{"id":"ca54f7c3-a15f-4553-9b9e-3e4ed30fe71c","arxiv_id":"2502.05527","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A fabrication routine using surface contacts and avoiding graphene etching preserves the triangular moiré lattice in marginally twisted double-bilayer graphene.","lead":"Fabricating twisted graphene devices usually deforms the delicate rotated lattice; this paper shows which fabrication steps cause the damage and offers a gentler contact method to avoid it. The result matters because preserving the twist angle is essential for studying exotic electronic states in these materials.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central claim depends on uncalibrated s-SNOM phase contrast as a proxy for stacking order; RIE-induced loss of contrast could be a plasma/contamination artifact, and 'virtually unchanged' is asserted without quantitative comparison or the cited supporting figures.","rationale":"The reader's weakest assumption and my concern coincide: the argument depends on s-SNOM third-harmonic phase contrast being a faithful, local readout of Bernal versus rhombohedral stacking in encapsulated, strained mTDBG. Refs. 17-18 provide genuine support for infrared contrast between stacking orders in few-layer graphene, and the triangular domain pattern is a known signature of atomic reconstruction, so the approach is not implausible. However, in this manuscript the readout is used without in-situ calibration under hBN encapsulation and with moiré-scale strain, and the central causal inference from image-contrast disappearance to stacking reversion is not uniquely determined. My critique is therefore about evidence sufficiency, not about internal inconsistency or authorial intent. Because the reader already assigned CONDITIONAL on essentially these grounds, I would keep that verdict: the paper should be accepted only after the calibration/control experiment is supplied, since the central preservation claim is otherwise unverified. The concrete test above would settle whether the concern lands, and if successful it would substantially strengthen the paper's central claims.","tokens_in":6614,"tokens_out":7793,"duration_ms":83033,"concrete_test":"On a fresh mTDBG device with alignment markers, acquire s-SNOM phase images of a fixed region and, in the same region, acquire an independent stacking-sensitive map (conductive-AFM current image, or STM where the surface is accessible). Then perform the same RIE edge-etch procedure used in Fig. 3, re-acquire both maps at the same coordinates, and compare the stacking labels. If the s-SNOM triangular contrast disappears after etching while the CAFM/STM labeling remains unchanged, the paper's RIE-induced stacking-reversion interpretation is not supported. If the independent probe shows ABA/ABC domain changes exactly coinciding with the s-SNOM contrast loss, the interpretation is confirmed. The same calibration should be applied to the 2D-contact comparison to verify that preserved phase contrast corresponds to preserved stacking rather than reduced optical sensitivity after metallization.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing concern is the s-SNOM readout itself. The paper identifies Bernal and rhombohedral stacking from third-harmonic phase images (Figs. 1F, 2-4), citing refs. 17-18, but never calibrates that phase against known stacking domains in the actual encapsulated mTDBG geometry. This matters most for the etching claim: the text says that after RIE, 'the triangular domains near the etching edge disappeared and relaxed to Bernal stacking,' and the sole evidence is the loss of triangular phase contrast near the edge. RIE is not a purely mechanical step; it involves plasma exposure, resist residue, and local defect/doping generation, any of which can suppress near-field optical contrast without changing stacking order. The manuscript rules out heating and external mechanical strain, but it does not rule out these optical or chemical confounds. For the positive claim, the statement that after 2D metal contacts 'the initial triangular lattices remained virtually unchanged' (Fig. 4B) is a qualitative visual assertion; no domain-period, domain-area, or image-correlation metric is reported, and the supporting comparison images cited for 1D vs 2D contacts (Supporting Information, Figs. S2-S3) are not included in the preprint. Thus the central preservation claim currently rests on an unvalidated and unquantified optical proxy for stacking order.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports a fabrication sequence for marginally twisted double-bilayer graphene (mTDBG) devices and uses scattering-type near-field optical microscopy (s-SNOM) third-harmonic phase images to monitor the reconstructed triangular stacking-order domains through successive fabrication steps. The authors find that transferring a top hBN layer can deform the triangular lattice, that reactive-ion etching of the sample boundary is accompanied by a loss of triangular contrast near etched edges, and that a two-dimensional surface-contact scheme that avoids etching the mTDBG layer leaves the triangular lattice \"virtually unchanged.