{"id":"06a2d8ee-24c4-4f3f-841f-37bc47f728c5","arxiv_id":"1908.04595","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":8.0,"correctness_risk":"low","formal_verification":"none","parameter_count":2,"one_line_summary":"Local twist-angle maps in magic-angle twisted bilayer graphene reveal 0.1-degree variations and gradients that generate unscreened electric fields and bulk quantum Hall edge states.","lead":"This paper maps how the twist angle between two graphene layers varies across real magic-angle devices, using a scanning magnetic sensor. It finds variations of about 0.1 degrees, large enough to alter the electronic state and create unexpected currents in the bulk of the sample.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Twist-angle conversion assumes a rigid n_s2 ∝ θ² relation; relaxation and heterostrain can shift the p/n LL midpoint and masquerade as θ gradients, so the gradient-driven QH picture rests on an uncalibrated conversion.","rationale":"The reader's weakest_assumption is exactly the rigid uniform-twist conversion, and the stress-test agrees that this is the load-bearing step. The paper's conclusions about unscreened electric fields and bulk quantum-Hall edge states are derived from spatial gradients of the extracted θ(r); any systematic bias in the conversion affects the gradients directly, not just the absolute angle. The concern is not that the measured p/n LL midpoint is noisy—the claimed precision is well supported—but that the mapping from midpoint to θ assumes no relaxation or strain corrections, despite the paper's own citations that both are strong band-structure modifiers. This is a model-accuracy issue rather than an internal inconsistency, and it is testable in the continuum model the paper already uses. A computational calibration with realistic strain and relaxation would settle whether the 0.1° span and the gradient networks are uniquely twist-angle phenomena or partly strain artifacts. Because the reader already assigned CONDITIONAL for this same assumption plus the lack of public data, the verdict remains UNCHANGED; if the proposed test passes, ACCEPT would be justified, and if it fails, the maps would need a strain-aware reinterpretation.","tokens_in":30300,"tokens_out":4855,"duration_ms":58727,"concrete_test":"Use the SI11 continuum Hamiltonian, augmented with realistic in-plane lattice relaxation and a uniform heterostrain of order 0.1%, to generate dispersive-band Landau-level fans at B ≈ 1.2 T for a known smooth θ(r) landscape. Apply exactly the paper's p/n midpoint extraction and compare the recovered θ(r) with the input geometric θ(r). If strain shifts the recovered midpoint by more than about 0.01° in equivalent θ (five times the claimed relative precision), the reported gradient network and the resulting bulk edge-state picture cannot be attributed uniquely to twist angle without a strain-aware deconvolution.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing step is the conversion in SI6 from the p/n dispersive-band LL midpoint to a local twist angle through n_s2 = 8θ²/(√3 a²), a rigid uniform-twist relation. The paper itself cites lattice relaxation within one supercell as significantly modifying the band structure (refs 12,13) and heterostrain as having strong predicted effects (refs 14,15), but the extraction routine applies no correction for either. For the midpoint to equal n_s2 exactly, the p and n dispersive LL fans must be particle-hole symmetric and their appearance density must be controlled solely by θ. Relaxation, uniaxial/biaxial heterostrain, and stacking-fault networks shift LL energies and the local density at which dispersive LLs appear without changing the geometric moiré period; these shifts would be read out as apparent θ(r) variations and gradients. Because the central claims—unscreened in-plane electric fields, bulk quantum-Hall edge states, and percolation—are driven by ∇θ(r), a strain/relaxation bias in the θ maps transfers directly to the main physical conclusion. The quoted ±0.002° accuracy is measurement precision, not model accuracy, and no independent calibration of the θ extraction against relaxed band structures or a second local probe is provided.