{"id":"ec49414a-830f-4e8a-bc49-aaafd0b8dd6a","arxiv_id":"2507.05560","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Core-level EELS reveals local variations in carbon K-edge van Hove singularity peaks within a moiré unit cell of twisted bilayer graphene, with linewidth broadening in AB and bridge regions.","lead":"Using a sharp electron probe, the authors mapped carbon K-edge electron energy loss spectra inside a single moiré unit cell of twisted bilayer graphene and found that van Hove singularity peaks broaden in AB and bridge stacked regions compared to AA. The result suggests core-exciton lifetimes vary with local stacking and highlights core-loss EELS as a local probe of electronic structure in moiré materials.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Linewidth differences of 5–50% are extracted from heavily overlapping Gaussians after deconvolution with a 170 meV zero-loss kernel; without a blinded synthetic test, the broadening and its core-exciton-lifetime interpretation are not yet secured.","rationale":"I retain the reader's conditional verdict, but the load-bearing blocker is not primarily the ADF-to-stacking assignment. At 9.8°, lattice relaxation in TBG is weak, and the simulated ADF image plausibly reproduces the three contrast levels; that concern is secondary. The decisive step is quantitative: the 5–50% linewidth broadening is extracted from five overlapping Gaussian components after maximum-entropy deconvolution of spectra acquired with 170 meV resolution and 50 meV/channel dispersion. The covariance-based error bars do not capture model degeneracy, and the paper provides no core-hole calculation connecting a stacking-dependent lifetime to a predicted width change. The experiment may well be correct, but the current analysis leaves a fitting artifact as an untested alternative explanation. A blinded synthetic test and replicate acquisitions would settle whether the broadening is real. This does not change the conditional verdict, so I recommend no verdict adjustment.","tokens_in":11206,"tokens_out":9574,"duration_ms":114988,"concrete_test":"Run a blinded synthetic test on the raw core-loss spectra: use the bright-region spectrum as a template, synthesize grey/dark spectra by adding only Poisson noise, small gain/background variations, and ±10–30 meV energy offsets with no physical broadening, then apply the exact deconvolution and five-Gaussian fitting with labels hidden. If false 5–50% width increases appear, the broadening claim is not established; if not, and synthetic broadened inputs are recovered, the concern is retired. A complementary check is to acquire ~10 repeated spectra from one bright region and compare run-to-run fitted-width scatter with the reported bright–dark difference.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim depends on comparing fitted widths of five overlapping van Hove singularity peaks (P1–P5) across bright, grey, and dark regions (Fig. 4c,d). The spectra are deconvolved with the simultaneously measured zero-loss peak (FWHM 170 meV) and then fit with five Gaussian components whose positions and widths are simultaneously free, plus broad π* and continuum terms. Several features make the extracted 5–50% broadening fragile. At 50 meV/channel dispersion, the smallest reported differences (5% of a ~0.2–0.4 eV width) are tens of meV, comparable to or below the instrument resolution before deconvolution; maximum-entropy deconvolution can introduce regularization-dependent line-shape changes. The five peaks overlap strongly, so a grey/dark spectrum with slightly different relative intensities, a small energy offset, or a different continuum background can be equivalently represented by broader components. The reported error bars are only variance–covariance fitting errors and do not include model degeneracy or run-to-run variation. No independent core-hole calculation or lifetime estimate is provided; the paper explicitly defers such theory to future work. Thus the assignment of the observed spectral differences to core-exciton lifetime variation is not yet secured, even if the stacking assignment is accepted.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports STEM-EELS measurements of the carbon K-edge in twisted bilayer graphene (TBG) with a monochromated 60 kV electron beam. The authors first show that both low-loss and core-loss spectra exhibit twist-angle-dependent features assigned to optical and core-level excitations involving van Hove singularities. They then compare core-loss spectra from different contrast regions within a moiré unit cell at 9.8° and report 5–50% broadenings of five Gaussian components (P1–P5) in AB/BA ('grey') and bridge ('dark') regions relative to AA ('bright') regions, with little peak-position change. They interpret the broadenings as local modifications of core-exciton lifetimes and band structure due to local stacking geometry. The paper includes a twist-angle series from 4.8° to 25.7°, supporting