\" The conclusion is that 2D contacts are a non-invasive route to preserving twist-angle during device fabrication.","tokens_in":6847,"tokens_out":5143,"duration_ms":48853,"significance":"The practical recipe is potentially valuable: monitoring the actual stacking-order lattice before, during, and after each fabrication step is exactly what is needed for twist-angle-sensitive device research, and the proposed 2D-contact workaround is simple and plausible. The paper's strengths are its direct step-by-step s-SNOM tracking and its clear identification of the top-hBN transfer as a deformation source. However, the evidence currently presented is qualitative and does not yet quantify twist-angle preservation or rule out non-stacking explanations for the observed contrast changes, so the central claims are not established at the level one would need before adopting the recipe.","major_comments":[{"comment":"The use of s-SNOM third-harmonic phase contrast as a direct readout of Bernal versus rhombohedral stacking is not calibrated in the encapsulated mTDBG geometry of this work; refs. 17-18 do not by themselves establish that the observed phase pattern is uniquely determined by stacking order. Without a control measurement (e.g., same-area comparison with a stacking-sensitive technique such as conductive AFM, Raman, or STM), the central preservation claim rests on an unvalidated optical proxy.","section":"Results and Discussion, Figs. 1F-4"},{"comment":"The attribution of the etching-induced loss of triangular contrast solely to boundary etching is too strong: the RIE step includes CF4/O2 plasma exposure, electron-beam resist residue, and potential local doping/defect generation, all of which can suppress near-field optical phase contrast without changing stacking order. The manuscript rules out macroscopic heating and external strain, but not these local chemical or electronic perturbations. A control experiment that revisits the same region after each sub-step (lithography, resist removal, plasma etch) or a stacking-sensitive measurement near the etched edge would be needed to support the statement that the domains \"relaxed to Bernal stacking.\"","section":"Results and Discussion, Figure 3B"},{"comment":"The key positive claim that 2D contacts preserve the triangular lattice is supported only by the qualitative phrase \"virtually unchanged\" and by representative images; no domain-period, domain-area, or image-correlation metric is reported, and the comparison images (SI Figs. S2-S3) are not included in the preprint. In addition, the before and after s-SNOM images are not demonstrated to be the same physical region, so the degree of preservation cannot be verified.","section":"Results and Discussion, Figure 4B and Supporting Information Figs. S2-S3"},{"comment":"The conclusion states that etching leads to \"a shift toward zero twist-angle and partial reversion to Bernal stacking,\" but no twist angle is measured anywhere in the paper; the triangular-lattice periodicity is used only implicitly as a twist-angle proxy. If the twist angle is to be inferred from domain periodicity, the relationship should be stated and the inference calibrated; otherwise, statements about \"zero twist-angle\" are unsupported.","section":"Conclusion"}],"minor_comments":[{"comment":"Supporting Information Figures S1-S3 are referenced in the main text but are not included in the submitted manuscript; please include these figures or state their availability.","section":"Supporting Information"},{"comment":"Some figure panels in the main text lack explicit scale bars and color-scale bars for the s-SNOM phase images; adding a uniform scale bar and color scale to every panel would make the claimed before/after comparison easier to evaluate.","section":"All figures"},{"comment":"The top hBN thickness of approximately 3 nm is stated as enabling imaging, but the effect of the encapsulation layer on the s-SNOM phase contrast is not discussed; one sentence on expected attenuation or contrast would strengthen the method description.","section":"Experimental Methods"},{"comment":"The phrases \"virtually unchanged\" and \"minimal deformation\" are used without a numerical threshold; define the tolerance (e.g., less than 5% change in domain periodicity) and report the measured values.","section":"Results and Discussion, Figure 4"},{"comment":"Reference 17 is a theoretical paper on topological confinement and does not provide an experimental calibration of infrared stacking-order contrast; please either replace it with a direct experimental calibration or clarify how it supports the contrast assignment.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"I see the manuscript as a fabrication-recipe report whose usefulness depends on the missing Supporting Information and on quantitative before/after comparisons. The issues listed in the major comments are addressable with additional data, so I recommend major revision rather than rejection. I would be willing to look at a revised version that includes the SI figures and a quantitative metric for lattice preservation."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nHere's my read of the mTDBG fabrication paper. The novel bit is real: they systematically image the reconstructed triangular lattice with s-SNOM at each fabrication step—dry transfer, top hBN capping, RIE boundary etch, metal deposition—and they propose a 2D surface-contact scheme that avoids etching the twisted layer. That is practically useful, and the observation that top hBN transfer itself can deform the lattice is worth knowing for anyone building these devices.