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents scanning nanoSQUID-on-tip imaging of Landau levels in two hBN-encapsulated magic-angle twisted bilayer graphene devices. From the spacing between p-type and n-type dispersive-band Landau level peaks, the authors extract maps of the local twist angle θ(r) via the rigid-twist relation between the flat-band density n_s2 and θ. They report that even transport-quality devices with correlated insulators and superconductivity contain local twist-angle variations spanning up to about 0.1°, networks of abrupt jumps, and mean gradients of order 0.05°/µm. They further argue that these gradients create unscreened in-plane electric fields, produce compressible/incompressible strip patterns and bulk quantum Hall edge states, and may influence correlated and superconducting phases. The claims are supported by line scans, tomographic movies, two-device consistency, and self-consistent electrostatic and Landau level calculations.","tokens_in":30564,"tokens_out":3840,"duration_ms":45444,"significance":"If the central interpretation holds, the work constitutes an important advance: it directly images local twist-angle disorder in realistic MATBG devices, provides a quantitative link between local θ(r) and quantum Hall structure, and proposes a concrete mechanism—twist-angle-gradient-induced built-in electric fields—that could affect transport and the correlated phase diagram. The experimental methodology is original and carefully executed, with long-duration tomographic acquisitions, spatial resolution near 50 nm, and machine-readable movies. The error analysis in SI6 is a real strength, and the paper is honest about regions where no MATBG physics is observed. The percolation interpretation of global transport in devices with large non-MA areas is falsifiable and already consistent with the contrasting device A/B behavior. The main reservation is not circularity—the θ extraction does not fit the theoretical model—but model bias in the conversion from measured LL spacings to local twist angle.","major_comments":[{"comment":"The entire θ(r) map rests on the relation n_s2 = 8θ²/(√3a²), which is the rigid uniform-twist result, and on the assumption that the midpoint of the p and n dispersive LL peaks directly gives n_s2(r). The manuscript itself notes in the introduction that in-plane relaxation within a supercell significantly modifies the band structure (refs 12,13) and that heterostrain has strong predicted effects (refs 14,15), yet no correction or calibration for either is applied. Because the paper's central physical consequences—unscreened electric fields, bulk QH edge states, and percolation—are driven by gradients of θ(r), any strain- or relaxation-induced shift of the apparent local n_s2 would transfer directly into the reported θ spans and gradients. I ask for a quantitative estimate of this systematic error: for example, a relaxed-band or heterostrained LL calculation showing how much the p/n midpoint moves relative to the geometric moiré density, or an independent local measurement of the moiré period in the same devices.","section":"SI6, twist-angle conversion"},{"comment":"The midpoint extraction assumes that the p and n dispersive LL fans are symmetric around the charge neutrality point and that all symmetry-breaking effects can be absorbed into a common midpoint shift. However, Fig. 2a displays a pronounced p-n asymmetry and position-dependent toggling between 4-fold and 8-fold LL degeneracies, which indicates that the single-particle band structure is not rigidly symmetric at the local level. The authors should justify explicitly why the lower LL midpoint remains an unbiased estimator of n_s2(r). A concrete test would be to compare θ(r) extracted from two different LL indices, or to compute the p/n LL midpoint in the SI11 model with a finite heterostrain included, and show that the difference is negligible relative to the 0.1° span claimed.","section":"Fig. 2a and SI6"},{"comment":"The conversion from gate voltage to density uses a single global backgate capacitance C_bg determined from transport, while the θ(r) maps cover areas near bubbles and over a device with spatial topography. Local variations in hBN thickness or gate geometry—most plausibly in the vicinity of the bubbles explicitly excluded from the maps—would change the local C_bg and could appear as apparent θ(r) variations. The paper does not provide a spatial calibration of C_bg or an estimate of its maximum variation over the mapped regions. Please add an upper bound on ΔC_bg/C_bg and show that it contributes negligibly to the reported θ span and to the gradient maps in Figs. 3c and 3g.","section":"SI2 and Fig. 3"}],"minor_comments":[{"comment":"The quoted accuracies are inconsistent across the manuscript: the main text states relative accuracy between locations of ±0.0002° and map accuracy better than ±0.001°, while SI6 quotes ±0.002° for the movie-derived maps and also mentions ±0.001°; these numbers should be harmonized with a clear statement of which uncertainty refers to line scans, which to maps, and which is statistical versus systematic.","section":"Main text and SI6"},{"comment":"The caption labels all three band structure panels as (c); the panels should be labeled (c), (d), and (e) or the caption text should be corrected.","section":"Fig. S10 caption"},{"comment":"Many inline equations and Greek symbols are rendered as garbled escape sequences in the submitted text, particularly in the derivations of θ(r) and in SI6; the final typeset version must be carefully checked for legibility.","section":"General