a global trend, and claims the first observation of local variations in core-level van Hove singularity peaks within a single moiré unit cell.","tokens_in":11426,"tokens_out":3443,"duration_ms":42580,"significance":"The reported measurements are technically challenging and potentially important: core-level EELS with 170 meV resolution and an atomically localized probe could in principle reveal local electronic-structure variations inside a moiré unit cell, going beyond global optical and STS studies. The twist-angle dependence of the carbon K-edge fine structure is itself a useful addition, and the authors are careful to acquire spectra from impurity-free regions and to calibrate energy axes with the simultaneously recorded zero-loss peak. However, the central quantitative claim of 5–50% linewidth broadening rests on a multi-Gaussian decomposition after maximum-entropy deconvolution, and the manuscript provides no synthetic or repeatability test to demonstrate that such broadenings are recovered reliably from the measured spectra. The stacking assignment also depends on a rigid-model STEM simulation that neglects relaxation. If the broadening claim survives validation, the paper would open a new route to local core-exciton physics in moiré materials; at present that claim is not yet secured.","major_comments":[{"comment":"The central result of 5–50% linewidth broadening is extracted from five-Gaussian fits performed after maximum-entropy deconvolution with the 170 meV zero-loss peak as the kernel. The smallest reported broadenings are tens of meV, comparable to or smaller than the instrument resolution before deconvolution, and the five components overlap strongly. Maximum-entropy deconvolution can introduce regularization-dependent line-shape changes, and a spectrum with slightly different relative intensities, a small energy offset, or a different continuum background can be equivalently represented by broader components. The error bars shown in Fig. 4d are variance–covariance fitting errors only and do not include model degeneracy, run-to-run variation, or deconvolution uncertainty. A blinded synthetic test—where known spectra with prescribed peak widths are convolved with the measured zero-loss peak, processed with the same deconvolution and fitting protocol, and compared to the input—is needed to establish that the reported 5–50% differences are recoverable. Without such a test, the linewidth differences and their core-exciton-lifetime interpretation are not secured.","section":"Fig. 4c,d and Data analysis"},{"comment":"The assignment of bright, grey, and dark ADF regions to AA, AB/BA, and bridge stacking relies on a rigid-model STEM simulation that explicitly ignores out-of-plane displacement and moiré relaxation, as stated in the text: 'no out-of-plane displacement nor the relaxation of moiré were considered' and the influence of higher-order moiré patterns is 'beyond the scope of this study.' Since the local stacking geometry is the independent variable in the paper's main claim, this assignment needs experimental validation—for example, a quantitative comparison between the simulated and experimental ADF profiles, or an assessment of how relaxation at 9.8° and 6.4° modifies the expected contrast in each region. If the actual local stacking differs from the rigid model, the correlation between linewidth broadening and stacking geometry would be undermined.","section":"Fig. 4a,b"},{"comment":"The interpretation that the observed broadening 'implies a reduction in the lifetime of core excitons' is presented without an independent estimate of core-hole lifetimes or a core-hole calculation; the paper explicitly defers such theory to future work. This leaves the possibility that the apparent broadening is caused by unresolved band-structure shifts, local strain, or changes in relative intensities and continuum background rather than by a genuine lifetime change. The TMDC charge-transfer analogy is suggestive but not a quantitative substitute. At minimum, the authors should provide an estimate of the expected core-exciton lifetime variation from published core-hole widths for carbon and from local interlayer coupling arguments, and should explicitly state that the current data cannot distinguish lifetime effects from band-structure contributions.","section":"Conclusion and Discussion of lifetime"}],"minor_comments":[{"comment":"The text says the fine structure 'includes four components of C1s excitations' to EA*, EB*, and Eg*, but then describes five components, P1–P5; please clarify the counting and the exact assignment of each component.","section":"Fig. 3a text"},{"comment":"The Data analysis section refers to 'Extended Figure 3' while the supplement labels the corresponding item Supplementary Figure 3; please make the referencing consistent.","section":"Data analysis"},{"comment":"The sentence 'EB starts to separate form EM' contains a typo ('form' should be 'from'); please correct