{"id":"54d8afa3-06ed-4acd-a0f6-fe8fe2457ac7","arxiv_id":"1909.01613","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"The paper predicts that edge-localized excitons in MoS2 nanoribbons appear below the bulk optical onset and are robust to edge termination and orientation.","lead":"Computer simulations predict that the edges of atomically thin MoS2 can host new low-energy bound electron-hole pairs, called excitons, that do not exist in the bulk material. These edge excitons may provide a spectroscopic way to probe edges and could influence nanoscale light-based devices.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The q=0 screened interaction W(0), supplied by an unvalidated analytic fit (SM S4), controls whether the metallic edge binds the intra-edge exciton; the paper reports neither fit parameters nor sensitivity of the E2 peak to this input.","rationale":"The reader's weakest assumption identifies SM S4's fitted W(q=0) as the most fragile premise, and I agree: this quantity enters the BSE direct term and controls whether the metallic edge can support bound excitons. The paper's own text flags q=0 as the point not well represented, and it provides no fitted parameters or sensitivity tests, so the concern is not manufactured. A legitimate counterpoint is that the analytic form Eq. (1) is physically motivated for a quasi-1D system and reproduces the computed W(q) for all q>0, as shown in Fig. S4; this gives some support to the extrapolation. However, the extrapolation to q=0 is exactly where the metallic intraband response matters, and the binding energies of order 0.4 eV could easily shift or disappear if W(0) is misestimated by a factor of order two. The k-grid convergence in Fig. S1 reduces discretization error for the nonzero-q points but does not test the model extrapolation itself. The width-independence claim is also an extrapolation from N=2 and N=3, but the q=0 screening issue is more fundamental because it threatens the existence of the predicted excitons rather than only their asymptotic width behavior. Since the reader already assigned CONDITIONAL on this basis, my assessment does not change the verdict; the recommendation is to keep CONDITIONAL and require either a sensitivity analysis of W(0) or a direct calculation of the q→0 screened interaction before full acceptance.","tokens_in":16276,"tokens_out":7893,"duration_ms":91331,"concrete_test":"Take the N=2 S-dimer ribbon and repeat the GW-BSE calculation with the q=0 contribution to the BSE direct term varied over a physically motivated range: specifically, (i) recompute the fit to Eq. (1) and report q0 and R with their statistical uncertainties; (ii) set W(0) to the best-fit value, to zero, and to twice the best-fit value; (iii) if the E2 peak position or binding energy changes by more than 0.1 eV, or the peak unbinds, the central claim is not robust to the q=0 model. A stronger check is to compute W(q) on a denser q mesh down to q0/10 and verify the fitted form directly before extrapolating.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central, strongest claim is that the width-independent E2 peak is an intrinsic, universal edge exciton that will survive in larger samples. The BSE kernel in a metallic ribbon is controlled by the q→0 head of the screened interaction. SM S4 states that q=0 is 'the only point that is not well represented' and that W(q=0) is therefore obtained by fitting the computed W(q>0) to the analytic 1D expression Eq. (1), with parameters q0 and R. The fitted W(0) is then used in the BSE direct term. Neither q0 nor R nor the resulting W(0) values are reported, and no sensitivity analysis is given. An overestimated W(0) would make bound edge excitons appear; an underestimated W(0) could reduce or destroy them. Because the prediction of a finite binding energy in a metallic 1D wire is exactly the non-standard result being claimed, this extrapolation is load-bearing, not a peripheral numerical detail. The k-point convergence check in Fig. S1 does not settle it, since it refines the q>0 sampling but does not validate the analytic continuation to q=0. The broad 'universal' claim also extrapolates from N=2 and N=3 ribbons with three ZZ terminations plus a citation for AC; but the q=0 screening issue is the more fundamental risk.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"Using G0W0 plus Bethe-Salpeter calculations on MoS2 zigzag nanoribbons (ZZ-NRs) with N=2 and N=3, and with S-dimer, S-monomer, and bare Mo-edge terminations, the authors predict two low-energy edge excitons below the bulk 2D exciton onset. E1 is an inter-edge exciton that becomes optically dark with increasing ribbon width; E2 is an intra-edge exciton that remains optically active and is claimed to be robust, universal, and observable in larger samples, with binding energies of about 0.2--0.4 eV attributed to ineffective screening of the metallic 1D edge states.","tokens_in":16583,"tokens_out":5969,"duration_ms":63389,"significance":"If correct, the predicted E2 peak would provide an experimentally accessible optical signature of edges in MoS2 and, plausibly, other group-VI TMDs, and it would be an interesting counterexample to the common expectation that metallic edges quench excitons. The manuscript has clear strengths: it uses a standard state-of-the-art G0W0-BSE framework, reports convergence tests (k-point sampling, number of bands, Tamm-Dancoff approximation), checks spin-orbit