typesetting"},{"comment":"The abstract says 'relative precision better than 0.002°' while the main text emphasizes 'absolute accuracy of ±0.005° and relative accuracy of ±0.0002°'; please clarify precision versus accuracy in the abstract to avoid misleading readers who do not consult SI6.","section":"Abstract"}],"recommendation":"major_revision","confidential_remarks":"This is a strong experimental paper with a plausible and important central claim. The required revision is not about circularity—the extraction is not fitting the model—but about systematic bias in the θ conversion. If the authors can bound strain/relaxation effects on the p/n LL midpoint and on C_bg, the paper would be publishable. The percolation and bulk-QH observations themselves are compelling and would survive even if the absolute θ values shift."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper delivers something genuinely new: the first 2D maps of local twist angle in hBN-encapsulated MATBG transport devices, with enough precision to show that even a superconducting device has a 0.1° spread in local θ. That is a real step beyond prior STM and TEM work, which could not access whole devices or the actual transport samples. The nanoSQUID-based LL imaging is clever and carefully executed, and the consistency between two devices and with continuum-model simulations (using published interlayer couplings) gives me confidence in the core picture: local band-structure variations are large, they correlate with the presence or absence of percolation paths, and they produce bulk QH edge states. The paper is worth serious attention for those results alone.\n\nThe soft spot is the one the stress test flags. The conversion from LL peak midpoint to twist angle uses the rigid uniform-twist relation n_s2 = 8θ²/(√3 a²), with no correction for lattice relaxation or heterostrain. Those effects are cited in the paper as strongly modifying the bands, but the extraction routine ignores them. So the quoted ±0.002° accuracy is measurement precision, not model accuracy. If strain or relaxation shifts the p/n midpoint, the apparent θ(r) gradients get biased, and since the unscreened electric fields and percolation claims are driven by ∇θ, the numbers could be off. This is not a fatal flaw: the qualitative conclusion that real devices are electronically inhomogeneous, and that this matters for global transport, survives even if the label is 'effective twist angle including strain.' But the authors should either calibrate the conversion against relaxed band structures or acknowledge the ambiguity explicitly and soften the quantitative claims about gradients and field strengths.\n\nA second, minor issue is that the raw data and analysis code are not public. The supplementary information is thorough, but the core θ maps would be more credible if the processed data were shared.\n\nFor the right reader, this is a high-value paper. Anyone working on moiré materials will want to know that local twist-angle disorder is at the 0.1° level in apparently good devices, and the technique itself will influence future scanning-probe work. I would bring it to a reading group and I expect it will be cited widely.\n\nMy recommendation for peer review: send it. The central result is significant and the experiments appear solid, but it needs a referee to press on the systematic error in the θ conversion and to ask for the data. That is a normal revision path, not a reason to desk reject.","headline":"Impressive first maps of local twist angle in real MATBG devices, but the θ extraction likely absorbs strain/relaxation effects and the quantitative gradients should be treated with caution.","tokens_in":31127,"tokens_out":2957,"would_cite":true,"duration_ms":35854,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["73.43.-f","73.22.Pr","72.80.Vp"],"model":"deepseek-v4-flash","headline":"This paper maps the local twist angle across magic-angle twisted bilayer graphene devices and shows that even the cleanest devices, including a superconducting one, carry roughly 0.1° of twist-angle disorder that reshapes the quantum Hall…","keywords":["twisted bilayer graphene","magic angle","twist-angle disorder","Landau levels","quantum Hall effect","nanoSQUID","moiré percolation"],"falsifier":"Measure the same device region with an atomic-resolution technique such as STM or TEM and compare the local moiré period with the SOT-derived $\\theta(\\mathbf{r})$; if the local band structure or reconstruction shifts the relation between $n_{s2}$ and $theta$ by more than the claimed 0.002° relative precision, the inferred angle maps and gradients would change. Alternatively, a calculation showing that heterostrain or lattice relaxation changes $n_{s2}(\\theta)$ by a comparable amount would falsify the angle calibration.","tokens_in":30125,"feed_emoji":"🧲","tokens_out":5470,"duration_ms":51966,"temperature":0.7,"pith_summary":"The paper shows