\n\nWhat it does well: the tear-and-stack plus flip procedure is described clearly; the device schematics and optical images match; they cite the relevant reconstruction and s-SNOM literature; and they are appropriately cautious about the strain sensitivity of the lattice. The method section gives enough detail to reproduce the recipe.\n\nThe soft spots are real but not fatal. First, the central claim—'virtually unchanged' after 2D contacts—rests on a single representative s-SNOM phase image, with no quantitative metric: no domain-period, domain-area, or image-correlation number, and no twist-angle extraction. Second, the attribution of lattice deformation 'solely' to RIE boundary etching is asserted after ruling out heating and external strain, but RIE also exposes the stack to plasma, resist residues, and defect/charge doping, any of which can suppress near-field optical contrast without flipping stacking order. They need a control (e.g., an unetched region far from the edge, or a sample that sees identical plasma exposure without edge removal) to back that causal claim. Third, the supporting figures S1–S3, which contain the annealing, 1D-contact, and 2D-contact comparisons, are missing from the preprint, so the most direct evidence is currently unavailable. The s-SNOM phase contrast is cited to prior work rather than calibrated in this exact encapsulated geometry; that is a minor issue if the prior assignments are solid, but a cross-check would tighten it.\n\nThe logic is coherent, the writing is honest, and there are no red flags in the reference list. This is a CONDITIONAL for me: the recipe is plausible and definitely worth a referee's time, but the paper needs quantification and a proper control for the etching artifact before I'd trust the headline claim. I'd send it to review rather than desk-reject, and I'd ask for those additions plus the SI.","headline":"A plausible, useful fabrication recipe for mTDBG, but the s-SNOM evidence is qualitative and the RIE deformation claim outruns the data.","tokens_in":7405,"tokens_out":2400,"would_cite":true,"duration_ms":23514,"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":"Marginally twisted double-bilayer graphene can be fabricated without losing its twist angle if the active layer is never etched and contacts are two-dimensional surface contacts.","keywords":["twisted graphene","s-SNOM","atomic reconstruction","superlattice","nanofabrication","double-bilayer graphene","twist angle","Bernal stacking"],"falsifier":"Repeat the boundary-etching step and check the local stacking with a technique independent of near-field optical phase, such as cross-sectional scanning transmission electron microscopy or low-temperature scanning tunnelling microscopy; if the triangular lattice vanishes after etching while the stacking order is unchanged, the claimed strain-induced relaxation to Bernal stacking is refuted.","tokens_in":6418,"feed_emoji":"🔬","tokens_out":5910,"duration_ms":87467,"temperature":0.7,"pith_summary":"Twisted double-bilayer graphene hosts correlated-electron physics only when its small twist angle stays uniform, and the paper's goal is to show that this uniformity can survive real device fabrication. Using infrared s-SNOM to image the reconstructed triangular lattice of alternating Bernal and rhombohedral domains, the authors track what happens at each step: transferring a top hBN layer deforms the lattice somewhat, reactive-ion etching of the stack boundary causes nearby domains to relax back to Bernal stacking, but depositing Cr/Au electrodes as two-dimensional surface contacts leaves the lattice virtually unchanged. The central conclusion is that the twist angle is preserved when the mTDBG layer itself is never etched, and that s-SNOM imaging provides a practical way to verify this at every step. This matters because it offers a concrete route to devices with homogeneous twist angles, which is a prerequisite for studying twist-angle-dependent physics.","feed_headline":"Twist angle survives fabrication if the graphene layer is never etched","feed_subtitle":"Near-field imaging shows boundary etching relaxes the triangular lattice, while 2D surface contacts preserve it.","key_machinery":"The central tool is third-harmonic phase imaging with scattering-type scanning near-field optical microscopy (s-SNOM), which distinguishes Bernal from rhombohedral tetralayer graphene by infrared optical contrast and maps the triangular domain lattice at tens-of-nanometre resolution. That image is what lets the authors watch the same lattice before and after each process step. The complementary fabrication principle is to avoid any etch that cuts through the mTDBG layer: a thin top hBN is selectively etched to expose the surface, and electrical contact is made by depositing metal onto that unetched graphene surface, so the strain that would otherwise relax through the etched boundary never gets released.","core_discovery":"On the paper's own terms, the discovery is that the reconstructed triangular lattice in marginally twisted double-bilayer graphene is not inherently destroyed by device fabrication; it is destroyed by specific steps, above all boundary etching. RIE of