effects, and examines several edge terminations. The central quantitative prediction, however, depends on a fitted q=0 screened interaction whose parameters and sensitivity are not reported, and the 'universal' width-independent claim is extrapolated from only two ribbon widths. These issues affect the load-bearing parts of the paper and need to be addressed before the prediction can be fully accepted.","major_comments":[{"comment":"The treatment of the metallic q→0 limit is load-bearing for the central claim. As stated in SM S4, \"the only point that is not well represented is the q=0 one,\" and the W(q=0) entering the BSE kernel is obtained by fitting the finite-q ab initio data to Eq. (1), with parameters q0 and R. The paper reports neither the fitted values (q0, R) nor the resulting W(0), and provides no sensitivity analysis. An overestimated W(0) would produce spurious bound excitons, while an underestimated value could artificially suppress or shift them; the k-point convergence test in Fig. S1 does not test this continuation. The authors should report the fit parameters and W(0) for each termination and width, and demonstrate that the existence and binding energies of E1 and E2 are stable when q0 and R are varied over the fitting uncertainty, or validate W(0) with an independent calculation that includes the intraband (1D metallic) contribution at small q.","section":"Supplemental Material, Sec. S4 (Eq. 1)"},{"comment":"The \"width-independent\" and \"larger samples\" claim rests on calculations for only two widths, N=2 (1.11 nm) and N=3 (1.65 nm). While the physical reasoning (same-edge localization of electron and hole and weak width dependence of the edge bands) is plausible, it is an extrapolation to state that the intra-edge exciton is \"expected in the optical spectrum not only of narrow ribbons, but also of larger samples.\" At least one additional width (e.g., N=4 or N=5) or a quantitative convergence model is needed to support this part of the conclusion; otherwise the statement should be weakened.","section":"Main text, Fig. 2 and concluding paragraph"}],"minor_comments":[{"comment":"There is a typo in the conclusion: \"width-indepedent\" should be \"width-independent.\"","section":"Concluding paragraph"},{"comment":"In the description of Fig. S5, the text says the E1 transition has \"an intra-edge character connecting left and right edges\"; this should be \"inter-edge character,\" because E1 connects states localized on opposite edges, consistent with the main text.","section":"Supplemental Material, Sec. S5"},{"comment":"In the Fig. S8 caption, \"empy\" should be \"empty.\"","section":"Supplemental Material, Fig. S8 caption"},{"comment":"The spin-orbit check is performed only at the LDA and independent-particle level, not at the GW-BSE level; a brief statement on why the conclusion is expected to hold for the bound excitons would strengthen the argument.","section":"Supplemental Material, Sec. S3"},{"comment":"The claim that analogous edge-related excitons exist for armchair edges relies on a citation to Ref. [88] rather than on calculations in this work; specifying the level of theory used in that reference would help the reader assess the universality claim.","section":"Main text, paragraph after Fig. 3"}],"recommendation":"major_revision","confidential_remarks":"This is a potentially important computational prediction, and the absence of fitted parameter values for the q=0 screening model is a genuine reproducibility and robustness concern that should be fixed in a revision. I do not see evidence of circular fitting to the target exciton peaks; the W(q) fit is matched to independent finite-q screening data. The paper would be suitable for publication after the q=0 sensitivity analysis is provided and the width-extrapolation claim is either supported by additional data or appropriately qualified."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is a solid computational paper with a genuinely new prediction: metallic zigzag MoS2 nanoribbons host edge excitons that sit well below the bulk A/B exciton peaks. The authors identify two classes, inter-edge (E1) and intra-edge (E2), and show across three edge terminations that E2 is stable and width-independent. If true, this gives experimentalists a non-invasive optical signature of edge quality, so the stakes are real.\n\nWhat is done well: the GW-BSE calculations are careful. They check k-point convergence, number of bands, Tamm-Dancoff, and spin-orbit effects, and they are explicit that a scissor correction is inadequate. The termination study (bare, S-monomer, S-dimer) is a genuine robustness test. The binding energies (~0.2-0.4 eV) are reported per peak, and the exciton wavefunctions are consistent with the inter- vs intra-edge assignment. The comparison to semimetallic carbon nanotubes is apt. The referencing is honest; the authors cite their own prior work on polar-discontinuity edge states because that is the relevant antecedent.\n\nThe soft spot is the q=0 screened interaction. The SM says the q=0 point is not well represented and they fit W(q>0) to a 1D analytic form, then use the fitted W(0) in the BSE kernel. That is load-bearing: in a metallic 1D wire, whether a bound exciton appears at all depends on the long-wavelength head of W. The paper gives no fit parameters (q0, R) and no sensitivity analysis. The k-point convergence check refines q>0 sampling but does not validate the analytic continuation to q=0. I do not think the concern is fatal, because the fitting form is physically motivated and matches finite-q data, but the reader cannot check how robust the 0.4 eV binding energy is to the extrapolation. A sensitivity scan or an independent screening model would settle it.