{"id":"e40fd2c3-2c2e-4940-8a3d-4117b2d1175d","arxiv_id":"2505.06968","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"In crossed twisted bilayer zigzag graphene nanoribbons, the number of localized edge states matches the number of symmetry-inequivalent boundary sites, and twist angle tunes four-terminal transport including 50/50 beam splitting.","lead":"This paper uses computer simulations of twisted pairs of graphene nanoribbons to show how their stacking pattern controls localized electronic states and current flow through four terminals. The results suggest that the twist angle can act as a switch, including a 50/50 beam-splitting effect.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claimed peak-site correspondence is not robust: eta, the central-miniband window, and the Appendix B exclusions are unspecified, so the central result may be a counting artifact.","rationale":"The reader's weakest assumption points to the same load-bearing concern: the peak-counting rule is unspecified in its regularization, energy window, and exclusion criteria. This is indeed the central issue because the abstract's headline finding is stated as an exact count. The paper provides multiple numerical examples that are internally consistent, but none of them tests whether the count survives changes in eta, window, or exclusion rules. The Appendix B exclusion is particularly problematic because it is applied post hoc: features that do not fit the correspondence are labeled as bulk or outside-stacking and removed. A faithful test would define these categories in advance. The number of non-equivalent sites is also not derived from a transparent algorithm, especially for incommensurate angles, so the comparison is not independently reproducible. These problems are addressable with additional numerical experiments, so the result should not be rejected outright, but it cannot be accepted as stated. The reader's CONDITIONAL verdict already captures this, so no change to the verdict is needed.","tokens_in":13460,"tokens_out":4458,"duration_ms":45625,"concrete_test":"Recompute the LDOS for theta = 50 degrees and theta = 30 degrees using eta = 1 meV, 5 meV, and 10 meV, and count peaks inside (i) a fixed window of +/- 0.1 eV around the Fermi level and (ii) the visually defined central miniband, both with and without the Appendix B exclusion rule. If the number of peaks changes under any of these choices, the claimed one-to-one correspondence is not robust. Additionally, derive the number of non-equivalent boundary sites directly from the atomic coordinates and the symmetry group of the finite cluster, and check whether that integer matches the peak count without human selection.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The abstract's central claim is that the number of LDOS peaks equals the number of symmetry-inequivalent edge sites of the stacking region (Figs. 6, 7, 9). This claim is not supported by a well-defined counting procedure. In Sec. II.C, eta is introduced only as a 'convenient regularization parameter' with no numerical value or convergence study. The LDOS peaks are counted inside an 'energy region of the central miniband' (Fig. 7(a)) without a criterion for that window. Appendix B then excludes peaks from the count when they are attributed to bulk or outside-stacking states (e.g., the -0.17 eV feature in 14-TBZGNR). Because the peak count is the dependent variable of the paper's main finding, any of these choices - a larger eta merging two nearby peaks, a wider window adding a bulk peak, or a stricter exclusion rule - can change the count. The number of non-equivalent sites is also presented as a visually determined integer (inset of Fig. 7(a)) without an algorithm for incommensurate angles, so the comparison lacks an independent operational definition. Without a convergence test and an a priori rule for the window and exclusions, the 'perfect correspondence' is not falsifiable and could be a counting artifact.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript studies twisted bilayer zigzag-graphene nanoribbons (TBZGNRs) using an extended tight-binding Hamiltonian with Slater-Koster transfer integrals, HHK recursion for real-space LDOS, and Landauer-Buettiker transport calculations for four-terminal devices. The authors analyze how in-plane stacking offsets, twist angle, and ribbon width affect localized edge states and conductance. The central claim is a 'perfect correspondence' between the number of LDOS peaks in the central miniband and the number of non-equivalent edge sites at the boundary of the stacking region (Figs. 6, 7, 9), together with the finding that twist angle is a powerful control of transport, including 50/50 electronic beam-splitting behavior.","tokens_in":13652,"tokens_out":5552,"duration_ms":54398,"significance":"If the claimed correspondence and beam-splitting control are robust, the paper would provide simple design rules for TBZGNR nanodevices: counting symmetry-inequivalent edge sites would predict the number of localized states, and the twist angle would offer a tunable knob for routing electrons. The work uses a well-established tight-binding formalism with long-range hoppings and a real-space Green function method, and it systematically explores a wider stacking-configuration space than earlier studies. The strengths are the reproducibility of the methodology (standard recursive Green functions and HHK LDOS) and the explicit treatment of symmetry, which yields plausible qualitative trends. However, the central counting claim is currently not supported by a well-defined, falsifiable procedure, as detailed in the major comments. The numerical