{"id":"c0c81a09-69fa-4418-80c2-1bebe72758b0","arxiv_id":"1908.09436","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"A semiempirical tight-binding GW-BSE model reproduces thickness-dependent exciton energies and binding energies in layered lead-iodide perovskites, showing a 2D-to-3D crossover.","lead":"Researchers built a cheap computer model to predict how the optical properties of layered lead-halide perovskites change with layer thickness. The model matches experiments for exciton energies and binding energies, which matters for designing better solar cells and LEDs.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The n-dependent quasiparticle correction is the least secure link: it rests entirely on a classical image-charge model with bulk-fitted GW parameters, and no layered ab initio benchmark validates it.","rationale":"I read the paper as a semiempirical prediction: bulk-fitted parameters, when combined with classical dielectric contrast, are used to predict thickness-dependent exciton energies without fitting to layered experimental data. That is a legitimate and falsifiable construction, and the layer-dependent comparison in Fig. 4(b) is a real prediction. The reader's conditional verdict is appropriate. My stress-test identifies the transferability of the layered self-energy as the single load-bearing link: all n-dependent quasiparticle effects enter through a classical delta-Sigma(z), and the same classical W appears in the BSE kernel. A direct ab initio GW-BSE calculation for n=1 or n=2 would settle whether this approximation is quantitatively reliable. I found two smaller issues worth flagging. First, the sentence defining the 2D/3D hydrogenic ratios is reversed: the text says 'in 2D E1s/E2s = 4 and in 3D E1s/E2s = 9', whereas the correct values are 2D = 9 and 3D = 4; the later usage around Fig. 4(c) is correct, so this appears to be a typo. Second, the abstract lists the n=infinity binding energy (37 meV) as 'calculated', but the body computes it from the hydrogenic model rather than from the tight-binding BSE; this should be stated explicitly. Neither issue moves the verdict, because the main concern is the same one the reader flagged and it is addressable by a targeted calculation.","tokens_in":14367,"tokens_out":10836,"duration_ms":110062,"concrete_test":"Perform a fully ab initio G0W0-BSE calculation for the n=1 layered perovskite (for example Cs2PbI4 or (BA)2PbI4 with an experimental or DFT-relaxed structure) and compare the quasiparticle gap, the 1s and 2s exciton energies, and the 1s binding energy against the model's n=1 predictions. If the ab initio quasiparticle gap differs from the model by more than about 0.1 eV, or if the BSE binding energy differs by more than about 30 meV, then the classical transferability assumption in Eqs. 5-6 is not sufficient and the central quantitative claim needs qualification.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim requires that the only substantial n-dependent quasiparticle effect is the classical electrostatic self-energy delta-Sigma(z) of Eqs. 5-6. The model transfers all eight parameters (tight-binding, SOC, five orbital self-energies, scissors shift) from bulk cubic CsPbI3/MAPbI3 to the layered Ruddlesden-Popper structures, and then adds delta-Sigma(z) from a uniform-slab dielectric model as a first-order perturbation (Eq. 6). Nothing in the paper checks this transferability against a layered ab initio GW calculation. In a monolayer, the GW self-energy can differ from the bulk not just through the long-range image potential but also through the change in RPA screening at short range, the modified wavefunction localization and confinement, and the atomistic electronic structure of the organic spacer; these effects are not captured by delta-Sigma(z). If they contribute more than roughly 0.1 eV to the quasiparticle gap or to the n=1 exciton binding energy, the agreement with experiment in Fig. 4(b) could be partly coincidental, and the mechanistic decomposition in Fig. 5 would be unreliable. A separate internal issue: the abstract's 37 meV bulk value is not a BSE result; the paper derives it from the hydrogenic formula with mu=0.10 m0 and epsilon=6.1.