\n\nThe \"universal\" claim is also slightly beyond the directly computed parameter space: N=2 and N=3 ribbons, three ZZ terminations, and a citation for AC edges. The width-independence is plausible from the edge-state picture, but it is an extrapolation. This is a minor issue relative to the W(0) question.\n\nWho this is for: anyone computing or measuring TMD edge states, and experimental groups doing low-temperature photoluminescence on flakes or ribbons. It deserves peer review. The question is well-posed, the methods are state-of-the-art, and the prediction is falsifiable. I would ask for the sensitivity analysis and a slightly softened \"universal\" phrasing, but this is not a desk reject. Engage with it.","headline":"Careful GW-BSE study predicting edge-localized excitons in MoS2 ribbons; the central prediction is plausible but rests on an unvalidated interpolation of W(q=0) that deserves sensitivity testing before I'd bet on it.","tokens_in":17139,"tokens_out":2959,"would_cite":true,"duration_ms":31537,"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":"MoS2 ribbon edges host a 1.5 eV exciton below the bulk peaks","keywords":["edge excitons","MoS2","transition metal dichalcogenides","zigzag nanoribbons","GW-BSE","Bethe-Salpeter equation","many-body perturbation theory","excitons in metallic systems"],"falsifier":"Measure local optical absorption or photoluminescence on a MoS2 flake with straight zigzag edges, scanning across the edge at energies below the 1.88 eV A exciton: the central prediction fails if no width-independent, edge-localized peak near 1.5 eV appears. A calculation that recomputes the Bethe-Salpeter spectrum using an independent, non-fitted value for W(q=0) would also settle whether the edge excitons remain bound.","tokens_in":16094,"feed_emoji":"🔬","tokens_out":8071,"duration_ms":77817,"temperature":0.7,"pith_summary":"Single-layer molybdenum disulfide, a semiconductor that absorbs light through tightly bound electron-hole pairs, is predicted to have additional excitons that belong to its edges rather than its interior. Using first-principles many-body calculations at the GW plus Bethe-Salpeter level, the paper finds two such excitons in zigzag nanoribbons, at roughly 0.55 eV and 1.49 eV, both well below the familiar A and B excitons of the two-dimensional sheet. The lower one binds electrons and holes on opposite edges and fades once ribbons exceed a few nanometres in width; the higher one lives on a single edge and is essentially unchanged as the ribbon widens. Because the same metallic edge states appear across edge terminations, orientations, and the whole family of group-VI transition-metal dichalcogenides, the authors argue this intra-edge exciton is a universal, intrinsic feature of edges rather than a ribbon-size artifact. If correct, it makes a naked edge optically visible at sub-gap energies in samples that range from nanoribbons to large flakes.","feed_headline":"MoS2 ribbon edges host a 1.5 eV exciton below the bulk peaks","feed_subtitle":"The peak is predicted to survive in wide flakes, giving a clear optical fingerprint of TMD edges.","key_machinery":"The load-bearing object is the set of mid-gap, metallic edge states that arise from the polar discontinuity where the MoS2 sheet terminates; these states supply both the single-particle transitions that form the excitons and the metallic screening that the calculation must handle at q→0. The machinery is the GW-BSE scheme with a truncated Coulomb interaction, in which the poorly represented W(q=0) term is supplied by fitting the computed finite-q screening to a one-dimensional analytic form, introducing an inverse screening length and transverse ribbon size. The argument turns on classifying the BSE eigenstates by their spatial electron-hole distribution—inter-edge versus intra-edge—and on showing that the intra-edge distribution survives as the ribbon width grows.","core_discovery":"On the paper's own terms, the central discovery is that the metallic edge states of MoS2 nanoribbons—normally expected to screen Coulomb interactions so strongly that no excitons survive—still bind electron-hole pairs with substantial energy. Classifying the two lowest optical peaks, the authors identify an inter-edge exciton (E1) whose electron and hole sit on opposite ribbon edges, with binding energy about 0.21–0.25 eV and optical activity that vanishes with increasing width, and an intra-edge exciton (E2) whose electron and hole sit on the same edge, with binding energy about 0.39–0.42 eV that persists at larger widths. They show E2's position, binding, and character are nearly independent of edge termination (bare, S-monomer, S-dimer) and also appear for armchair ribbons, and they conclude that this intra-edge excitation is robust and universal—an intrinsic consequence of the edge's existence. The finite binding is attributed to screening being ineffective in one dimension, mirroring earlier findings on semimetallic carbon nanotubes.","pith_inferences":["If E2 is width-independent, edge-mapping optical techniques such as near-field photoluminescence on flakes with straight