results are internally consistent within the specified model, but the headline 'perfect correspondence' remains a visually inferred pattern rather than a demonstrated law.","major_comments":[{"comment":"The central correspondence between the number of LDOS peaks and the number of non-equivalent edge sites is not well-defined because the LDOS regularization parameter eta is never specified. In Sec. II.C, rho_j(E) is defined with the formal limit eta -> 0^+, but the practical computation uses 'a finite eta value, used as a convenient regularization parameter,' and no actual value is reported anywhere in the text or captions. Since the number of resolved peaks in Figs. 6, 7, and 9 is the dependent variable of the main claim, a peak count is meaningful only if it is shown to be stable with respect to eta (for example, by repeating the count for a range of eta values or by resolving peak positions using a converged continued-fraction calculation with a stated tolerance). Without such a convergence test, the statement in Sec. III.B that 'the number of LDOS peaks revealed in the LDOS calculation is the same as the number of non-equivalent sites' is not a reproducible quantitative claim.","section":"Section II.C and Figs. 6, 7, 9"},{"comment":"The counting procedure for both sides of the correspondence is under-specified. The LDOS peaks are counted 'in the energy region of the central miniband' (Fig. 7(a) caption), but no criterion defines the boundaries of this window; for instance, whether the three 'emerging DOS peaks surrounding the central miniband' described for Model A in Sec. III.A are included is decided per figure. Likewise, the number of non-equivalent sites shown in the inset of Fig. 7(a) is presented as a visually determined integer, and no operational algorithm is given for incommensurate angles such as 50 degrees, where the stacking region is aperiodic and a rigorous equivalence test for edge sites is nontrivial. Without a priori rules for the energy window and for what makes two edge sites inequivalent, the claimed 'perfect correspondence' is not falsifiable as stated.","section":"Sec. III.B, Fig. 7(a) and inset"},{"comment":"The equality between the LDOS peak count and the non-equivalent-site count relies on post-hoc exclusion of peaks that are classified as not belonging to the edge of the stacking region. Appendix B explicitly removes the E - E_F = -0.17 eV feature of the 14-TBZGNR because it 'does not correspond to a state located at the edge line of the stacking region,' and also removes the 'linear bulk' state. Because this classification is performed after inspecting the LDOS spatial distribution, the subsequent comparison in Fig. 9(b)-(c), which keeps the count of 'edge states of the stacking limited regions' equal to the number of non-equivalent boundary sites, risks being circular. The authors should provide a pre-registered criterion (for example, a threshold on the integrated LDOS weight inside the geometric stacking region) for whether a DOS feature counts as an edge-stacking state, and apply that criterion uniformly to all widths and angles before comparing with the site count.","section":"Appendix B and Sec. III.C, Fig. 9"}],"minor_comments":[{"comment":"There is an inconsistency in the use of eta: Eq. (4) and the surrounding text call eta an 'infinitesimal number,' while Sec. II.C states that a finite eta is used as a regularization parameter. Please clarify the difference and report the actual value used in the LDOS calculations.","section":"Section II.B.2 vs II.C"},{"comment":"The claim that the number of non-equivalent sites at the boundary increases with ribbon width (Fig. 9(c)) is given as a list of integers, but no explanation is given for the non-monotonic sequence (3, 3, 5, 5, 6, 7, 8) in terms of the dodecagonal-ring geometry; a short description of how these numbers are obtained would help.","section":"Section III.C, Fig. 9"},{"comment":"The '50/50' beam-splitting events at theta = 45 degrees (E - E_F ~ -0.28 eV) and theta = 15 degrees are asserted without a tolerance or a quantitative test; the authors should state how close T13 and T14 are to 0.5 and over what energy range the splitting holds.","section":"Section III.B, Fig. 5"},{"comment":"The phrase 'in accordance with reported scanning tunneling spectroscopy measurements' is only qualitatively supported by Ref. [11]; a direct comparison of calculated peak energies or relative intensities with the experimental spectra of that reference would strengthen the claim.","section":"Abstract and Sec. I"},{"comment":"The term 'linear bulk state' is used without definition; please define it (or cite a reference) so that the exclusion rule in the peak-counting procedure is transparent.","section":"Appendix B"},{"comment":"No computational parameters are reported for the HHK recursion (number of Lanczos steps, continued-fraction truncation) or for the transport self-energies (convergence criterion). Adding these parameters is necessary for reproducibility of the numerical results.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The manuscript reports a competent tight-binding study with a plausible physical picture, but the headline 'perfect correspondence' is not yet established at the claimed level. The main risk is the circularity introduced by the post-hoc exclusion of peaks in Appendix B; a revision that defines the counting rules a priori and reports eta plus convergence checks would make the claim solid. The paper fits the journal's scope, and the transport results are interesting even if the strongest claim is softened."