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript presents a semiempirical tight-binding model for layered Ruddlesden-Popper lead-iodide perovskites. The mean-field Hamiltonian is built from a DFT/Wannier tight-binding parametrization with spin-orbit coupling; quasiparticle corrections are introduced as orbital-dependent diagonal self-energies fitted to a bulk GW band structure and are combined, for layered systems, with a classical electrostatic self-energy delta-Sigma(z). Excitonic spectra are obtained by solving the Bethe-Salpeter equation with a classical screened Coulomb interaction. The authors report thickness-dependent 1s and 2s exciton energies for n=1-5, compare them with experiment in Fig. 4(b), extract binding energies, and analyze the nonhydrogenic Rydberg series and the 2s-2p degeneracy breaking.","tokens_in":14650,"tokens_out":6258,"duration_ms":60448,"significance":"If the central claim is accepted, this work provides a computationally affordable atomistic route to excitonic properties of layered perovskites and a useful quantitative decomposition into carrier-confinement, dielectric, and exciton-binding contributions. The agreement with experiment for both 1s and 2s energies across n is a concrete, falsifiable prediction, and the term-by-term mechanism analysis is a genuine strength. The explicit treatment of the full Brillouin zone goes beyond effective-mass models. However, the strength of the claim depends on the transferability of bulk-fitted self-energy parameters and on the classical model of dielectric screening; the paper does not benchmark either against an independent layered first-principles calculation.","major_comments":[{"comment":"The sentence 'At n=5, the ratio is approximately 4, indicating a conventional 3D exciton within the hydrogenic model' is inconsistent with the definitions given two sentences earlier, where the 3D hydrogenic ratio is 9 and the 2D ratio is 4. A ratio of approximately 4 at n=5 would indicate 2D-like behavior, contradicting the paper's own conclusion that conventional hydrogenic 3D behavior is recovered at large n. The text and Fig. 4(c) should be checked and the interpretation corrected.","section":"Fig. 4(c) and following paragraph"},{"comment":"The bulk value E_1s^b = 37 meV for n=infinity is not produced by the BSE calculation used for n=1-5; it is obtained from the hydrogenic formula Ry = mu/(2 epsilon^2) with mu = 0.10 m0 and epsilon = 6.1. Listing this in the abstract as 'calculated to be' obscures the distinction between a many-body calculation and an analytic estimate, and the text should label it accordingly.","section":"Abstract and 'Excitons in semiconductors' paragraph"},{"comment":"The thickness-dependent quasiparticle correction is the classical electrostatic self-energy delta-Sigma(z) of Eq. (5), added perturbatively via Eq. (6) to self-energies fitted to bulk GW. No comparison to an independent layered ab initio GW or other first-principles calculation is provided to validate the transferability of the bulk self-energy constants to monolayers. Since the claimed agreement with experiment within 0.1 eV in Fig. 4(b) depends on this assumption, the authors should either provide a benchmark for at least one small-n system or clearly state the resulting uncertainty in the predicted quasiparticle gaps and binding energies.","section":"Eqs. (5)-(6) and Fig. 4(b)"}],"minor_comments":[{"comment":"In the sentence reporting binding energies, '112 mev' should be '112 meV'.","section":"Results after Fig. 4"},{"comment":"The caption phrase 'a ratio of 9 or 4 is the prediction of an ideal 2D or 3D hydrogenic model' is ambiguous; it should be rephrased to state explicitly that 9 corresponds to the 3D model and 4 to the 2D model.","section":"Fig. 4(c) caption"},{"comment":"The term 'GW-BSE study' may overstate the level of theory for the layered systems, where the self-energy is a fitted diagonal shift plus a classical electrostatic image-potential correction rather than an ab initio GW self-energy; this should be stated more prominently.","section":"Title and abstract"},{"comment":"The numerical values of the fitted parameters (Delta_Pb_SOC, Delta_I_SOC, the five Sigma_mu, and Sigma_CB) are not reported, which would improve reproducibility and facilitate comparison with other tight-binding parametrizations.","section":"Bulk parameterization paragraph"}],"recommendation":"major_revision","confidential_remarks":"The paper is promising and the central physical picture is likely sound, but the current text contains a clear inversion of the 2D/3D ratio interpretation in the discussion of Fig. 4(c), which may indicate an error in the figure or in the text. The manuscript would also benefit from an explicit layered benchmark or a clearly quantified caveat about the transferability of bulk-fitted self-energies, and the abstract should not present the hydrogenic bulk estimate as a BSE result. These issues are fixable within the scope of the manuscript."