{"id":"3156757f-94a6-4f8c-b59c-37e7dc7db564","arxiv_id":"2607.25172","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Parent valley character plus stacking symmetry, not chemistry, organizes bandwidth scaling and topological bands across more than 1,000 twisted semiconductor moiré structures.","lead":"By computing over 1,000 moiré band structures for 43 twisted semiconductor materials, the authors built a catalogue of flat and topological bands. They find that the parent material's valley character and bilayer stacking, rather than chemical composition, decide when twisting creates useful topological minibands.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Taxonomy rests on unvalidated transferability of the TAPW/MLFF pipeline to all 43 materials; a stratified full-DFT validation would settle whether the valley hierarchy is real or a pipeline artifact.","rationale":"The reader's weakest assumption—that the TAPW tight-binding Hamiltonians from GSFE/MLFF-relaxed structures faithfully reproduce low-energy moiré physics for all 43 materials—is exactly the load-bearing point I would stress. The paper's central claim is a predictive taxonomy, and a taxonomy built on approximate methods can only be as strong as the transferability of those methods. The absence of per-material validation in the main text, together with the dependence on an 'In preparation' method reference, makes this the most important uncertainty. I agree with the reader's conditional verdict: the taxonomy is plausible and internally consistent, but it should not be accepted as established until the pipeline is validated on a representative sample. The concrete test I propose is feasible because moderate twist angles (5–7°) yield supercells that are still accessible to direct DFT for many of the materials in the database. I do not see a reason to reject the paper outright; the concern is about the strength of the evidence, not about internal inconsistency. Thus the reader's CONDITIONAL verdict remains appropriate, and I would not change it.","tokens_in":11333,"tokens_out":3927,"duration_ms":39980,"concrete_test":"Select a stratified sample of 6–9 bilayers spanning Γ-, K-, and M-valley systems, including at least one Janus termination (e.g., BiTeCl) and one AA/AB pair. For each, at the smallest commensurate twist angle with a tractable supercell (θ≈5–7°), perform full DFT structural relaxation and band-structure calculations on the moiré cell, then compare the parent valley character, the lowest-miniband bandwidth W(θ), and the Chern/Z2 invariants against the TAPW/MLFF pipeline results. If any Γ/K/M classification flips, if any bandwidth differs by more than ~30%, or if any topological invariant changes, the reported valley hierarchy is not yet robust; if all match, the transferability concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central taxonomy—that moiré bands are organized by parent valley character and stacking symmetry—depends on every material-level label and moiré band feature produced by one pipeline: GSFE/MLFF relaxation followed by TAPW tight-binding bands. The benchmark support cited (Refs. 22, 44, 52) is concentrated on MoTe2/WSe2, and the TAPW implementation itself is given only as Ref. [54], 'In preparation.' No per-material convergence or validation against full DFT or experiment is shown for the broader 43-material set. If the MLFF relaxation or TAPW interlayer hybridization transfers imperfectly, two things can happen: (a) the parent valley character is misassigned (Γ vs K vs M ordering changes), or (b) the moiré potential depth and hence the bandwidth scaling class is misestimated. Either error propagates directly into the purported universal hierarchy and into the Chern/Z2 assignments. In addition, the W(θ)∝θ² claim in the 'Universal scaling and valley-controlled hierarchy' section is asserted without a quantitative fit or a no-potential baseline, so it is not independently checkable as presented. The pipeline-validation gap is the more load-bearing issue because all three valley classes—Γ, K, and M—are claimed to follow distinct behaviors, and a single class misassignment would weaken the taxonomy's predictive claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a high-throughput computational framework that combines screening of experimentally known monolayers, GSFE/MLFF-based structural relaxation, and truncated atomic plane-wave (TAPW) tight-binding simulations to construct a database of moiré band structures for 43 twisted semiconductors and 91 bilayer prototypes, yielding over 1,000 angle-resolved band structures. The central claim is that the low-energy moiré electronic structure is organized primarily by the parent monolayer's valley character together with stacking symmetry: Γ-valley systems show a nearly quadratic twist-angle scaling W(θ)∝θ² attributed to folding-dominated kinetics; K-valley systems exhibit stacking-controlled isolated valley Chern minibands; M-valley systems are material-specific and display non-symmorphic kagome-like or quasi-one-dimensional behavior. The paper also reports topology (Chern, Z₂) and quantum geometry for selected minibands, and demonstrates that surface termination in Janus BiTeCl can switch the relevant valley character. The work is framed as a predictive taxonomy and is accompanied by a public web database.","tokens_in":11581,"tokens_out":4340,"duration_ms":45561,"significance":"If the reported hierarchy is robust, this is a useful organizing principle for a rapidly growing field, and the database itself is a valuable community resource. The paper makes explicit, falsifiable predictions: for