{"id":"58ebbe91-3060-4488-8ae7-aa3556f9764a","arxiv_id":"1908.06284","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"ABC-stacked few-layer graphene can be converted to ABA stacking by Joule heating or by a train of laser pulses, with the moving domain wall imaged at nanometer scale.","lead":"This paper demonstrates two ways to switch few-layer graphene between two atomic stacking patterns: heating with an electric current or with many short laser pulses. The pattern change is tracked with Raman microscopy and electron microscopy, and matters because stacking controls whether graphene acts as a semiconductor or a semimetal.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The TEM partial-transition claim rests on a diffraction ratio calibrated for trilayers and on no post-train spatial image; layer-count and laser-damage effects are not ruled out.","rationale":"The paper's central qualitative claim is plausible: the Joule-heating experiment shows a voltage-dependent evolution of the 2D Raman line shape and of the R parameter, and the TEM section does contain a nominally ABC region whose diffraction ratio changes after a 100 kHz pulse train. That said, the TEM component is the weakest link because every quantitative statement about the laser-induced transition is mediated by a single scalar R calibrated against simulated trilayers. The sample is not a trilayer, and the supplement itself documents layer-number sensitivity of exactly the peak used for discrimination. In addition, the only spatial images of laser exposure (Fig. 4j-k) show contrast variations not tied to stacking, so a diffraction-ratio-only readout is not enough to distinguish ABC-to-ABA conversion from optically induced damage or rippling. This does not disprove the transition, but it means the abstract's 'laser-pulse train' route and the 'thermal' mechanism are conditional on a validation that is absent. The reader's weakest_assumption already pointed at calibration and uncertainty; I agree in direction but sharpen it to the layer-count mismatch and missing post-train spatial verification. Because the qualitative claim has independent support from the Raman Joule-heating data, I do not move the verdict; I retain CONDITIONAL.","tokens_in":12024,"tokens_out":9248,"duration_ms":100476,"concrete_test":"Repeat the type-2 pulse-train irradiation on a fresh ABC region, then acquire a dark-field image of the same region after the train before recording the diffraction pattern, and compute F_ABA with a calibration simulated for the actual layer number (about L = 7), using unilluminated ABC and ABA regions of the same flake as references. If the post-train dark-field image does not show a brighter ABA domain expanding from the exposed area, or if the unilluminated control's R drifts by more than about 0.02 under repeated diffraction-pattern acquisition, the TEM partial-transition claim is not sustained.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Electron-diffraction evidence for the laser-pulse-induced ABC-to-ABA transition (Sec. 2.2) is the change in the ratio R defined by Eq. (1), converted to an ABA fraction by F_ABA = (R - 0.005)/(0.444 - 0.005) using simulated pure ABA and ABC trilayer patterns. Two unaddressed effects make this conversion unreliable for the actual sample. First, the FLG is about 2.5 nm thick, i.e. about 7 layers, and Supplemental Fig. S5 states that for ABC stacking with layer number not a multiple of 3, the discriminating first-order peak at 4.7 nm^-1 is 'very weak' rather than absent; the pure-trilayer baseline therefore cannot be applied without an L-dependent recalibration. The measured pre-illumination R = 0.019 is already assigned F_ABA about 0.038 for a region claimed to be entirely ABC, indicating that the baseline is not known to that accuracy. Second, no dark-field image or Raman map of the same area after the pulse train is presented; the only post-train evidence is the scalar R. The single-pulse dark-field controls in Fig. 4j-k show laser-induced contrast changes that the authors attribute to rippling, so a non-stacking contribution to R (rippling, damage, contamination) is a live alternative. If the baseline shift or a laser-damage contribution accounts for the R = 0.019 to 0.144 change, the claimed partial transition is not established, and with it the 'thermal' heat-build-up conclusion loses its TEM support.