{"id":"259ee318-152e-4b0e-ba9d-1c1aeaf4614b","arxiv_id":"2607.19789","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"high","formal_verification":"none","parameter_count":5,"one_line_summary":"A 3D finite-element model predicts 26.7–28.2% power conversion efficiency for ZnSe/MgXS3/Sb2S3 solar cells, but key input parameters are estimates and the reported 'optimized' conditions conflict with the paper's own parameter sweeps.","lead":"This paper simulates MgTiS3, MgZrS3, and MgHfS3 chalcogenide-perovskite solar cells in COMSOL and reports predicted efficiencies of 26.7–28.2%. The numbers hinge on estimated material properties and are not experimentally validated.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The paper's own doping and defect sweeps (Fig. 3b,c) show higher PCE at 10^19 cm^-3 doping and 10^12 cm^-3 defects than the reported 'optimized' values, so the headline efficiency claim is not internally supported as optimal.","rationale":"Central claim requires the reported PCEs to be optimized over the investigated parameters. The paper's own data show that they are not: both doping and defect-density sweeps have better PCEs at parameter extremes. That is a direct internal inconsistency, not a matter of outside consensus. It undermines the specific numbers in the abstract and conclusion. Table 1's discrepancy adds to the unreliability. The reader's weakest_assumption identified the 1D optical model; that is a legitimate external-model concern, but the optimization contradiction is more decisive because it is visible within the manuscript itself and cannot be rescued by better absorption physics. Agreement is partial because the overall REJECT is supported, but for a different primary reason. Since the central claim is not reproducible or internally consistent, the reader's REJECT stands; no change to verdict needed.","tokens_in":22426,"tokens_out":3557,"duration_ms":35499,"concrete_test":"Using the authors' model parameters, reproduce the sweeps in Fig. 3(b)-(c) and then run a joint grid over acceptor doping (10^17-10^19 cm^-3) and defect density (10^12-10^14 cm^-3) at 800 nm for each absorber. If any grid point yields PCE above the reported 26.72%/28.18%/28.16%, the 'optimized conditions' claim is refuted. If code is unavailable, a minimal analytic check is to compare the reported PCEs with the sweep endpoints in Fig. 3; the endpoints already exceed them, so the authors must either disclose a constraint that invalidates those endpoints or revise the optimization claim.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The headline claim is that 'under optimized conditions' the devices reach 28.18%, 26.72%, and 28.16% PCE. The paper specifies (Sec. 3.2) the optimized values as absorber thickness 800 nm, doping 10^17 cm^-3, and defect density 10^14 cm^-3. But the same section's sweeps contradict that choice: Fig. 3(b) shows PCE increasing monotonically with doping from 10^15 to 10^19 cm^-3, reaching 31.67% (MgTiS3), 33.24% (MgZrS3), and 31.78% (MgHfS3) at 10^19 cm^-3. Fig. 3(c) shows PCE decreasing as defect density rises, with 32.41%, 32.81%, and 31.16% at 10^12 cm^-3. Thus the 'optimized' points are worse than the endpoints of both one-dimensional sweeps; no joint-optimization procedure, constraint, or cost function is given to explain why 10^17/10^14 is chosen. Table 1 further lists acceptor concentrations of 10^18 cm^-3 for all absorbers, conflicting with the text's 10^17 cm^-3. Because the central claim is explicitly about optimized performance, this internal inconsistency is load-bearing: the reported PCEs may simply be arbitrary points on the sweep, not optima. The 1D Beer-Lambert optical treatment is a separate risk, but this contradiction does not depend on external physics.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper presents COMSOL Multiphysics simulations of n-ZnSe/p-MgXS3/p+-Sb2S3 chalcogenide-perovskite solar cells (X = Ti, Zr, Hf). It sweeps absorber thickness, doping, and defect density; window and BSF layer properties; temperature; series/shunt resistance; and compares devices with and without an Sb2S3 back-surface field. The headline result is that under the stated optimized conditions (800 nm absorber, 10^17 cm^-3 doping, 10^14 cm^-3 defects) the simulated devices achieve PCEs of 28.18% (MgZrS3), 26.72% (MgTiS3), and 28.16% (MgHfS3), with VOC 0.94/0.74/1.07 V and JSC 34.46/42.69/29.89 mA/cm2. The paper also reports coupled electro-thermal maps showing small steady-state temperature rises and argues that the 3D framework provides insight inaccessible to 1D tools.","tokens_in":22852,"tokens_out":7349,"duration_ms":69613,"significance":"The forward device model is not circular: the efficiency numbers are outputs of a drift-diffusion simulation using material parameters taken from prior DFT and experimental literature, and no fitting to a target efficiency is claimed. If the inputs and optimization were reliable, these results would make a useful case for MgXS3 as a lead-free absorber and demonstrate a multiphysics modeling workflow. The systematic sweeps and BSF quantum-efficiency comparison are informative. However, the central 'optimized' claim is contradicted by the paper's own sweeps, and the optical generation model is effectively 1D despite the 3D framework claim. The absolute PCEs should therefore be