that the local twist angle in magic-angle twisted bilayer graphene is far from uniform: maps made with a scanning nanoSQUID reveal variations spanning close to 0.1° even in devices that show high-quality MATBG transport and superconductivity. These variations are not passive disorder. They generate unscreened in-plane electric fields that pull quantum Hall edge states into the bulk and replace the textbook edge picture with interlaced compressible and incompressible strips. The authors argue this explains why transport in MATBG rarely shows full conductance quantization, and why global magic-angle behavior can survive through percolation even when large parts of a device are not at the magic angle.","feed_headline":"Magic-angle graphene hides 0.1° twist disorder even in best devices","feed_subtitle":"SQUID imaging shows twist gradients create bulk quantum Hall states and unscreened electric fields.","key_machinery":"The central object is the nanoSQUID-on-tip (SOT), a superconducting quantum interference device at the apex of a sharp pipette that can image the tiny magnetic fields of equilibrium currents flowing in incompressible quantum Hall strips. With backgate modulation, the SOT records sharp peaks whenever an incompressible strip passes beneath the tip, and the gate voltages of these peaks give the local density of full Landau levels, from which the local flat-band density and thus the local twist angle are extracted through the rigid-twist relation $n_{s2} = 8\\theta^2/(\\sqrt{3}\\,a^2)$, where $a$ is graphene's lattice constant. A separate self-consistent electrostatic and band-structure calculation converts the measured $\\theta(\\mathbf{r})$ maps into chemical-potential, electric-field, and persistent-current maps that explain the observed bulk quantum Hall strips.","core_discovery":"By imaging the equilibrium currents carried by incompressible quantum Hall strips with a nanoSQUID-on-tip, the authors obtain tomographic maps of the local Landau levels and, from the spacing between p- and n-band peaks, maps of the local flat-band density $n_{s2}(\\mathbf{r})$ and hence the local twist angle $\\theta(\\mathbf{r})$ with relative precision better than 0.002° and spatial resolution of a few moiré periods. They find twist-angle spans of 0.13° and 0.10° in two devices, with gradients around 0.05°/µm and networks of ~0.01° jumps, and they show that these gradients produce electric fields up to ~$10^{5}$ V/m inside narrow incompressible strips and move quantum Hall edge states into the bulk. The central claim is that twist-angle disorder is a dominant, distinct disorder type in MATBG, one that changes the local band structure and substantially affects the correlated insulator and superconducting phases.","pith_inferences":["If twist-angle disorder is the dominant disorder in MATBG, then improving only the global alignment accuracy will not fix device quality; controlling local gradients or relaxation-induced jumps may matter more than the average angle.","The same nanoSQUID-based tomographic method could be applied to other moiré systems to test whether the observed ~0.01° jumps correspond to stacking-fault networks seen in transmission electron microscopy.","The gate-tunable built-in planar electric fields produced by twist-angle gradients could be deliberately engineered for band-structure tuning, photovoltaic, or thermoelectric applications—an extension the paper mentions as a direction but does not itself demonstrate.","A direct comparison between the SOT-derived $\\theta(\\mathbf{r})$ maps and atomic-resolution measurements on the same device would test whether the inferred angle gradients are intrinsic to the lattice or partly an artifact of the rigid-twist density relation."],"forward_implications":["Even devices that show superconductivity can contain substantial non-magic-angle regions; global MATBG behavior can be carried by percolating paths of favorable local twist angle.","Twist-angle gradients create unscreened in-plane electric fields that move quantum Hall edge states into the bulk, forming narrow (~50 nm) incompressible strips carrying persistent currents.","The observed $\\theta(\\mathbf{r})$ gradients and associated electric fields should affect the stability of correlated insulators, superconductivity, and magnetism in MATBG.","The usual absence of full conductance quantization in MATBG transport follows naturally from the coexistence of several different Landau levels crossing the Fermi level in the bulk.","At high enough magnetic field, when the Landau-level degeneracy exceeds the local $n_{s2}$ variation, conventional quantum Hall quantization should be restored."],"supporting_citations":[{"why":"Provides the continuum model that defines the flat bands at the magic angle and the moiré band structure used throughout the analysis.","marker":"[9]"},{"why":"Supplies the continuum-model framework for the twisted bilayer