example, AA-stacked K-valley systems should host isolated valley Chern minibands, certain Γ-valley AB stackings should host robust Z₂ phases, and Janus termination can switch valley character. The authors should be credited for constructing a large, publicly accessible dataset with band structures and topological data, and for grounding the claim in a systematic enumeration of stacking configurations. However, the significance is conditional on the transferability of the TAPW/MLFF pipeline to the full 43-material set; the current evidence is strong mainly for MoTe₂/WSe₂. The quantitative scaling law and the class sizes are also not yet established from the data as presented, which limits the degree to which the taxonomy can be accepted as predictive.","major_comments":[{"comment":"The central claim W(θ)∝θ² for Γ-valley systems is asserted as 'consistent with a folding-dominated regime' but no quantitative fit is provided. Fig. 2 shows heatmaps only; the text does not report power-law exponents, residuals, or a no-potential folding baseline for individual compounds. Because this scaling is a headline result, please provide log-log fits per material with quoted exponents and goodness-of-fit measures, and overlay a folding-only reference curve in Fig. 2 or a supplement figure. Without this, the reader cannot distinguish a genuine quasi-universal law from an approximate trend.","section":"Universal scaling and valley-controlled hierarchy"},{"comment":"The entire database relies on GSFE/MLFF relaxation followed by TAPW tight-binding Hamiltonians. The TAPW implementation is cited as Ref. [54], 'In preparation,' and the method appears to have been benchmarked mainly on MoTe₂ and WSe₂ via Refs. [22,44,52]. No per-material validation against direct DFT supercell calculations, independent continuum models, or experiment is shown for the broader set of halides and chalcogenides. A stratified validation is needed: at least one representative per valley class (Γ, K, M, Q) with a full-DFT or alternative moiré calculation at a small twist angle, checking parent band-edge ordering, moiré bandwidth, and topological invariants. If the pipeline misassigns valley character or misestimates interlayer hybridization in less-studied compounds, the reported valley hierarchy and the associated Chern/Z₂ assignments could be properties of the pipeline rather","section":"Computational workflow"},{"comment":"The paper argues for a 'universal' valley-and-stacking hierarchy, but explicitly prioritizes high-symmetry valleys and sets aside off-symmetry Q-valley systems, describing their behavior only as 'intricate' and driven by 'sensitive interplay.' The manuscript does not state how many of the 43 monolayers or 91 prototypes fall into the Q-valley class, nor whether the remaining materials actually obey the taxonomy. Without this information, the claimed organizing principle is not testable for a potentially large fraction of the database. Please report the class sizes and provide a systematic account of Q-valley systems, or explicitly restrict the universality claim to high-symmetry-valley systems.","section":"Valley classification"},{"comment":"For M-valley systems the text states they 'host a sophisticated topological landscape governed by emergent non-symmorphic symmetries' and 'manifest unique kagome or quasi-one-dimensional physics,' citing Refs. [49,50], but no M-valley moiré band structure, symmetry analysis, or topological quantity from the database is shown in this work. Since the taxonomy includes M-valley systems as a distinct class, at least one representative M-valley example from the new database should be presented, or the text should clearly attribute this statement to prior literature rather than to the current high-throughput results.","section":"Twist-induced topology from valley structure"}],"minor_comments":[{"comment":"Several strings such as '/uni0000001a/uni00000011/...' appear in the figure caption and in the text near Fig. 2. This appears to be an encoding artifact and must be fixed before publication.","section":"Fig. 2 caption and surrounding text"},{"comment":"Ref. [54] is cited as 'In preparation' for the TAPW implementation. This is not verifiable; please either provide the full technical details in the Supplementary Information or update the reference once available.","section":"References"},{"comment":"The Fig. 2 caption states that absence of symbols identifies minibands that 'lack energetic isolation' in K- and Γ-valleys, but the numerical criterion for energetic isolation is not defined in the main text. Please specify the band-isolation energy threshold and the twist-angle sampling grid used for the database.","section":"Methods / Definition of 'energetic isolation'"},{"comment":"The text uses both 'quadratic scaling' and 'nearly quadratic' for W(θ). Please define the quantitative range of twist angles over which this scaling is claimed, and clarify how 'nearly' is assessed.","section":"Scaling wording"},{"comment":"The Data Availability statement refers to Supplementary Information [56], but the reference is a generic placeholder. Please point to the specific sections and file names of the database.","section":"Data availability"}],"recommendation":"major_revision","confidential_remarks":"The paper is a large computational survey with a potentially valuable public database. The main concern is whether the taxonomy is an artifact of an insufficiently validated pipeline; the requested stratified validation and quantitative scaling fits are essential. If the authors supply those, the work could be suitable for publication. The 'In preparation' reference for the central method is also an issue for a journal submission."