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a useful systematic tight-binding study of crossed zigzag graphene nanoribbons, and the transport maps are the best part. The headline claim—the “perfect correspondence” between LDOS peaks and non-equivalent edge sites—is not yet pinned down well enough to be compelling.\n\nWhat’s actually new: the A1/A2 stacking configurations, the angle/width scan of LDOS peak counts, and the four-terminal conductance behavior (full suppression, lead preference, 50/50 beam splitting). The methods are standard, the HHK real-space Green function approach is well-suited, and I don’t see any sign of parameter fitting. Models A/B/C were in the experimental paper [11]; the extension and the transport analysis are the contribution.\n\nWhere it gets soft: the counting rule. Eta is called a “convenient regularization parameter” but no value is given and no convergence test is shown. Peaks are counted inside a “central miniband” window without a criterion, and Appendix B removes features post hoc when they are assigned to bulk or outside-stacking states. At incommensurate angles the number of non-equivalent sites is read off visually, so the independent variable of the correspondence is also under-defined. That means the central claim could be an artifact of these choices. This is a real soft spot, not a manufactured one. The transport claims are more robust: they follow directly from the Green functions and don’t depend on the counting. The agreement with STS in Ref. [11] is qualitative, which is fine but shouldn’t be oversold. Lattice relaxation is neglected, and the paper justifies it by citing Ref. [11]; that is minor.\n\nOverall, this is a serious piece of computational work. The symmetry-counting idea may be right, but it needs a sharper operational definition before I would trust it as a design rule. It deserves peer review, with requests for eta/window convergence studies, an algorithmic definition of non-equivalent sites at incommensurate angles, and ideally code/data release.","headline":"Useful transport study of crossed bilayer zigzag GNRs, but the headline symmetry-counting rule is not yet operationally defined enough to be trusted.","tokens_in":14234,"tokens_out":2094,"would_cite":true,"duration_ms":22168,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"The number of localized states in a twisted graphene nanoribbon junction is predetermined by the symmetry-distinct sites on the edge of its stacking region.","keywords":["twisted bilayer graphene","zigzag graphene nanoribbons","localized edge states","tight-binding model","local density of states","Landauer-Büttiker transport","electron beam splitter","stacking symmetry"],"falsifier":"For a single TBZGNR configuration, e.g., a 14-TBZGNR at 90°, compute the LDOS with η ten times smaller and with the energy window defined by the full central band rather than a chosen miniband, then count resolved peaks on the stacking-region edge. If the count differs from the number of non-equivalent boundary sites, or if the peak count changes when η is halved, the claimed perfect correspondence fails. A complementary test: compare the predicted count with STS peak counts on an atomically characterized TBZGNR junction.","tokens_in":13184,"feed_emoji":"🌀","tokens_out":7160,"duration_ms":65963,"temperature":0.7,"pith_summary":"The paper argues that in a junction of two crossed zigzag graphene nanoribbons, every symmetry-distinct site on the boundary of the overlap region hosts a localized electronic state. Counting those sites therefore predicts the number of low-energy LDOS peaks without diagonalizing the full system, a rule the authors find holds across twist angles from 90° to 15° and ribbon widths from 4 to 16. The same real-space tight-binding model shows that the twist angle, more than stacking offset or ribbon width, controls four-terminal conductance, including complete transmission suppression, energy filters, and 50/50 electron beam splitting. If correct, the count gives a simple geometric design rule for these devices and for the localized states already seen in scanning tunneling spectroscopy.","feed_headline":"Symmetry count predicts localized states in twisted nanoribbons","feed_subtitle":"A one-to-one rule ties LDOS peaks to edge-site symmetry, and twist angle tunes 50/50 splitting.","key_machinery":"The central object is the stacking region, the overlap of the two crossed nanoribbons, whose boundary sites are grouped into symmetry-equivalence classes using the region's mirror, C2, and 12-fold local symmetries. The load-bearing identity is the count: number of LDOS peaks equals number of non-equivalent edge sites. It is computed with an extended Slater–Koster tight-binding Hamiltonian with long-range hoppings, solved by the Haydock–Heine–Kelly recursive Green's function method for LDOS and by decimation-based Green's functions for Landauer–Büttiker transmission.","core_discovery":"On its own terms, the paper establishes a perfect correspondence: the number of non-equivalent sites on the edge of the stacking region equals the number of LDOS peaks in the central miniband near the Fermi level. The equivalence classes are set by the point-group symmetries of the overlap region, namely mirror lines, C2 rotations, and 12-fold local-symmetry patches. This is demonstrated for five stacking offsets at 90°, for twist angles from 90° down to 15°, and for ribbon widths n = 4 to 16, after excluding features that belong to bulk or outside-stacking states. The authors further show that twist angle is the strongest control knob: at θ = 45° the device splits an incoming current 50/50 toward one interlayer terminal, at θ = 15° the favored direction flips, and certain stackings