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript reports experimental observations of a local, controlled transition from ABC (rhombohedral) to ABA (Bernal) stacking in few-layer graphene. Two methods are presented: Joule heating of an hBN-encapsulated FLG device, monitored by spatially resolved Raman 2D-peak mapping, and illumination with trains of femtosecond laser pulses in a transmission electron microscope, monitored by selected-area electron diffraction and dark-field imaging. The authors interpret the results as a thermally driven transition accompanied by motion of a domain wall, and they image a domain wall at roughly 9 nm resolution in dark-field mode.","tokens_in":12331,"tokens_out":4263,"duration_ms":37226,"significance":"If the central claim holds, the paper offers a practical route to locally engineer stacking order in few-layer graphene, which is relevant for creating in-plane heterostructures with tunable electronic properties. The results also provide insight into the transition mechanism, suggesting a thermally driven shear process. The paper's strengths include the combination of Raman and TEM characterization, the use of simulated diffraction patterns as a reference, and the direct dark-field visualization of a domain wall. The thermal interpretation is supported by the observation that single pulses do not induce the transition while pulse trains do, consistent with heat build-up.","major_comments":[{"comment":"The conversion of the diffraction ratio R to an ABA fraction F_ABA uses a calibration from simulated trilayer (L=3) patterns, but the experimental FLG is about 2.5 nm thick (roughly 7 layers). The supplementary text states that for ABC stacking with L not a multiple of 3, the discriminating first-order peak at 4.7 nm^-1 is 'very weak' rather than absent, so the pure-trilayer baseline cannot be applied without an L-dependent recalibration. The measured pre-illumination R=0.019 already yields F_ABA≈0.038 for a region claimed to be entirely ABC, indicating the baseline is not known to the required accuracy. An error analysis or a layer-count-adjusted calibration is needed before the change to R=0.144 can be interpreted as a partial transition.","section":"Sec. 2.2, Eq. (1), Supplemental Fig. S5"},{"comment":"The only post-train evidence for the partial transition is the scalar R obtained from selected-area diffraction; no dark-field image or Raman map of the same region after the pulse train is presented. The single-pulse dark-field controls in Fig. 4j-k show laser-induced contrast changes that the authors attribute to rippling, so a non-stacking contribution to R (e.g., from rippling, damage, or contamination) is a live alternative. Spatially resolved post-train data, or an explicit test ruling out non-stacking contributions, would be needed to substantiate the claim of a partial transition and the associated domain-wall motion.","section":"Sec. 2.2, Figs. 4d-e and 4j-k"},{"comment":"The Raman classification uses hard thresholds R<1.3 (ABC) and R>1.7 (ABA) without quoted uncertainties. The intermediate values, R between about 1.3 and 1.6, are interpreted as gradual partial transition and a shifting DW, but without error bars or a calibrated mixing model this quantitative interpretation is not firmly supported. Please provide error bars on R and a calibration for mixed stacking if the gradual-shift claim is to be made.","section":"Sec. 2.1, Fig. 2c"}],"minor_comments":[{"comment":"The phrase 'ABS FLG' should be 'ABC FLG'.","section":"Sec. 2.1"},{"comment":"The sentence describing the series resistors says 'one resistor (the ABC domain) having a resistance an order of magnitude higher than the other (the ABC domain)'; the second domain should be ABA.","section":"Sec. 2.1, thermal model"},{"comment":"Several typographical errors are present, including 'indentify' (should be 'identify') and 'sheer movements' (should be 'shear movements').","section":"Throughout"},{"comment":"The definition of R is not typeset clearly; a more explicit presentation of the ratio of summed intensities would improve readability.","section":"Eq. (1)"},{"comment":"The authors note that the radial intensity profiles 'can only serve as an approximate guide'; this caveat should be reflected when drawing quantitative conclusions from the ratio R.","section":"Sec. 2.2"}],"recommendation":"major_revision","confidential_remarks":"The TEM quantification issue is a genuine concern that should be addressed before publication. The paper otherwise presents interesting observations. I would not reject, but the authors need to either provide a layer-count-corrected calibration with error bars or soften the quantitative claims to qualitative ones."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper's central claim is credible: ABC few-layer graphene can be locally and controllably turned into ABA stacking by Joule heating and by trains of femtosecond laser pulses. The qualitative observations hold up, and the two methods cross-support each other. But the quantitative numbers, especially in the TEM section, are softer than the abstract suggests.