{"id":"b8235796-3297-416f-aeae-4e19cb9c3da6","arxiv_id":"2411.16248","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"SEM electron channelling contrast imaging reveals twist domains in TMD heterostructures, with contrast inversions explained by a semi-empirical elastic plus inelastic scattering model.","lead":"This paper demonstrates that a standard scanning electron microscope can image the twisted crystal domains in atomically thin TMD heterostructures, even when sealed under an encapsulation layer. The method offers twistronics researchers a widely accessible, non-destructive tool for checking local stacking order in finished devices.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The predicted ~20° contrast-inversion angle in Fig. 2d is not independent of the data: it relies on an 'empirically determined weighting' whose value and fitting procedure are not reported, so the mechanistic claim is weaker than the abstract implies.","rationale":"The reader's weakest_assumption correctly identifies the empirically determined weighting in Fig. 2d as the key point where the mechanistic claim may be circular. The paper's own text confirms that the elastic contribution alone cannot explain the ~20° inversion at 1500 eV, and the inelastic contribution is added with an unspecified weight. Since the experimental data define the zero crossing, a fitted weight can trivially reproduce it. The stress-test pass finds this is indeed the most load-bearing concern: it affects the advertised theoretical insight, but not the practical imaging result, which is supported by multiple detectors, sample geometries, materials (MoS2, WS2), SEM instruments, tilt series, and encapsulation studies. The experimental core is convincing and justifies a conditional acceptance. The manuscript should either provide the weighting and a sensitivity analysis, or temper the abstract's language from 'reveals' to something like 'is consistent with' the observed contrast inversions. The verdict should remain CONDITIONAL, so no change from the reader's verdict is recommended.","tokens_in":24248,"tokens_out":3528,"duration_ms":37217,"concrete_test":"Report the number and values of the empirically determined weight(s) used for the black dots in Fig. 2d, then perform a leave-one-out cross-validation on the Fig. 2d dataset: refit the weight using all experimental points except the two that bracket the zero-contrast inversion (approximately the STEM2 point near 4-16° and the neighbouring point above 20°), and predict the sign of the held-out points. If either held-out point is predicted with the wrong sign, or if the inferred inversion angle shifts by more than one detector angular bin (~4°), the claimed 'specified scattering angle' for contrast inversion is fit-dependent rather than a robust physical prediction.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central mechanistic claim is that theoretical calculations reveal how elastic and inelastic scattering combine to produce contrast inversions at specified detector scattering angles. In the 'Domain contrast in suspended samples' section, the black theoretical points in Fig. 2d are said to be obtained 'by combining elastic and inelastic scattering contributions with an empirically determined weighting.' The paper does not state how many weights were adjusted, their fitted values, the fitting criterion, or any sensitivity analysis. Because the experimental data in Fig. 2d themselves contain the zero-contrast crossing between STEM2 (negative) and STEM5 (positive), and because the computed elastic contribution at 1500 eV is negative for all relevant angles (SI Fig. S3.4d and S3.5), the location of the predicted inversion is controlled by the relative amplitude of the oscillatory inelastic signal (trough near ~16°, peaks near ~22° and ~30°, Fig. 2c). A single weight chosen to match the low- and high-angle contrast magnitudes can place the zero crossing at ~20° essentially by construction. The abstract's phrase 'reveals ... contrast inversions at specified detector scattering angles' is therefore not an independent prediction unless the weight is fixed a priori or shown to be insensitive to reasonable variation. This concern does not undermine the experimental demonstration that SEM ECCI images twist domains, including under encapsulation, but it weakens the theoretical explanation that is advertised as a central finding.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript demonstrates that scanning electron microscope electron channelling contrast imaging (SEM ECCI) can reveal reconstructed twist domains in twisted TMD heterostructures, including samples sealed beneath hBN encapsulation, on bulk substrates, and for thicker TMD stacks. The experimental work is extensive: it quantifies domain contrast with propagated errors for multiple detectors (EsB, InL, ETD, and annular STEM segments), maps contrast versus scattering angle, studies the dependence on working distance, accelerating voltage, tilt, rotation, substrate, and encapsulation, and validates the approach on several SEM platforms and on both MoS2 and WS2. The paper also presents a theoretical model that combines elastic Bragg scattering and a semi-classical Monte Carlo treatment of inelastic scattering to explain the observed sign of domain contrast and its inversion near 20° scattering angle. The central mechanistic claim, however, rests on an 'empirically determined weighting' that is not described or justified, and the theoretical account of tilt-dependent contrast is absent. The experimental demonstration is convincing and likely useful, but the theoretical