it.","section":"Fig. 2c"},{"comment":"For the 6.4° and 4.8° data, the shifts of 200–300 meV and the local-structure dependence are stated without a table or plot of the fit parameters with correlation information; adding a supplementary table of all fitted positions, widths, and intensities would improve reproducibility.","section":"Data analysis and Supplementary Fig. 4"},{"comment":"The choice of a broad Gaussian component for energy losses above 286.5 eV and the fixed overall π* envelope extracted from single-layer graphene should be described more explicitly, including how sensitive the P1–P5 widths are to reasonable variations of these fixed components.","section":"Experimental Section"}],"recommendation":"major_revision","confidential_remarks":"The paper presents a plausible and technically impressive measurement, but the central quantitative claim—position-dependent linewidth broadening of overlapping van Hove singularity peaks—requires validation through synthetic tests and a more rigorous treatment of the stacking assignment. The readers' concerns about the five-Gaussian analysis and the rigid-model STEM simulation are well founded and should be addressed before publication. This is fixable within the scope of the manuscript, so I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First, what is new: this is the first report of core-loss EELS fine structure mapped within a single moiré unit cell of twisted bilayer graphene. The twist-angle dependence of the carbon K-edge vHs peaks (P1–P5) is also a nice extension of earlier valence-loss work. The experimental side is careful: spectra were taken on impurity-free areas, the ZLP was acquired simultaneously, and the delocalization argument for core-loss at 60 kV is sound. Looking at Fig. 4c, the bright/grey/dark spectra do look different in the π* region, so the basic observation of local spectral variation is plausible and probably real.\n\nThe soft spots are all in the interpretation. The five-Gaussian fit has many free parameters, and the error bars only reflect the fitting covariance. The smallest broadening values (5% of ~0.2–0.4 eV) are tens of meV, at or below the 50 meV/channel dispersion and well below the 170 meV resolution before deconvolution. The maximum-entropy deconvolution can change line shapes in regularization-dependent ways, and no synthetic test is shown to prove that the broadening is recoverable. The stacking assignment relies on a rigid-model simulation that ignores out-of-plane relaxation, which the authors acknowledge; that is a real concern, though at 9.8° the effect is probably modest. And the paper explicitly defers core-hole calculations, so the lifetime interpretation is an inference, not a test. These are not fatal—the observation may well stand—but they mean the headline claim is stronger than the evidence.\n\nFor whom is this written? Experimental EELS and moiré-materials people will get value. It deserves a serious referee, and I would send it out with requests for a blinded synthetic fit test, replicate measurements on another sample, and ideally a core-hole calculation. The central observation is worth publishing, but the lifetime story needs more support.","headline":"Local carbon K-edge EELS mapping inside a TBG moiré cell is new and the raw spectra do differ by region; the 5–50% linewidth broadening and its core-exciton-lifetime interpretation, however, are not yet secured given the heavy fitting and deconvolution.","tokens_in":12014,"tokens_out":2502,"would_cite":true,"duration_ms":26937,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["79.20.Uv","73.22.Pr","71.35.-y"],"model":"deepseek-v4-flash","headline":"The carbon K-edge fine structure of twisted bilayer graphene varies within a single moiré unit cell, showing local linewidth broadening of van Hove peaks caused by stacking-dependent core-exciton lifetimes.","keywords":["twisted bilayer graphene","moiré superlattice","core-loss EELS","van Hove singularity","exciton lifetime","monochromated STEM","carbon K-edge","local stacking geometry"],"falsifier":"Perform the same core-loss mapping on a moiré cell whose local stacking is independently determined by atomic-resolution STEM images that include lattice relaxation, and check whether the 5 to 50 percent linewidth broadening still tracks the rigid-model AA/AB/bridge assignment; alternatively, compute carbon K-edge core-exciton lifetimes for relaxed AA and AB stackings and compare the predicted linewidth difference with the measured one.","tokens_in":10982,"feed_emoji":"🔬","tokens_out":5174,"duration_ms":56813,"temperature":0.7,"pith_summary":"The paper claims that the carbon K-edge fine structure of twisted bilayer graphene changes from point to point within a single moiré unit cell, so that core-level excitations to van Hove singularities have different linewidths depending on local stacking. Using core-loss electron energy loss spectroscopy (EELS) with a monochromated 60 kV scanning transmission electron