electronic structure that underlies the Landau-level calculations.","marker":"[10]"},{"why":"Demonstrates nanoscale imaging of equilibrium quantum Hall edge currents in monolayer graphene, the technique directly adapted here.","marker":"[28]"},{"why":"Introduces the scanning SQUID-on-tip sensor whose sensitivity enables detection of narrow incompressible strips.","marker":"[29]"},{"why":"Calculates lattice relaxation and band modulation in twisted bilayer graphene, motivating the expected local band-structure changes from twist gradients.","marker":"[12]"},{"why":"Observes atomic and electronic reconstruction and stacking-fault networks at the twisted interface, which the measured $\\theta$ jumps resemble.","marker":"[13]"},{"why":"Establishes correlated insulator behavior at half-filling in magic-angle graphene, the transport signature the paper's devices exhibit globally.","marker":"[1]"},{"why":"Reports unconventional superconductivity in magic-angle graphene, the global feature shown to coexist with local twist-angle disorder.","marker":"[2]"},{"why":"Describes the tear-and-stack fabrication method used to make the devices that are imaged.","marker":"[26]"}],"fun_headline_variants":["NanoSQUID maps twist disorder in magic-angle graphene","Twist gradients induce bulk quantum Hall edge states","0.1° twist variations remain in best graphene devices","Imaging twist angles reveals giant electric fields","Magic graphene's twist disorder alters correlated phases"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The maps assume that the local density needed to fill the flat bands, $n_{s2}(\\mathbf{r})$, is tied to the local twist angle by the rigid-rotation formula $n_{s2} = 8\\theta^2/(\\sqrt{3}\\,a^2)$ and that the midpoint of the p- and n-band Landau-level peaks gives $n_{s2}(\\mathbf{r})$ exactly, with negligible bias from lattice relaxation, heterostrain, charge disorder, or local variations of the backgate capacitance.","fun_headline_variants_meta":{"raw":{"variants":["NanoSQUID maps twist disorder in magic-angle graphene","Twist gradients induce bulk quantum Hall edge states","0.1° twist variations remain in best graphene devices","Imaging twist angles reveals giant electric fields","Magic graphene's twist disorder alters correlated phases"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00029,"raw_usage":{"total_tokens":1762,"prompt_tokens":1073,"completion_tokens":689,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":689,"completion_tokens_details":{"reasoning_tokens":615}},"tokens_in":689,"tokens_out":689,"duration_ms":8275,"temperature":1.0,"reasoning_tokens":615,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:37:25.399925+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the same device region with an atomic-resolution technique such as STM or TEM and compare the local moiré period with the SOT-derived $\\theta(\\mathbf{r})$; if the local band structure or reconstruction shifts the relation between $n_{s2}$ and $theta$ by more than the claimed 0.002° relative precision, the inferred angle maps and gradients would change. Alternatively, a calculation showing that heterostrain or lattice relaxation changes $n_{s2}(\\theta)$ by a comparable amount would falsify the angle calibration.","supporting_citations":[{"cited_title":"Bistritzer and A","cited_arxiv_id":null,"evidence_quote":"Provides the continuum model that defines the flat bands at the magic angle and the moiré band structure used throughout the analysis."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the continuum-model framework for the twisted bilayer electronic structure that underlies the Landau-level calculations."},{"cited_title":"Nanoscale imaging of equilibrium quantum Hall edge currents and of the magnetic monopole response in graphene","cited_arxiv_id":"1908.02466","evidence_quote":"Demonstrates nanoscale imaging of equilibrium quantum Hall edge currents in monolayer graphene, the technique directly adapted here."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Calculates lattice relaxation and band modulation in twisted bilayer graphene, motivating the expected local band-structure changes from twist gradients."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Observes atomic and electronic reconstruction and stacking-fault networks at the twisted interface, which the measured $\\theta$ jumps resemble."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes correlated insulator behavior at half-filling in magic-angle graphene, the transport signature the paper's devices exhibit globally."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reports unconventional superconductivity in magic-angle graphene, the global feature shown to coexist with local twist-angle disorder."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Describes the tear-and-stack fabrication method used to make the devices that are imaged."}],"review_version":1}