explanation is not yet at the level implied by the abstract.","tokens_in":24504,"tokens_out":2690,"duration_ms":29083,"significance":"If the experimental claims are accepted, this is a practical and widely accessible characterization route for twist domains: it is non-destructive, works on conventional substrates, and crucially works under hBN encapsulation, which is the geometry relevant to real devices. The systematic quantification of contrast (Michelson contrast with propagated errors) across detectors, tilts, thicknesses, and instruments is a genuine strength. The paper also shows that the method does not require precise azimuthal alignment at zero tilt, and it quantifies contrast degradation from encapsulation and contamination. The theoretical modelling is less convincing: the elastic-scattering analysis is ab initio and parameter-light, but the inelastic model is semi-classical and is combined with the elastic contribution using an unspecified empirical weight that is evidently chosen to match the very data it is used to explain. Consequently the claimed prediction of the ~20° contrast-inversion angle is not independent. The strength of the paper is therefore primarily experimental; the mechanistic explanation needs substantial reframing or additional validation.","major_comments":[{"comment":"The theoretical points in Fig. 2d are obtained 'by combining elastic and inelastic scattering contributions with an empirically determined weighting', but the manuscript does not state the value of the weight, how many weights were adjusted, the fitting criterion, or any sensitivity analysis. Since the elastic contribution at 1500 eV is negative for all relevant angles (SI Fig. S3.4d and S3.5) and the inelastic contribution has a trough near 16° and peaks near 22° and 30° (Fig. 2c), a single weight chosen to reproduce the experimental low-angle and high-angle contrast magnitudes can place the zero crossing at ~20° essentially by construction. The abstract's claim that the calculations 'reveal' contrast inversions at specified detector scattering angles is therefore not a parameter-free prediction. Please report the fitted weighting, show the sensitivity of the predicted inversion angle to that weight, or fix it a priori from the model rather than from the data being explained.","section":"Domain contrast in suspended samples; Fig. 2d; SI Section 3.2"},{"comment":"The abstract also claims that the calculations reveal contrast inversions 'at specified detector scattering angles and sample tilts', but no theoretical calculation of contrast versus specimen tilt is presented. Figure 4 shows experimental tilt dependence (inversion near 11°, optimum near 21°), and SI Fig. S3.16 shows emission intensity versus azimuthal angle for fixed emission angles, which is not the same as a tilt series. The theoretical model in Fig. 2c treats scattering angle relative to the incident beam, not the sample tilt angle. Either add a calculation that predicts the tilt dependence or remove 'sample tilts' from the abstract's mechanistic claim.","section":"Domain contrast as a function of sample tilt; Fig. 4; SI Section 3.2.4"},{"comment":"The inelastic scattering model is semi-classical and omits core-electron ionization and plural scattering, which are invoked to explain the discrepancy at high scattering angles. The paper does not show that the inelastic model, without the empirical weighting, reproduces any independent observable (e.g., absolute angular distributions or energy dependence of the contrast). Given that the contrast inversion is the key mechanistic result, the model should be validated against data other than the single fitted curve in Fig. 2d, or the theoretical claim should be softened to a qualitative consistency check rather than a quantitative prediction.","section":"Conclusions; SI Sections 3.2 and 3.2.4"}],"minor_comments":[{"comment":"'preferrable' should be 'preferable'.","section":"Conclusions"},{"comment":"Typo: 'c4ontributions' should be 'contributions'.","section":"SI Section 3.1"},{"comment":"The contrast metric is spelled 'Michaelson' in several places but 'Michelson' in the SI; please standardize.","section":"Throughout"},{"comment":"The main text defines contrast using IXM' and IMX' while the SI uses I1 and I2; please align the notation so the reader can map between the two.","section":"SI Section 2"},{"comment":"'there is a contrast is reversal at ~11°' is ungrammatical; should read 'there is a contrast reversal at ~11°'.","section":"Figure 4 caption"},{"comment":"The explanation for the absence of ETD contrast mentions the need to include high-energy inelastic core-electron scattering for a full treatment; this limitation should be stated in the main text near the model description, not only in the SI.","section":"Domain contrast in suspended samples"}],"recommendation":"major_revision","confidential_remarks":"The stress-test concern about the empirical weighting is valid and lands on the central mechanistic claim. The experimental contribution is strong enough that the paper is likely salvageable, but the theory section needs to be reframed as an illustrative model with the fitted weight explicitly reported and its influence on the inversion angle quantified, or the theoretical claims in the abstract must be softened. I would also check whether the journal's scope tolerates a paper where the main novelty is experimental characterization; in my view the encapsulation and multi-platform demonstration is sufficient for a materials-science journal."