microscope probe, the authors resolve spectra from AA, AB/BA, and bridge regions and find 5 to 50 percent broadening of the van Hove peaks in the AB and bridge regions relative to AA. The broadening is interpreted as a shorter core-exciton lifetime where local interlayer coupling and charge transfer are stronger. If correct, this turns core-level EELS into a local probe of excitonic and band-structure variation inside moiré cells, complementing the global twist-angle picture that previous studies have provided.","feed_headline":"Core-loss peaks broaden where stacking changes inside a moiré cell","feed_subtitle":"Carbon K-edge spectra show 5–50% wider van Hove peaks in AB and bridge regions relative to AA stacking.","key_machinery":"The load-bearing object is the fine structure of the carbon 1s K-edge, measured by monochromated core-loss EELS at 170 meV energy resolution with an atomically sharp probe whose signal delocalization is about 0.2 nm. This localization allows spectra to be assigned to specific stacking regions, AA (bright), AB/BA (grey), and bridge (dark), within the 1.5 nm moiré cell. The argument proceeds through a five-Gaussian line-shape analysis (P1 to P5) that maps the van Hove peaks to unoccupied saddle-point states $E_A^*$, $E_B^*$, and the minigap state $E_g^*$, and through comparison of the extracted linewidths, which report core-exciton lifetimes. The twist-angle series of the same peaks establishes that the core-level excitations track the same band hybridization seen in optical absorption.","core_discovery":"The central discovery is that the core-level van Hove singularity peaks in the carbon K-edge are not uniform across a moiré unit cell. In a 9.8 degree twisted bilayer graphene, spectra from the ADF-bright (AA-stacked), grey (AB/BA), and dark (bridge) regions at equivalent positions show essentially unchanged peak energies but systematically broader linewidths in the AB and bridge regions. The paper attributes this to shorter core-exciton lifetimes caused by stronger local interlayer interaction and charge transfer, together with modifications of the local band structure. The same position-dependent broadening appears at 6.4 degrees and becomes more pronounced at 4.8 degrees, where lattice reconstruction starts, whereas at 9.8 degrees the peak positions remain nearly fixed across the cell. The authors present this as the first corroboration of locally modified core-exciton behavior inside a moiré unit cell.","pith_inferences":["If the lifetime interpretation holds, core-exciton linewidths could be used to estimate local dielectric screening or electron-hole interaction strength across moiré cells, a quantity the paper does not directly extract.","The global twist-angle data suggest that core-level van Hove energies track the same band-structure evolution as optical transitions, making core-loss spectroscopy a potential proxy for predicting optical response at twist angles where direct optical measurement is difficult.","The observed broadening in AB and bridge regions is attributed to faster charge transfer, so a testable extension is to correlate the same spectral maps with local electrochemical or transport measurements at the moiré scale.","At 4.8 degrees, reconstruction effects appear; a systematic mapping across twist angles near the magic angle could reveal whether lattice relaxation changes the local exciton-lifetime pattern in a detectable way."],"forward_implications":["Core-loss EELS can serve as a sub-nanometer local probe of unoccupied band structure and exciton lifetimes in moiré materials.","The twist-angle dependence of the carbon K-edge van Hove peaks means core-level spectroscopy is sensitive to the same interlayer hybridization that controls the optical properties of twisted bilayer graphene.","Linewidth broadening inside a moiré cell offers a position-dependent measure of interlayer coupling and local charge-transfer rate.","Extending the method to smaller twist angles could track the onset of lattice reconstruction through local spectral broadening and shifts.","The same approach could map local excitonic effects in other van der Waals heterostructures beyond graphene."],"supporting_citations":[{"why":"Provides the band theory of twisted bilayer graphene, including the van Hove singularities and minigap that the core-level peaks are assigned to.","marker":"7"},{"why":"Reports optical absorption measurements of van Hove singularities and excitonic effects, serving as the baseline for the twist-angle dependence the paper reproduces.","marker":"16"},{"why":"Documents atomic and electronic reconstruction in twisted bilayer graphene, used to explain the stronger local shifts at small twist angles.","marker":"20"},{"why":"Shows rotation-angle-dependent excitonic absorption in van der Waals heterostructures measured by EELS, supporting the analogy between stacking geometry and exciton lifetime.","marker":"21"},{"why":"Demonstrates hyperspectral imaging of