the hBN/mTDBG/hBN stack releases the intrinsic strain stored between neighbouring triangular domains, and the domains near the etched edge disappear as the local stacking relaxes to Bernal. The paper further claims that this damage can be avoided by etching only the top hBN and depositing Cr (5 nm)/Au (80 nm) contacts directly onto the intact mTDBG surface, forming two-dimensional contacts; the initial triangular lattices then remain virtually unchanged, even after a 350 °C anneal to clean the surface. Thus the method establishes that twist-angle preservation is a matter of fabrication geometry rather than a fundamental limit.","pith_inferences":["The paper's own images show that the top hBN transfer step also deforms the triangular lattice, which the authors attribute to mechanical strain; a natural next experiment would quantify how release temperature, stamp material, or hBN thickness changes that strain.","Because the reported s-SNOM data establish lattice preservation but not electrical performance, an implied but untested consequence is whether these 2D surface contacts are ohmic and low-resistance enough for transport measurements of the correlated states.","The etch-edge relaxation implies a measurable length scale, the distance from the boundary over which triangular domains are lost, that could be mapped as a function of twist angle and compared with predictions for strain relaxation in reconstructed moiré lattices."],"forward_implications":["Devices made with 2D surface contacts and no mTDBG etching should retain a homogeneous twist angle through metallization and annealing at up to 350 °C.","s-SNOM can be used as an inline inspection tool: regions with a uniform triangular lattice can be identified before etching and the etch boundary placed far enough away to keep them intact.","Boundary etching should be treated as a source of local stacking relaxation, meaning any device geometry that requires etching near the active area will degrade the flat-band physics of mTDBG.","The same preservation recipe, if correct, should extend to other strain-sensitive twisted van der Waals stacks where atomic reconstruction creates competing stacking orders."],"supporting_citations":[{"why":"Establishes that marginally twisted graphene undergoes atomic reconstruction into alternating stacking domains, the triangular lattice this paper tracks.","marker":"[11]"},{"why":"Shows intrinsic strain between adjacent triangular domains and domino-like stacking order switching, the basis for attributing etch-induced relaxation to strain release.","marker":"[14]"},{"why":"Supplies the s-SNOM technique with nanoscale infrared imaging that is used to observe the triangular lattice.","marker":"[16]"},{"why":"Demonstrates that stacking order in few-layer graphene can be distinguished by infrared optical contrast, the readout principle behind the phase images.","marker":"[18]"},{"why":"Provides the tear-and-stack method used to fabricate the marginally twisted double-bilayer graphene stacks.","marker":"[19]"},{"why":"Documents that metal deposition can induce strain and change stacking order in graphene multilayers, motivating the non-invasive contact design.","marker":"[20]"},{"why":"Defines the conventional one-dimensional edge-contact method that requires etching the graphene, which the paper avoids.","marker":"[21]"}],"fun_headline_variants":["Skip the etch: twist angle survives in double-bilayer graphene","Etching relaxes twist angle, surface contacts keep it intact","Twist angle preserved by avoiding graphene etching in fabrication","How to keep twist angle: don't etch the graphene layer","Near-field imaging finds the fabrication step that kills twist angle"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the s-SNOM third-harmonic phase image is a faithful local readout of Bernal versus rhombohedral stacking order, so that the disappearance of triangular contrast after etching really means the lattice relaxed back to Bernal stacking rather than reflecting a tip, contamination, or dielectric artefact.","fun_headline_variants_meta":{"raw":{"variants":["Skip the etch: twist angle survives in double-bilayer graphene","Etching relaxes twist angle, surface contacts keep it intact","Twist angle preserved by avoiding graphene etching in fabrication","How to keep twist angle: don't etch the graphene layer","Near-field imaging finds the fabrication step that kills twist angle"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000592,"raw_usage":{"total_tokens":2718,"prompt_tokens":833,"completion_tokens":1885,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":449,"completion_tokens_details":{"reasoning_tokens":1801}},"tokens_in":449,"tokens_out":1885,"duration_ms":12211,"temperature":1.0,"reasoning_tokens":1801,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-08T18:57:15.205093+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Repeat the boundary-etching step and check the local stacking with a technique independent of near-field optical phase, such as cross-sectional scanning transmission electron microscopy or low-temperature scanning tunnelling microscopy; if the triangular lattice vanishes after etching while the stacking order is unchanged, the claimed strain-induced relaxation to Bernal stacking is refuted.","supporting_citations":[],"review_version":1}