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Punchline: this is a clever, usable model, and the main result—thickness-dependent 1s and 2s exciton energies matching Blancon et al. within ~0.1 eV—survives a careful reading. It is not a fully ab initio GW-BSE paper, but it doesn't claim to be.\n\nWhat's new: a Wannier/tight-binding Hamiltonian on the full perovskite lattice, with spin-orbit coupling, fitted to bulk GW, plus a classical image-charge self-energy for the layered geometry. That gives the first systematic atomistic sweep across n=1–5 and bulk for layered lead iodides, including the nonhydrogenic Rydberg ratio, the 2s–2p splitting, and a clean decomposition of the optical gap into confinement, dielectric, and exciton-binding terms. The agreement with experiment is not circular: the parameters are pinned to a bulk CsPbI3/MAPbI3 GW reference, not to the layered exciton data being predicted.\n\nWhere I'd push: the layer-dependent quasiparticle correction is the least secure piece. δΣ(z) is pure electrostatics. In a monolayer, RPA screening at short range, wavefunction localization, and the atomistic organic/inorganic interface can modify the self-energy beyond the image-charge term. There is no layered ab initio GW benchmark to test that, so the Fig. 5 decomposition should be read as model-based rather than measured. I don't think this breaks the central claim—the experimentally checked energies come from the full calculation, and the discrepancies are small—but it caps how much mechanism you can attribute. Also, the abstract's 37 meV bulk binding energy is a hydrogenic estimate (μ=0.10m0, ε=6.1), not a BSE output; that should be labeled. Minor: no code, data, or error bars for the parameters; only two valence and two conduction bands in the BSE; and no sensitivity scan over εi/εo, which is worth adding since the screening model carries real weight.\n\nBottom line: anyone modeling 2D halide perovskites will want this on hand, and I'd send it out. I'd ask the authors to add a transferability check against even one layered ab initio GW-BSE result, and to relabel the bulk value so readers don't mistake it for a BSE outcome.","headline":"A cheap, non-circular tight-binding GW-BSE model that gets the n-dependent exciton energies of layered lead-iodide perovskites right, but whose classical dielectric self-energy would benefit from a layered ab initio transferability check.","tokens_in":15217,"tokens_out":2463,"would_cite":true,"duration_ms":27854,"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":"A semiempirical tight-binding GW-BSE model reproduces the thickness-dependent exciton binding energies of layered lead-halide perovskites, from 302 meV at n=1 to 37 meV in bulk.","keywords":["Ruddlesden-Popper perovskites","layered perovskites","exciton binding energy","GW approximation","Bethe-Salpeter equation","tight-binding model","spin-orbit coupling","dielectric confinement"],"falsifier":"Measure the 2s-2p splitting of the n=1 exciton by high-field magneto-optics or two-photon spectroscopy: the model predicts it is about 1 meV (2p lower), and the 1s binding energy near 302 meV in bulk crystals. If experiment finds near-degenerate 2s/2p or a binding energy far outside 250-350 meV, the dielectric-confinement picture fails.","tokens_in":14135,"feed_emoji":"⚛️","tokens_out":6836,"duration_ms":63061,"temperature":0.7,"pith_summary":"This paper aims to show that the thickness-dependent optical properties of Ruddlesden–Popper lead-iodide perovskites—the family A′2An−1PbnI3n+1—can be captured by a semiempirical model that is far cheaper than a full ab initio many-body calculation. The model starts from a tight-binding band structure with spin-orbit coupling, corrects it with GW self-energies borrowed from the bulk