exciton confinement within a moiré unit cell, establishing the precedent for local excitonic mapping that this paper extends to core-level excitations.","marker":"23"},{"why":"Defines the fundamental limits to spatial resolution in EELS, justifying the claim that core-loss signals are localized to roughly 0.2 nm.","marker":"24"},{"why":"Establishes atom-by-atom core-loss spectroscopy, the methodological foundation for using core-level EELS as a local electronic-structure probe.","marker":"25"},{"why":"Shows that monochromated core-loss EELS can resolve unoccupied van Hove singularities in carbon nanostructures, providing the experimental template for the present measurements.","marker":"26"},{"why":"Provides evidence that local charge-transfer rates vary in twisted bilayer graphene, supporting the paper's explanation of shorter core-exciton lifetimes in AB and bridge regions.","marker":"32"}],"fun_headline_variants":["Moiré unit cell hosts locally modified core-exciton lifetimes","AB stacking shortens core-exciton lifetime in twisted bilayer graphene","Core-loss peaks reveal stacking-dependent exciton lifetimes in moiré","First corroboration of core-exciton variation inside moiré cell","Twisted bilayer graphene: core-level excitations vary by local stacking"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The mapping of the three ADF contrast regions to AA, AB/BA, and bridge stacking rests on a rigid-model simulation that ignores out-of-plane displacement and moiré relaxation; if the real local stacking differs from that model, the claimed relation between linewidth and stacking geometry would not follow.","fun_headline_variants_meta":{"raw":{"variants":["Moiré unit cell hosts locally modified core-exciton lifetimes","AB stacking shortens core-exciton lifetime in twisted bilayer graphene","Core-loss peaks reveal stacking-dependent exciton lifetimes in moiré","First corroboration of core-exciton variation inside moiré cell","Twisted bilayer graphene: core-level excitations vary by local stacking"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000336,"raw_usage":{"total_tokens":1851,"prompt_tokens":926,"completion_tokens":925,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":542,"completion_tokens_details":{"reasoning_tokens":833}},"tokens_in":542,"tokens_out":925,"duration_ms":10474,"temperature":1.0,"reasoning_tokens":833,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T19:22:54.584650+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Perform the same core-loss mapping on a moiré cell whose local stacking is independently determined by atomic-resolution STEM images that include lattice relaxation, and check whether the 5 to 50 percent linewidth broadening still tracks the rigid-model AA/AB/bridge assignment; alternatively, compute carbon K-edge core-exciton lifetimes for relaxed AA and AB stackings and compare the predicted linewidth difference with the measured one.","supporting_citations":[{"cited_title":"Physical Review B 2013, 87, 205404, DOI: 10.1103/PhysRevB.87.205404","cited_arxiv_id":null,"evidence_quote":"Provides the band theory of twisted bilayer graphene, including the van Hove singularities and minigap that the core-level peaks are assigned to."},{"cited_title":"W.; Liang, Y.; Brown, L.; Yang, L.; Park, J., Van Hove singularities and excitonic effects in the optical conductivity of twisted bilayer graphene","cited_arxiv_id":null,"evidence_quote":"Reports optical absorption measurements of van Hove singularities and excitonic effects, serving as the baseline for the twist-angle dependence the paper reproduces."},{"cited_title":"K.; Lin, Y","cited_arxiv_id":null,"evidence_quote":"Shows rotation-angle-dependent excitonic absorption in van der Waals heterostructures measured by EELS, supporting the analogy between stacking geometry and exciton lifetime."},{"cited_title":"H.; Blach, D","cited_arxiv_id":null,"evidence_quote":"Demonstrates hyperspectral imaging of exciton confinement within a moiré unit cell, establishing the precedent for local excitonic mapping that this paper extends to core-level excitations."},{"cited_title":"F., Limits to the spatial, energy and momentum resolution of electron energy-loss spectroscopy","cited_arxiv_id":null,"evidence_quote":"Defines the fundamental limits to spatial resolution in EELS, justifying the claim that core-loss signals are localized to roughly 0.2 nm."},{"cited_title":"Nature 2010, 468, 1088-1090, DOI: 10.1038/nature09664","cited_arxiv_id":null,"evidence_quote":"Establishes atom-by-atom core-loss spectroscopy, the methodological foundation for using core-level EELS as a local electronic-structure probe."},{"cited_title":"Nano Letters 2016, 16, 3661-3667, DOI: 10.1021/acs.nanolett.6b00825","cited_arxiv_id":null,"evidence_quote":"Shows that monochromated core-loss EELS can resolve unoccupied van Hove singularities in carbon nanostructures, providing the experimental template for the present measurements."}],"review_version":1}