treated with caution until these issues are resolved. No code or input files are provided, which limits reproducibility of the 3D model.","major_comments":[{"comment":"The central efficiency claim is not internally supported as an optimum. The text states the optimized absorber parameters are 800 nm thickness, 10^17 cm^-3 doping, and 10^14 cm^-3 defect density, and the abstract/conclusion report PCEs of 28.18% (MgZrS3), 26.72% (MgTiS3), and 28.16% (MgHfS3) 'under optimized conditions.' However, the doping sweep in Fig. 3(b) shows PCE increasing monotonically to 31.67% (MgTiS3), 33.24% (MgZrS3), and 31.78% (MgHfS3) at 10^19 cm^-3, and the defect sweep in Fig. 3(c) shows PCEs of 32.41%, 32.81%, and 31.16% at 10^12 cm^-3, all higher than the reported optima. The thickness sweep is described as monotonically increasing over 400-1200 nm, so 800 nm is not established as optimal either. No joint-optimization procedure, constraint, or cost function is provided to explain the choice. Table 1 further lists acceptor densities of 10^18 cm^-3 for all absorbers, con","section":"§3.2, Fig. 3, Table 1"},{"comment":"The claimed '3D optical-electrical-thermal framework' is not a 3D optical model. The depth-dependent photogeneration rate G_tot(z) is obtained from a wavelength-resolved Beer-Lambert attenuation law with a direct-transition Tauc-type absorption coefficient and is then imported into the Semiconductor Module (Sec. 2.2, refs. 43-45). No 3D electromagnetic wave simulation is performed; interference, reflection, texture, and lateral optical nonuniformities are ignored. Since J_SC and therefore the headline PCEs are set by G_tot(z), the abstract's claims of 'geometry-aware predictions ... inaccessible to 1D tools' and 'self-consistently couple[d] optical generation' overstate what was done. The optical input is effectively 1D; the 3D claim should be revised or a full-wave optical solve added, and the Tauc parameterization should be checked against the first-principles absorption spectra cited","section":"§2.2, Abstract, §1"},{"comment":"Absolute efficiency and voltage predictions depend on recombination parameters that are not reported. The paper sweeps defect density but does not specify SRH capture cross-sections, carrier thermal velocities, or resulting lifetimes for electrons and holes in any layer; Table 1 lists only defect concentrations. Equation (10) uses U_SRH and U_Auger, but Auger coefficients are also not given in Table 1 or the text. These inputs control V_OC, FF, and the efficiency values at the reported operating points. The authors should state all recombination parameters and provide a sensitivity analysis; otherwise the absolute PCEs cannot be assessed. This is a missing-support issue, not a claim that the simulation is circular.","section":"Table 1, §2.2, Eq. (10)"}],"minor_comments":[{"comment":"For position-dependent thermal conductivity, the steady-state heat equation should be written as -∇·(k∇T) = Q (or with the storage term); the form '-k∇²T + Q = ρ_p C_p dT/dt' assumes constant k and uses nonstandard notation ρ_p. Please correct.","section":"Eq. (9)"},{"comment":"Reference [28] is incompletely formatted; the phrase 'The definitive citation to contrast 1D limitations against your 3D COMSOL framework' appears to be an editorial note rather than a citation. Several other references also lack volume/page details.","section":"References"},{"comment":"The text says J_SC increases steadily with thickness and that PCE increases with thickness, yet calls 800 nm the optimized thickness. Please clarify whether the PCE actually peaks at 800 nm or whether the reported 'optimized' point is simply a selected operating point; if the latter, revise the abstract/conclusion accordingly.","section":"§3.2, Fig. 3"},{"comment":"There are numerous grammatical and typographical errors ('achieves', 'fro', 'inaugurating', 'scratched from published literature', inconsistent spacing around equations). A careful language edit is needed.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The internal inconsistency in the 'optimized' claim is serious enough that the paper cannot be accepted as is, but it is addressable by rerunning or re-reporting the simulations at genuinely optimized points and by correcting the optical-model description. I therefore recommend major revision rather than rejection. The forward-modeling nature of the work and the systematic parameter sweeps are strengths; the lack of disclosed recombination parameters and the 1D optical treatment are the main technical weaknesses."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a fairly standard COMSOL device study applied to a promising lead-free perovskite family, with one big internal inconsistency that makes the headline numbers unreliable as reported.\n\nWhat's new and worthwhile: this is the first device-level simulation of ZnSe/MgXS3/Sb2S3 stacks for these three absorbers, with systematic sweeps of thickness, doping, defect density, plus QE, recombination maps, and a coupled electro-thermal analysis. That is useful scoping work. The parameter tables and mesh details are given in enough detail to reproduce most of the model, except for the recombination parameters.