\n\nWhat's new here is the local control. Prior work did thermal ABC-to-ABA conversion at 1300 °C or above, and Jiang et al. moved domain walls with an AFM tip. This paper adds two practical routes: resistive heating with sub-micron Raman mapping that tracks the domain wall as it shifts, and laser pulse trains with in-situ electron diffraction and dark-field imaging. The dark-field image of the DW at ~9 nm resolution is a nice technical step. The Raman maps in Fig. 2 are the strongest part: the R parameter increases with applied bias, and the spatial progression from one end of the channel to the other looks like a genuine moving boundary. The authors also deserve credit for stating that single pulses don't switch, which strengthens the heat-buildup argument.\n\nNow the soft spots. The TEM evidence for the laser-switching claim is a ratio R computed from diffraction intensities, converted into an ABA fraction using a formula calibrated from simulated trilayer patterns. But the sample is about seven layers thick, and the authors themselves note in Fig. S5 that for ABC stacking with layer number not a multiple of three the discriminating first-order peak is weak but not absent. So the calibration baseline is off. The pre-illumination R = 0.019 already yields F_ABA = 0.038 for a region supposed to be pure ABC, which tells you the conversion has no built-in zero. More importantly, there is no post-train dark-field image or Raman map of the same area; the only post-train evidence is the scalar R. The dark-field control images after single pulses show contrast changes attributed to rippling, so a non-stacking contribution is a live possibility. On the Joule heating side, the transition pressures (6–36 GPa) come from a one-dimensional thermal model with a fitted out-of-plane conductance and a constant thermal expansion coefficient, so treat those as order-of-magnitude.\n\nNone of this kills the central result. The Raman experiment is fairly convincing on its own, and the TEM results are consistent with it. But the TEM partial-transition claim is not established with the same confidence. This paper deserves serious refereeing. The right referee will ask for error bars on R and F_ABA, a spatial map after the pulse train, and a control for laser damage or rippling. Given the new capability and the clean Raman switching, I'd send it out.","headline":"Local switching of ABC to ABA is credible; the TEM quantification is not.","tokens_in":12941,"tokens_out":3560,"would_cite":true,"duration_ms":35593,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["81.05.ue","81.05.uf","78.67.Bf","68.37.-d","61.05.J-"],"model":"deepseek-v4-flash","headline":"ABC-stacked few-layer graphene can be switched to common ABA stacking locally and controllably by Joule heating or by laser-pulse trains, with the transition driven by thermal stress and propagating through a moving domain wall.","keywords":["graphene","graphite","van der Waals heterostructures","domain wall","Raman spectroscopy","transmission electron microscopy","electron diffraction","structural transition"],"falsifier":"A decisive and directly observable test is to take a flake that the paper's classifiers assign a partial conversion — for instance about 30% ABA after laser illumination — and image the same area at atomic resolution with scanning tunneling microscopy or cross-sectional annular dark-field scanning transmission electron microscopy, both established for graphene. If the true areal fraction of converted stacking disagrees with the fraction that the Raman ratio thresholds and the linear diffraction formula assign, then the empirical calibrations are mislabeling partial transitions, and the claimed stepwise switching progression and moving-domain-wall picture would not be established; agreement would confirm them.","tokens_in":11799,"feed_emoji":"🔥","tokens_out":20971,"duration_ms":154706,"temperature":0.7,"pith_summary":"This paper claims that the metastable ABC (rhombohedral) stacking of few-layer graphene can be converted, locally and controllably, into the common ABA (Bernal) stacking by two heating routes: passing a current through the device, or illuminating it with a rapid train of femtosecond laser pulses. In both cases the authors argue that the switch is thermal, that it starts at a small spot and spreads by shifting the boundary (domain wall) between the two stackings, and that the boundary can be watched while it moves. The claim matters because the two stackings have very different electronic properties — ABC trilayers host an electrically tunable band gap, while ABA trilayers behave as a semimetal — so a writable, local switch between them would allow stacking patterns to be engineered inside a single flake. The paper also reports that single laser pulses, even ones that transiently heat the lattice above 3000 K, do not trigger the transition, which places the switching