perovskite, and accounts for the layered dielectric environment with a purely classical electrostatic correction. The central quantitative claims are 1s exciton binding energies of 302 meV for n=1, 97 meV for n=5, and 37 meV for bulk MAPbI3, together with a nonhydrogenic Rydberg series that relaxes to ordinary hydrogenic behavior as n grows. If the model is right, it provides a practical route to predicting exciton physics in large-unit-cell layered materials without resolving every atom at the GW-BSE level.","feed_headline":"Exciton binding in layered perovskites: 302 meV down to 37 meV","feed_subtitle":"Thickness-dependent Ruddlesden-Popper lead-iodide perovskites match experiment within 0.1 eV across the 2D-to-3D crossover.","key_machinery":"The engine of the calculation is a semiempirical tight-binding Hamiltonian for the Pb 6s/6p and I 5s/5p orbitals, built from a bulk DFT calculation plus two nonperturbative spin-orbit coupling constants; into that Hamiltonian the paper inserts five orbital-resolved GW self-energies and a conduction-band scissors shift, all fitted to reproduce the ab initio GW band structure of bulk MAPbI3. For a layered structure the only additional ingredient is the electrostatic self-energy δΣ(z) of Eq. (5), obtained from the classical screened Coulomb interaction in a stack of alternating inorganic and organic dielectric slabs, applied through the perturbative expectation values of Eqs. (6a)-(6b). Optical spectra follow from solving the Bethe-Salpeter equation with the two-electron integrals built from the same classical W, using tight-binding momentum and dipole matrix elements. This machinery is what lets the paper treat unit cells with many atoms while keeping the mean-field, self-energy, and exciton problems on the same footing.","core_discovery":"The core claim is that a single set of bulk-derived tight-binding and GW parameters, combined with a classical dielectric-confinement self-energy, quantitatively explains the evolution of excitons from two to three dimensions in layered lead-iodide perovskites. The paper reports 1s exciton binding energies of 302 meV (n=1), 177 meV (n=2), 135 meV (n=3), 112 meV (n=4), and 97 meV (n=5), with the bulk limit estimated at 37 meV, and shows that the calculated 1s and 2s excitation energies track experimental values within 0.1 eV for all n. It also finds a nonhydrogenic exciton series: at n=1 the 1s/2s binding-energy ratio is about 6, intermediate between the 2D limit of 4 and the 3D limit of 9, and the 2s-2p degeneracy is broken by about 1 meV at n=1 while recovering degeneracy at large n. In an exfoliated n=1 bilayer the model predicts a 1s binding energy of 444 meV and an absorption peak at 2.50 eV, matching the measured 2.53 eV.","pith_inferences":["If the electrostatic-confinement picture holds, the same classical W could be used to predict trion and biexciton binding energies in these layered perovskites, giving definite numbers to compare with the multi-carrier complexes the paper mentions as future work.","A sharp test of the model would be a direct measurement of the 2s-2p splitting in n=1; the predicted ~1 meV splitting is small but resolvable with high-field magneto-optics, and would discriminate a purely dielectric model from models with additional quantum-confinement corrections.","Because the model retains atomistic orbital character (e.g., the different roles of in-plane versus vertical iodine orbitals at the valence band edge), it could be extended to study how chemical substitutions on the halide site shift the exciton series, a connection the paper does not pursue.","The near-cancellation of dielectric self-energy and exciton binding suggests a general design rule for layered semiconductors: changes in the environment will move the band gap but barely move the lowest exciton peak, which could matter for heterogeneous device architectures such as perovskite-on-substrate junctions."],"forward_implications":["The thickness dependence of the optical gap is set mainly by quantum confinement, not by dielectric contrast, because the increase in exciton binding energy almost cancels the dielectric self-energy shift.","The 1s absorption energy is nearly independent of the organic spacer length, so chemical tuning of the band gap (e.g., from butyl to dodecyl ammonium) should leave the