edges should see a sub-gap peak localized at the edge; this is a direct, testable consequence the paper does not itself propose.","The strong sensitivity to screening geometry suggests the edge-exciton energy should shift with the dielectric environment, such as substrate or capping layers; a systematic study of that shift could sharpen or falsify the predicted universality.","Because the fitted W(q=0) controls binding, the quantitative values of 0.21–0.42 eV carry an uncertainty not quantified here; recomputing the spectra with an independent treatment of metallic screening at q=0 would test whether the edge excitons are indeed bound.","If the intra-edge exciton proves real, it could seed sub-gap optical nonlinearities or low-threshold gain in TMD nanostructures, since the transition is spatially localized yet coupled to the metallic edge continuum."],"forward_implications":["The intra-edge E2 exciton should appear as a clear, edge-localized peak below the bulk A and B excitons in large MoS2 samples, not just in nanoribbons.","The inter-edge E1 exciton is a fingerprint of ultranarrow ribbons, and its disappearance with width tracks the vanishing wavefunction overlap between opposite edges.","Excitons can remain bound at metallic one-dimensional edges, so metallic screening does not automatically quench optical resonances in these systems.","Because edge mid-gap states are shared across edge terminations and orientations, similar sub-gap edge absorption is expected for armchair ribbons and for other group-VI TMDs such as MoSe2 and WS2.","Accurate quasiparticle corrections across the whole Brillouin zone are required; simplified scissor-style corrections would misplace the edge states relative to bulk bands."],"supporting_citations":[{"why":"Supplies the reference A and B exciton peaks (1.88/2.03 eV) that define the 'below bulk onset' claim.","marker":"[15]"},{"why":"Establishes mid-gap edge states in MoS2 nanoribbons, the single-particle basis of the new excitons.","marker":"[49]"},{"why":"Traces the metallic edge states to a polar discontinuity at zigzag edges, making them intrinsic to termination.","marker":"[51]"},{"why":"Provides the GW-BSE many-body framework used for all reported spectra and binding energies.","marker":"[77]"},{"why":"Documents the fitted one-dimensional screening model that supplies the critical W(q=0) in the BSE kernel.","marker":"[82]"},{"why":"Gives the ~1 eV binding energy of 2D MoS2 bulk excitons used to contrast with the smaller edge-exciton bindings.","marker":"[87]"},{"why":"Shows excitons can bind in metallic one-dimensional systems, the precedent invoked for finite binding despite metallic edges.","marker":"[93]"},{"why":"Extends the carbon-nanotube exciton analysis and is cited alongside [93] as evidence that 1D metals can sustain bound electron-hole pairs.","marker":"[94]"}],"fun_headline_variants":["Metallic MoS2 edges still bind 0.4 eV excitons","Universal edge excitons in MoS2 ignore termination","Intrinsic MoS2 edge excitons with 0.4 eV binding","MoS2 edge excitons survive metallic screening"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the fitted one-dimensional screening formula correctly gives the screened electron-hole attraction at the very long wavelength (zero momentum) that the calculation cannot sample directly; if that extrapolation is too attractive, the predicted edge excitons could fail to bind.","fun_headline_variants_meta":{"raw":{"variants":["Metallic MoS2 edges still bind 0.4 eV excitons","Universal edge excitons in MoS2 ignore termination","Intrinsic MoS2 edge excitons with 0.4 eV binding","MoS2 edge excitons survive metallic screening"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000562,"raw_usage":{"total_tokens":2636,"prompt_tokens":880,"completion_tokens":1756,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":496,"completion_tokens_details":{"reasoning_tokens":1684}},"tokens_in":496,"tokens_out":1756,"duration_ms":15091,"temperature":1.0,"reasoning_tokens":1684,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T05:13:09.040852+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure local optical absorption or photoluminescence on a MoS2 flake with straight zigzag edges, scanning across the edge at energies below the 1.88 eV A exciton: the central prediction fails if no width-independent, edge-localized peak near 1.5 eV appears. A calculation that recomputes the Bethe-Salpeter spectrum using an independent, non-fitted value for W(q=0) would also settle whether the edge excitons remain bound.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes mid-gap edge states in MoS2 nanoribbons, the single-particle basis of the new excitons."},{"cited_title":"Gibertini and N","cited_arxiv_id":null,"evidence_quote":"Traces the metallic edge states to a polar discontinuity at zigzag edges, making them intrinsic to termination."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Documents the fitted one-dimensional screening model that supplies the critical W(q=0) in the BSE kernel."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Shows excitons can bind in metallic one-dimensional systems, the precedent invoked for finite binding despite metallic edges."},{"cited_title":"Deslippe, C","cited_arxiv_id":null,"evidence_quote":"Extends the carbon-nanotube exciton analysis and is cited alongside [93] as evidence that 1D metals can sustain bound electron-hole pairs."}],"review_version":1}