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is a useful, mostly convincing experimental paper. The SEM ECCI method for mapping twist domains in TMDs is a real practical advance, and the paper does a thorough job of showing when it works and when it doesn't. The theoretical story is more fragile than the abstract implies, but that doesn't sink the main result.\n\nWhat's new here is the systematic scope. Earlier work had shown ECCI on a few twisted 2D samples, but this paper maps contrast across detector type and angle, sample tilt, substrate, encapsulation, layer thickness, and two material systems, and demonstrates transfer across several SEM platforms. The encapsulation result is the one that people will actually use: being able to see twist domains non-destructively in a finished device is a genuine capability gain.\n\nThe experimental core is solid. Contrast values come with propagated errors, the sample-flip control confirms the domain assignments, and the qualitative picture—negative elastic contrast at low scattering angles, positive inelastic contrast at high angles—is a useful organizing framework. The EsB detector working without tilt is a practical point that deserves emphasis.\n\nThe soft spot is the 'empirically determined weighting' in Figure 2d. One (or a few) fitted numbers connect the elastic and inelastic contributions, and the paper does not report their values, the fitting criterion, or any sensitivity test. Because the experimental data themselves contain the zero crossing, locating the predicted inversion at ~20° is at least partly constructed, not derived. The abstract's 'reveal' overstates the predictive strength. To their credit, the authors admit the high-angle discrepancy and mark the substrate ETD explanation as a hypothesis, so this is an omission of transparency rather than a hidden fatal flaw. A referee should ask for the weight to be reported, a sensitivity sweep, and more careful wording in the abstract and conclusions.\n\nWho is this for: anyone doing twistronics device characterization. The method is cheap, fast, and non-destructive, and the paper gives enough detail to reproduce it. It deserves a serious referee, not a desk reject. With the fitting made transparent and the claims softened, I'd take it as published.","headline":"A practical, experimentally strong SEM imaging method for twist domains; the mechanistic model has a fitting soft spot that needs disclosure and softer claims.","tokens_in":25069,"tokens_out":3172,"would_cite":true,"duration_ms":30636,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["68.37.Hk"],"model":"deepseek-v4-flash","headline":"Conventional SEM, via electron channelling contrast, can map reconstructed twist domains in TMD heterostructures even when they are sealed beneath an encapsulation layer.","keywords":["twistronics","transition metal dichalcogenides","electron channelling contrast imaging","scanning electron microscopy","moiré superlattice","3R stacking domains","hBN encapsulation","Michelson contrast"],"falsifier":"Measure the STEM detector domain contrast versus collection angle for a twisted bilayer whose stacking assignment is known from an independent probe, with the elastic/inelastic mixing weight fixed by the scattering calculation alone rather than fitted to the same images; if the zero-contrast crossing does not fall near 20 degrees, the proposed balance of elastic and inelastic scattering is wrong.","tokens_in":24067,"feed_emoji":"🔬","tokens_out":7326,"duration_ms":87263,"temperature":0.7,"pith_summary":"This paper argues that an ordinary scanning electron microscope, using electron channelling contrast, can image the reconstructed stacking domains of twisted transition-metal dichalcogenide bilayers, including domains buried under a hexagonal boron nitride encapsulation layer. The central experimental claim is that this non-destructive method works on as-produced samples on conventional substrates and does not require knowing the crystal orientation in advance. To explain why domain brightness reverses between detectors and with sample tilt, the paper combines elastic Bragg scattering between the two layers with a Monte Carlo model of inelastic secondary-electron generation and attenuation. If the picture is right, the twistronics community gains a cheap, widely available tool for correlating local domain architecture with device behaviour.","feed_headline":"Standard SEM sees twist domains hidden under encapsulation","feed_subtitle":"Electron channelling contrast maps reconstructed TMD stacking without destroying the device.","key_machinery":"The load-bearing object is a two-layer interference model for elastically Bragg-scattered partial waves, expressed as $I_n = 2A^2(1 + \\cos(\\mathbf{G}_n \\mathbf{r}_0 - \\Delta\\varphi_n))$, where $\\Delta\\varphi_n = d(k_0 - \\sqrt{k_0^2 - G_n^2})$ is the phase difference accumulated because the top and bottom layers sit a distance $d$ apart, and the interlayer offset $\\mathbf{r}_0$ is what distinguishes the XM' and MX' stackings. This predicts which Bragg reflections carry domain contrast and at which scattering angles they appear. A separate semi-classical Monte Carlo model treats inelastic signal as secondary electrons generated in proportion to local electron density and attenuated along their escape path using empirical inelastic mean-free-path data. Combining the two scattering channels with an empirically determined weighting reproduces the measured angular dependence of STEM detector contrast, including the inversion near 20 degrees.","core_discovery":"For freely suspended twisted MoS2 bilayers at 1500 V, low-angle transmitted electrons give negative domain contrast dominated by the {1-100} Bragg