as the CrBr3 energy gap, which this paper reinterprets as a deeper inter-band transition."},{"cited_title":"19, 3138 (2019)","cited_arxiv_id":null,"evidence_quote":"Reports the 1.35 eV photoluminescence peak that opened the controversy over whether the true gap is much smaller than 1.68 eV."},{"cited_title":"K.Kanazawa andG","cited_arxiv_id":null,"evidence_quote":"Provides the 2.1 eV optical gap and transport context that the new 0.57 eV result directly contradicts."},{"cited_title":"M.Feenstra, Phys","cited_arxiv_id":null,"evidence_quote":"Supplies the normalization formula used to convert raw dI/dV spectra into a quantity proportional to the local density of states."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives comparable STM/S measurements on MBE-grown CrBr3, providing the contrast in sample preparation and tip-sample conditions discussed for the imaging differences."}],"review_version":1}