give full three-terminal suppression. The transport follows from the same localized edge states that the counting rule enumerates.","pith_inferences":["The geometric nature of the count suggests it may survive beyond the specific tight-binding model: any method that preserves the symmetry of the stacking region should find the same number of boundary-localized states, a point that could be tested with a first-principles LDOS calculation on one of the modeled junctions.","The correspondence hints at a deeper symmetry-index statement: the number of edge-localized modes equals the number of orbits of the stacking-region boundary under its point group, which would make the rule a topological invariant rather than an empirical trend.","A practical extension the paper does not pursue is a design chart of beam-splitting energies versus twist angle and ribbon width, which would let experiments pick a geometry and doping level that place a 50/50 splitting at a desired energy.","A time-resolved wave-packet simulation could test the counting rule dynamically: inject a broad wave packet from one lead and count the distinct edge-bound modes it excites; that number should match the non-equivalent-site count."],"forward_implications":["The number of localized edge states in a twisted nanoribbon junction is fixed by geometry: count the symmetry-distinct boundary sites of the stacking region, and that is how many LDOS peaks to expect near the Fermi level.","Twist angle can act as a switch: at the same device, rotating from 45° to 15° changes which interlayer terminal receives the current and moves the 50/50 splitting energy.","In-plane stacking offsets that break mirror or C2 symmetry create extra localized states and change which terminals conduct, enabling complete blocking of all three outgoing terminals for electrons injected from lead 1 in Model A2.","Ribbon width tunes the correspondence by changing the number of non-equivalent sites, so wider ribbons have more localized states and sharper interlayer conductance onsets.","The same localized states that the counting rule predicts control four-terminal transport, so the rule doubles as a transport-design checklist."],"supporting_citations":[{"why":"Reports the experimental TBZGNR junctions and STS-observed tunable edge states that the paper's counting rule is built to explain and extend.","marker":"[11]"},{"why":"Supplies the Haydock–Heine–Kelly recursive Green's function method used to compute the LDOS whose peaks are counted.","marker":"[18–20]"},{"why":"Provides the Landauer–Büttiker Green's function formalism used for all four-terminal conductance calculations.","marker":"[21]"},{"why":"Defines the orthogonal pz tight-binding Hamiltonian that underlies the electronic-structure calculations.","marker":"[29, 30]"},{"why":"Gives the Slater–Koster parameterization of distance- and orientation-dependent transfer integrals.","marker":"[31]"},{"why":"Sets the long-range hopping amplitudes and decay length used to produce the LDOS peaks and transport curves.","marker":"[25, 32–35]"},{"why":"Identifies the quasicrystalline 12-fold local-symmetry regions and Stampli tiling used to define the stacking-region boundary and its non-equivalent sites.","marker":"[22–28]"},{"why":"Establishes the prior 50/50 electronic beam splitter in graphene nanoribbons that the paper extends to twist-angle control.","marker":"[16]"}],"fun_headline_variants":["Twist angle tunes 50/50 splitting in twisted nanoribbons","Edge symmetry count predicts localized states in twisted ribbons","Twist angle is strongest knob for 4-terminal transport","Counting non-equivalent edge sites predicts LDOS peaks"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The counting rule assumes that peaks visible in the LDOS at a finite broadening η, inside an arbitrarily chosen energy window, give a stable enumeration of localized edge states; no convergence test for η or the window is reported, and some features are later excluded by hand in Appendix B.","fun_headline_variants_meta":{"raw":{"variants":["Twist angle tunes 50/50 splitting in twisted nanoribbons","Edge symmetry count predicts localized states in twisted ribbons","Twist angle is strongest knob for 4-terminal transport","Counting non-equivalent edge sites predicts LDOS peaks"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000211,"raw_usage":{"total_tokens":1398,"prompt_tokens":910,"completion_tokens":488,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":526,"completion_tokens_details":{"reasoning_tokens":420}},"tokens_in":526,"tokens_out":488,"duration_ms":4858,"temperature":1.0,"reasoning_tokens":420,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T22:28:59.202246+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For a single TBZGNR configuration, e.g., a 14-TBZGNR at 90°, compute the LDOS with η ten times smaller and with the energy window defined by the full central band rather than a chosen miniband, then count resolved peaks on the stacking-region edge. If the count differs from the number of non-equivalent boundary sites, or if the peak count changes when η is halved, the claimed perfect correspondence fails. A complementary test: compare the predicted count with STS peak counts on an atomically characterized TBZGNR junction.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reports the experimental TBZGNR junctions and STS-observed tunable edge states that the paper's counting rule is built to explain and extend."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the prior 50/50 electronic beam splitter in graphene nanoribbons that the paper extends to twist-angle control."}],"review_version":1}