lowest exciton peak roughly fixed.","Exfoliated ultrathin layers are predicted to show substantially larger exciton binding energies (444 meV for a single n=1 bilayer) and a 2s state that blue-shifts more than the 1s state.","The nonhydrogenic 1s/2s ratio and the ~1 meV 2s-2p splitting at n=1 are specific fingerprints that high-resolution magneto-optical experiments could confirm.","The same parameterization can be transferred to other large-unit-cell layered compounds, avoiding full ab initio GW-BSE calculations in materials where the bulk parameters are known."],"supporting_citations":[{"why":"Supplies the ab initio GW band structure of bulk MAPbI3 that the tight-binding parameters, SOC constants, orbital self-energies, and scissors shift are fitted to reproduce.","marker":"[16]"},{"why":"Provides the experimental thickness-dependent 1s and 2s exciton energies used to validate the model across n.","marker":"[9]"},{"why":"Gives the closed-form screened Coulomb interaction for an infinite stack of dielectric layers, the basis for W in the layered perovskite model.","marker":"[13]"},{"why":"Supplies the electrostatic transfer-matrix method used to compute W for finite layer stacks, including the exfoliated bilayer case.","marker":"[43]"},{"why":"Supplies the on-site spin-orbit coupling Hamiltonian for Pb and I p orbitals, which the paper includes nonperturbatively.","marker":"[25]"},{"why":"Provides the image-charge self-energy expression used to define the electrostatic correction δΣ(z) for a charge in a dielectric layer.","marker":"[44]"},{"why":"Extends the self-energy correction to dielectric quantum-well geometries, forming the basis for the layered-substrate comparison.","marker":"[45]"},{"why":"Provides the magneto-absorption measurement of the exciton reduced mass (0.104m0) used to validate the band effective masses.","marker":"[41]"},{"why":"Supplies the experimental spectrum of an exfoliated ultrathin n=1 bilayer used to test the predicted 2.50 eV peak and 444 meV binding energy.","marker":"[10]"},{"why":"Gives a prior all-inorganic GW-BSE calculation for n=1, providing a reference point for the semiempirical approach.","marker":"[21]"}],"fun_headline_variants":["Thickness tunes exciton binding: 302 meV (2D) to 37 meV (bulk)","Layered perovskites: exciton binding follows 2D-to-3D crossover","Exciton series in perovskites: nonhydrogenic from 2D to bulk","Perovskite exciton binding: theory matches experiment within 0.1 eV","2D to 3D: exciton binding drops from 302 to 37 meV"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The model assumes that the only difference between bulk and layered perovskites is the classical electrostatic environment; if the atomic-scale electronic screening in thin layers is genuinely different from bulk, the predicted exciton energies would move.","fun_headline_variants_meta":{"raw":{"variants":["Thickness tunes exciton binding: 302 meV (2D) to 37 meV (bulk)","Layered perovskites: exciton binding follows 2D-to-3D crossover","Exciton series in perovskites: nonhydrogenic from 2D to bulk","Perovskite exciton binding: theory matches experiment within 0.1 eV","2D to 3D: exciton binding drops from 302 to 37 meV"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00082,"raw_usage":{"total_tokens":3648,"prompt_tokens":1059,"completion_tokens":2589,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":675,"completion_tokens_details":{"reasoning_tokens":2472}},"tokens_in":675,"tokens_out":2589,"duration_ms":16614,"temperature":1.0,"reasoning_tokens":2472,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:11:00.560263+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the 2s-2p splitting of the n=1 exciton by high-field magneto-optics or two-photon spectroscopy: the model predicts it is about 1 meV (2p lower), and the 1s binding energy near 302 meV in bulk crystals. If experiment finds near-degenerate 2s/2p or a binding energy far outside 250-350 meV, the dielectric-confinement picture fails.","supporting_citations":[{"cited_title":"The 2s exciton state, having a smaller binding energy, exhibits a larger increase by 90 meV","cited_arxiv_id":null,"evidence_quote":"Supplies the experimental spectrum of an exfoliated ultrathin n=1 bilayer used to test the predicted 2.50 eV peak and 444 meV binding energy."}],"review_version":1}