time at nanoseconds or longer.","feed_headline":"Joule heat and laser pulses flip graphene's ABC stacking to ABA","feed_subtitle":"Two heating routes convert rhombohedral ABC to Bernal ABA, letting the domain wall be tracked at 9 nm resolution.","key_machinery":"The load-bearing mechanism is thermally driven, stress-mediated motion of a stacking domain wall. Under Joule heating the device is modeled as two resistors in series, with the ABC domain more resistive and therefore hotter; because thermal expansion is restricted, compressive stress builds at the wall and pushes it once a threshold pressure is reached — the paper estimates 6–36 GPa across the three observed transition steps, in the same range as a previously reported value of about 22 GPa. Stacking identification rests on two discriminators: the Raman 2D-peak ratio $R = I_3/I_2$ after fitting three Lorentzians ($R < 1.3$ for ABC, $R > 1.7$ for ABA), and the electron-diffraction ratio of first-order to second-order peak intensities, which is converted to an ABA fraction by the linear formula $F_{\\mathrm{ABA}} \\approx (R - 0.005)/0.444$ (first-order diffraction peaks are nearly extinct for ABC stacking). Dark-field imaging that selects the (1-100) Bragg peak makes the ABA domain bright and the ABC domain dark, allowing the wall to be located at about 9 nm resolution.","core_discovery":"The paper's central claim is that ABC-stacked few-layer graphene can be switched to ABA stacking locally and controllably by Joule heating or by laser-pulse illumination, with the transition driven by compressive stress and progressing by domain-wall motion rather than by a single simultaneous shift of all layers. In the Joule-heating experiment on an hBN-encapsulated ABC flake, Raman mapping of the 2D peak — using the intensity ratio $R = I_3/I_2$ of fitted Lorentzian sub-peaks, with $R < 1.3$ indicating ABC and $R > 1.7$ indicating ABA — shows conversion starting at one end of the channel at 3.5 V and spreading until the whole device becomes ABA at 9 V. In the electron microscope, single 790 nm pulses at fluences up to about 150 mJ/cm$^2$ produced no transition, while a 120 ms train at 100 kHz repetition rate converted part of an ABC region to ABA, showing that heat accumulation over many pulses is required; a linear calibration from simulated diffraction patterns, $F_{\\mathrm{ABA}} \\approx (R - 0.005)/0.444$, converts the measured first-to-second-order peak ratio into an ABA fraction and yields a partial transition with about 32% ABA after illumination. Dark-field imaging selecting the (1-100) diffraction spot renders ABA bright and ABC dark, resolving the domain wall at about 9 nm. The authors conclude that both switching routes are thermal, that the wall moves as conversion proceeds, and that the transition takes nanoseconds or longer.","pith_inferences":["If the stress-threshold picture is correct, interrupting a Joule-heating sweep at an intermediate bias should freeze a partially converted pattern in place, letting stacking domains be written in arbitrary shapes rather than only fully converted; this is a testable extension the paper does not perform.","The transition's irreversibility blocks stroboscopic pump–probe measurement, as the authors note, but a two-stage scheme — a train of heating pulses followed by a delayed electron probe — could still bracket the switching time, turning their single-pulse null result (a lower bound in the nanosecond range) into an actual measurement.","Because the estimated domain-wall pressures of 6–36 GPa straddle the previously reported ~22 GPa threshold, purely mechanical stress, such as bending or pressing the flake, should also drive ABC-to-ABA conversion at room temperature without global heating, connecting this work to earlier tip-based manipulation of stacking solitons."],"forward_implications":["Stacking order becomes a writable, local degree of freedom: Joule heating switches regions on the micrometer scale, while laser-pulse trains switch areas of roughly 20–40 μm.","The transition is thermal and accumulative: single femtosecond pulses that transiently exceed 3000 K do not switch stacking, whereas a 100 kHz train that builds up heat does, placing the transition on a nanosecond-or-longer timescale.","Conversion proceeds by domain-wall motion: the wall shifted progressively along the channel during Joule heating until the entire device became ABA, and was imaged directly at about 9 nm resolution in dark-field electron microscopy.","Diffraction ratios quantify partial conversion: the formula $F_{\\mathrm{ABA}} \\approx (R - 0.005)/0.444$ assigns about 32% ABA stacking after laser illumination, showing that the switch can be partial rather than all-or-nothing.","Dark-field imaging tracks the wall at roughly 9 nm resolution, an order of