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"What you should know: this is a big high-throughput computation catalog, not a deep dive on any single material. The database itself is the contribution: 43 monolayers, 91 bilayer prototypes, over 1,000 angle-resolved band structures, with an online repository. That is a real resource for the twistronics community, and the organizing taxonomy—moiré bands controlled by parent valley character plus stacking symmetry—is plausible and consistent with the well-studied MoTe2/WSe2 cases. The Janus termination example (BiTeCl) is a nice illustration of how surface chemistry can shift valley character and change topology.\n\nWhat it does well: it turns a case-by-case literature into a screening capability. The specific predictions (K-valley AA stacking gives isolated Chern bands, certain Γ-valley stackings give Z2 = 1) are not baked in by construction; they follow from the calculations. The paper is honest about scope: Q-valley systems and M-valley universality are explicitly excluded or qualified. The references to the underlying methods are appropriate, and the data availability statement is concrete.\n\nSoft spots, in order of weight. First, the pipeline transferability. The TAPW implementation is only given as Ref. [54], \"In preparation,\" and the validation cited is concentrated on MoTe2/WSe2. For a paper claiming a universal hierarchy across 43 materials, the absence of any per-material or stratified validation against full DFT or experiment is a genuine gap. If the MLFF relaxation or the tight-binding construction misassigns the parent valley or misestimates the moiré potential depth for even one material class, the taxonomy loses its predictive force. Second, the headline W(θ) ∝ θ² scaling is asserted as \"consistent with a folding-dominated regime\" but no fits, residuals, or a no-potential baseline are shown. That is a testable quantitative claim, and the paper does not provide the test. It might hold, but as presented it is a hand-wave. Third, the classification-by-valley-then-reporting-valley-behavior is mildly circular as an organizing statement, though the specific topological predictions escape that circularity.\n\nNone of this kills the paper. The database is valuable enough that the field would benefit from having it vetted and corrected. But the universal scaling claim should be either quantified or softened, and the validation gap needs to be addressed—ideally with a supplementary section showing convergence checks and benchmark comparisons for a representative subset of non-TMD materials.\n\nVerdict: send to peer review, but the referee should demand the fits and the validation. If those come through, this becomes a standard reference for twisted semiconductor screening.","headline":"A genuinely useful materials database with a plausible valley-based taxonomy, but the central scaling law rests on assertion rather than fits, and the pipeline transferability is the weak link.","tokens_in":12142,"tokens_out":1888,"would_cite":true,"duration_ms":21623,"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":"The low-energy moiré bands of twisted semiconductors are governed by the valley character of the parent band edge together with stacking symmetry.","keywords":["twisted bilayer semiconductors","moiré bands","valley character","topological minibands","Chern number","Z2 index","high-throughput computation","Janus bilayers"],"falsifier":"Measure the bandwidth versus twist angle for three Γ-valley compounds classified as quadratic using a method that does not assume the truncated atomic plane-wave tight-binding model (e.g., full DFT on small commensurate cells); if the actual exponent deviates significantly from 2, the scaling rule fails. For K-valley AA compounds predicted to host isolated |C_K| = 1 minibands, an independent probe such as ARPES or exact diagonalization that finds no isolated Chern band would falsify the stacking-hybridization mechanism.","tokens_in":11164,"feed_emoji":"🌀","tokens_out":5607,"duration_ms":50942,"temperature":0.7,"pith_summary":"This paper tries to establish that the wide variety of moiré minibands in twisted semiconductors follows a simple hierarchy: the momentum-space location of the parent monolayer's band edge (the valley) plus the stacking symmetry of the twisted bilayer determines whether the bands are flat, how their width scales with twist angle, and whether they carry nontrivial topology. To back this up, the authors built a high-throughput pipeline that relaxes twisted structures and computes moiré electronic structure for 43 experimentally realized monolayers and 91 bilayer prototypes, yielding over 1,000 angle-resolved band structures. The payoff is a predictive taxonomy: Γ-valley systems show nearly quadratic bandwidth scaling with twist angle, K-valley systems can host isolated valley Chern minibands depending on stacking, and M-valley systems are idiosyncratic with kagome-like physics. If correct, this gives materials designers a rule of thumb for choosing parent semiconductors that will produce flat topological moiré bands.","feed_headline":"Valley character rules the flat bands of twisted semiconductors","feed_subtitle":"A 43-material census shows which parent valleys yield flat topological bands, a shortcut for designing twisted-matter devices.","key_machinery":"The central object is the valley character of the parent monolayer band edge — the high-symmetry momentum (Γ, K, M, or off-symmetry Q) where the valence or conduction band extremum sits — combined with the stacking symmetry (parallel AA vs antiparallel AB) of the twisted bilayer. The argument is carried by a high-throughput computational pipeline: structural relaxation via generalized stacking-fault energy surfaces and machine-learned force fields, followed by construction of first-principles tight-binding Hamiltonians and moiré band calculations using a truncated atomic plane-wave method, which yields bandwidths, Chern numbers, Z₂ indices, and quantum geometry without fitting.","core_discovery":"The central discovery is that the low-energy moiré electronic structure of twisted semiconductors is organized primarily by the valley character of the parent band edge together with stacking symmetry. In Γ-valley systems, miniband width follows a nearly quadratic twist-angle scaling W(θ) ∝ θ², reflecting a folding-dominated kinetic-energy scale. In K-valley systems, parallel (AA) stacking permits spin-aligned interlayer tunneling that lifts valley degeneracy and yields isolated minibands with nonzero valley Chern numbers |C_K| = 1, whereas antiparallel (AB) stacking preserves inversion symmetry and suppresses such isolation. M-valley systems are material-specific, governed by non-symmorphic","pith_inferences":["A testable extension is to use the valley-and-stacking rule as a screening criterion: given a candidate monolayer, reading off its band-edge valley from a standard DFT band structure would predict whether flat topological minibands are likely at small twist angles, without running a full moiré calculation.","The quadratic W(θ) ∝ θ² scaling for Γ-valley systems suggests a simple folding picture in which the moiré potential plays a minor role; if so, even non-twisted periodic modulations that fold the Γ point would produce similarly flat bands, a connection the paper does not explore.","The surface-termination knob in Janus bilayers points toward a general design strategy: asymmetric termination can deliberately move band extrema off high-symmetry points, turning a systematic Γ-valley material into a complex Q-valley one, and vice versa, which may be used to toggle between flat-band and dispersive regimes in the same compound."],"forward_implications":["If the valley-and-stacking hierarchy holds, Γ-valley twisted semiconductors are the most predictable class: their miniband widths shrink roughly as θ², and certain stacking configurations harbour Z₂ = 1 insulating minibands over wide twist-angle windows.","K-valley homobilayers with parallel (AA) stacking become the natural platform for isolated valley Chern minibands, since AA stacking allows spin-aligned interlayer tunneling; AB-stacked K-valley bilayers are expected to remain topologically trivial in the topmost minibands.","M-valley systems will not fit a single-scaling description; their moiré bands are dominated by non-symmorphic symmetries and can realize kagome or quasi-one-dimensional physics, requiring a different classification toolbox.","In Janus bilayers, choosing which surface faces the partner layer (e.g., Te–Te vs Te–Cl vs Cl–Cl in BiTeCl) can flip the parent valley character and thereby switch the resulting moiré topology, providing an experimental tuning knob.","Quantum-geometry screening of the database identifies candidates for fractionalized phases — AA-stacked MoTe₂ among Chern systems and AB-stacked BiTeI, MgBr₂, CaI₂ among Z₂ systems — whose flat bands are more Landau-level-like."],"fun_headline_variants":["Valley character decides which twisted semiconductors host flat bands","43 twisted semiconductors reveal valley rule for flat topological bands","Stacking and valley identity pick flat bands in 43 twisted materials","Γ, K, or M valley sets the fate of twisted semiconductor bands","Twisted-matter census: valley character orders 1000+ moiré bands"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The entire classification assumes that the machine-learning-relaxed structures and the truncated atomic plane-wave tight-binding Hamiltonians built from them faithfully represent the true low-energy moiré physics of all 43 materials; if the relaxation or the tight-binding approximation misassigns a valley or hybridization in any compound, its category in the hierarchy would be an artifact of the pipeline.","fun_headline_variants_meta":{"raw":{"variants":["Valley character decides which twisted semiconductors host flat bands","43 twisted semiconductors reveal valley rule for flat topological bands","Stacking and valley identity pick flat bands in 43 twisted materials","Γ, K, or M valley sets the fate of twisted semiconductor bands","Twisted-matter census: valley character orders 1000+ moiré bands"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000179,"raw_usage":{"total_tokens":1157,"prompt_tokens":781,"completion_tokens":376,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":525,"completion_tokens_details":{"reasoning_tokens":286}},"tokens_in":525,"tokens_out":376,"duration_ms":4409,"temperature":1.0,"reasoning_tokens":286,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T03:11:40.546968+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the bandwidth versus twist angle for three Γ-valley compounds classified as quadratic using a method that does not assume the truncated atomic plane-wave tight-binding model (e.g., full DFT on small commensurate cells); if the actual exponent deviates significantly from 2, the scaling rule fails. For K-valley AA compounds predicted to host isolated |C_K| = 1 minibands, an independent probe such as ARPES or exact diagonalization that finds no isolated Chern band would falsify the stacking-hybridization mechanism.","supporting_citations":[],"review_version":1}