reflection, while high-angle transmitted and backscattered electrons give positive contrast dominated by inelastic scattering; the zero crossing sits near 20 degrees scattering angle. The sign of contrast is therefore not intrinsic to a domain but depends on which scattering population the detector collects. On bulk substrates the reflected signals behave similarly for both low- and high-energy detectors, because the substrate adds positive elastic and inelastic contributions from electrons crossing the bilayer in reverse; on ultra-thin supports the ETD signal can appear inverted. A specimen tilt near 21 degrees multiplies contrast by roughly four and reverses its sign, and encapsulation by 3.5 nm of hBN reduces contrast by about half while leaving domains visible. The paper concludes that optimum imaging is obtained by maximising collected signal while avoiding the angular and tilt conditions where contrast inverts.","pith_inferences":["Beyond the paper, a robust inversion angle would give a scattering-based fingerprint of stacking registry that does not depend on image histogram calibration.","Beyond the paper, the ETD result implies that in atomically thin samples the conventional secondary-electron detector signal is largely reporting channelled higher-energy electrons, so ETD micrographs of such samples should not be read as pure surface topography.","Beyond the paper, the same combined elastic and inelastic contrast logic should apply to other layered systems with two distinct registries, such as twisted graphene on hBN, wherever a low-index Bragg reflection falls inside the accessible detector angle.","Beyond the paper, serial imaging after plasma cleaning showed contrast recovery, suggesting that quantitative contrast versus contamination thickness could turn ECCI into a calibrated probe of surface cleanliness."],"forward_implications":["Twist domains in finished, encapsulated devices can be checked non-destructively, so domain architecture can be correlated with measured electronic properties of the same device.","Detector choice matters: annular backscattered-electron detectors give consistent positive contrast at zero tilt and do not require rotating the sample toward the detector, while ETD images can mislead unless the tilt and substrate regime is understood.","Tilting the sample by about 21 degrees can quadruple domain contrast, and the accompanying sign flip provides an internal consistency check that the observed contrast really comes from the stacking domains.","The same BSE channelling approach should map domains in WS2 with higher contrast than MoS2, and remains usable for lower-layer thicknesses up to at least three layers."],"supporting_citations":[{"why":"Prior demonstration of channelling-contrast imaging of twisted domains that required precise tilting; this paper's non-tilted BSE result is measured against it.","marker":"[2]"},{"why":"Earlier ETD imaging of stacking domains in encapsulated WSe2 at 40 degrees tilt; the present work generalises and quantifies encapsulation contrast loss.","marker":"[31]"},{"why":"Establishes the XM'/MX' domain assignment used to fix the sign of domain contrast for the imaging conditions.","marker":"[32]"},{"why":"Supplies the tear-and-stack fabrication route for the twisted TMD bilayers studied here.","marker":"[34]"},{"why":"Provides the inelastic mean-free-path description that the semi-classical inelastic model is built to reproduce.","marker":"[36]"},{"why":"Supplies the electron-scattering mean-free-path and cross-section relations used in the Monte Carlo treatment of secondary electrons.","marker":"[37]"}],"fun_headline_variants":["SEM trick reveals twist domains under encapsulation","Twist domains in TMDs seen through SEM contrast","New SEM method maps domains in twisted 2D materials","Non-destructive imaging of twist domains with SEM","See twist domains hidden beneath encapsulation"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The calculated position of the contrast inversion depends on an empirically chosen weighting that mixes the elastic and inelastic signals to match the measured contrast values, so the theory's prediction near 20 degrees is not fully independent of the data it explains.","fun_headline_variants_meta":{"raw":{"variants":["SEM trick reveals twist domains under encapsulation","Twist domains in TMDs seen through SEM contrast","New SEM method maps domains in twisted 2D materials","Non-destructive imaging of twist domains with SEM","See twist domains hidden beneath encapsulation"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000177,"raw_usage":{"total_tokens":1275,"prompt_tokens":909,"completion_tokens":366,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":525,"completion_tokens_details":{"reasoning_tokens":296}},"tokens_in":525,"tokens_out":366,"duration_ms":112516,"temperature":1.0,"reasoning_tokens":296,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T13:19:40.678985+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the STEM detector domain contrast versus collection angle for a twisted bilayer whose stacking assignment is known from an independent probe, with the elastic/inelastic mixing weight fixed by the scattering calculation alone rather than fitted to the same images; if the zero-contrast crossing does not fall near 20 degrees, the proposed balance of elastic and inelastic scattering is wrong.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Prior demonstration of channelling-contrast imaging of twisted domains that required precise tilting; this paper's non-tilted BSE result is measured against it."}],"review_version":1}