magnitude better than the about 1 μm Raman maps, opening nanoscale study of stacking solitons during a thermally driven transition."],"supporting_citations":[{"why":"Establishes the precedent that rhombohedral graphite transforms to Bernal stacking under heating above 1300 °C, the bulk thermal transition that the paper reproduces locally in few-layer devices.","marker":"[2]"},{"why":"Supplies the Raman 2D-peak analysis used to map ABC versus ABA stacking and the benchmark that no conversion was seen below 800 °C in tri- and tetralayer graphene.","marker":"[4]"},{"why":"Provides the first-principles Raman fingerprint of rhombohedral graphite that underpins identifying ABC domains from 2D-peak spectra.","marker":"[5]"},{"why":"Shows that stacking domain-wall solitons in bi- and trilayer graphene can be moved, erased, and split by an AFM tip, the manipulation precedent for the controlled wall shift reported here.","marker":"[19]"},{"why":"Supplies the Joule-heating protocol for hBN-encapsulated graphene and the heat-diffusion model used to estimate device temperatures and domain-wall pressures.","marker":"[22]"},{"why":"Demonstrates that an external influence (molecular adsorption) can transform trilayer graphene stacking, the basis for the device design with a small intentional ABA end section.","marker":"[23]"},{"why":"Gives the ~22 GPa transformation pressure for rhombohedral graphite against which the paper's 6–36 GPa domain-wall pressure estimates are compared.","marker":"[25]"},{"why":"Describes the femtosecond-resolved transmission electron microscope that enabled in-situ laser illumination with diffraction and dark-field imaging.","marker":"[26]"}],"fun_headline_variants":["Heat flips graphene's ABC stacking to ABA","Joule heat and laser pulses switch graphite stacking","ABC to ABA: heating drives domain wall motion","Local heating turns rhombohedral graphite into Bernal","Thermal pulses controllably re-stack graphene layers"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the two measuring recipes used to tell the two stackings apart — the shape of the scattered-light peak in one experiment and the brightness ratio of electron-diffraction rings in the other — are accurate enough to identify a partly converted sample; if either recipe mislabels a partially converted region, the claimed stepwise switching and the moving boundary between the two stackings would not be established.","fun_headline_variants_meta":{"raw":{"variants":["Heat flips graphene's ABC stacking to ABA","Joule heat and laser pulses switch graphite stacking","ABC to ABA: heating drives domain wall motion","Local heating turns rhombohedral graphite into Bernal","Thermal pulses controllably re-stack graphene layers"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000253,"raw_usage":{"total_tokens":1626,"prompt_tokens":1070,"completion_tokens":556,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":686,"completion_tokens_details":{"reasoning_tokens":483}},"tokens_in":686,"tokens_out":556,"duration_ms":5993,"temperature":1.0,"reasoning_tokens":483,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:49:49.685916+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A decisive and directly observable test is to take a flake that the paper's classifiers assign a partial conversion — for instance about 30% ABA after laser illumination — and image the same area at atomic resolution with scanning tunneling microscopy or cross-sectional annular dark-field scanning transmission electron microscopy, both established for graphene. If the true areal fraction of converted stacking disagrees with the fraction that the Raman ratio thresholds and the linear diffraction formula assign, then the empirical calibrations are mislabeling partial transitions, and the claimed stepwise switching progression and moving-domain-wall picture would not be established; agreement would confirm them.","supporting_citations":[{"cited_title":"Shabalin, Ultra-High Temperature Materials","cited_arxiv_id":null,"evidence_quote":"Establishes the precedent that rhombohedral graphite transforms to Bernal stacking under heating above 1300 °C, the bulk thermal transition that the paper reproduces locally in few-layer devices."},{"cited_title":"Thermal expansion coefficients of graphite crystals,","cited_arxiv_id":null,"evidence_quote":"Supplies the Raman 2D-peak analysis used to map ABC versus ABA stacking and the benchmark that no conversion was seen below 800 °C in tri- and tetralayer graphene."},{"cited_title":"Convergent beam electron holography for analysis of van der Waals heterostructures,","cited_arxiv_id":null,"evidence_quote":"Provides the first-principles Raman fingerprint of rhombohedral graphite that underpins identifying ABC domains from 2D-peak spectra."}],"review_version":1}