\n\nThe main problem: the paper says \"under optimized conditions,\" but the sweeps in Fig. 3(b,c) show higher PCE at 10^19 cm^-3 doping and 10^12 cm^-3 defects than at the claimed optima of 10^17/10^14. Table 1 also lists NA=10^18, contradicting the text. No cost function or joint optimization is given. This is not a minor typo—it undermines the central claim of optimized performance. The thickness sweep also appears monotonic, so choosing 800 nm needs justification. A referee should ask for either a proper optimization procedure or a clear statement that these are \"representative\" rather than optimized values.\n\nSecond: the \"3D\" optical framework is actually 1D Beer-Lambert generation with a Tauc-type absorption coefficient. That is fine for a first pass, but it does not match the \"geometry-aware 3D\" claim and may miss interference or texture effects. The paper should be honest about that.\n\nThird: SRH lifetimes/capture cross-sections and Auger coefficients are never specified, so the absolute efficiency numbers are not reproducible. The heat source uses U_Aug but no Auger parameters are given. That is a reproducibility issue.\n\nAlso, reference [28] contains an unremoved editorial note: \"The definitive citation to contrast 1D limitations against your 3D COMSOL framework.\" That looks like an internal instruction left in the manuscript; it does not kill the science, but it signals the finalization was rushed.\n\nOverall: the paper is a serious, useful simulation study with a clear fixable flaw. I would send it to peer review—the material class matters and the framework is mostly coherent—but only after the authors resolve the optimization inconsistency and provide the missing recombination parameters. If those numbers are not justified, the efficiency values should not be reported as optimized.","headline":"A useful device-level scoping study of MgXS3 perovskites, but the 'optimized' PCEs are contradicted by the paper's own sweeps and the model is less 3D than claimed.","tokens_in":23394,"tokens_out":4122,"would_cite":false,"duration_ms":35851,"reading_group":"maybe","serious_thinker":"no","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Simulations show lead-free MgXS3 perovskites can reach about 28% solar-cell efficiency.","keywords":["chalcogenide perovskites","MgXS3","solar cells","COMSOL simulation","drift-diffusion","lead-free photovoltaics","back surface field","opto-electro-thermal modeling"],"falsifier":"Measure the absorption coefficient of MgZrS3 thin films and run the same device simulation with that measured spectrum; if the resulting J_SC deviates by more than a few mA/cm2 from the 34.46 mA/cm2 predicted here, the headline efficiency claim is not robust.","tokens_in":22248,"feed_emoji":"☀️","tokens_out":2020,"duration_ms":21004,"temperature":0.7,"pith_summary":"This paper argues that three lead-free chalcogenide perovskite compounds, MgTiS3, MgZrS3, and MgHfS3, can serve as high-efficiency solar absorbers when placed in a ZnSe/MgXS3/Sb2S3 cell stack. Using a three-dimensional finite-element model that couples optical generation, charge transport, and heat flow, the authors predict optimized power conversion efficiencies of 26.72%, 28.18%, and 28.16%, respectively, under standard one-sun illumination. The study also claims that the 3D framework reveals spatial effects—like localized recombination heating and interface hot spots—that simpler one-dimensional models cannot capture. If the simulations hold up, these materials offer a nontoxic, thermally stable route beyond lead-halide perovskites.","feed_headline":"Lead-free MgZrS3 solar cells modeled at 28.18% efficiency","feed_subtitle":"3D simulations of MgTiS3, MgZrS3, and MgHfS3 show a nontoxic perovskite path—if real absorption matches the model.","key_machinery":"The load-bearing tool is a 3D finite-element model in COMSOL Multiphysics that self-consistently couples Poisson's equation and drift-diffusion carrier transport with optical photogeneration and steady-state heat transfer. Optical generation is computed from a depth-dependent Beer-Lambert attenuation law using a direct-transition Tauc-type absorption coefficient, and this profile drives the electrical and thermal solutions. The device stack itself—ZnSe window, MgXS3 absorber, Sb2S3 back-surface field—is the second key element, with band offsets chosen to keep conduction- and valence-band discontinuities below 0.3 eV.","core_discovery":"The central claim is that optimized n-ZnSe/p-MgXS3/p+-Sb2S3 solar cells achieve simulated efficiencies of 28.18% for MgZrS3, 26.72% for MgTiS3, and 28.16% for MgHfS3, with open-circuit voltages of 0.94 V, 0.74 V, and 1.07 V and short-circuit current densities of 34.46, 42.69, and 29.89 mA/cm2. The paper further claims that adding an Sb2S3 back-surface-field layer improves rear carrier collection and boosts quantum efficiency, and that coupled electro-thermal simulations show only millikelvin steady-state temperature rises under illumination.","pith_inferences":["The headline efficiencies depend on the assumed absorption shape; using the actual first-principles absorption spectra of MgXS3 instead of a Tauc-model approximation could shift J_SC and PCE noticeably.","A natural next step is to fabricate MgXS3 films and measure their absorption coefficients and carrier lifetimes to test whether the simulated 28% efficiency is physically reachable.","The same coupled 3D framework could be extended to textured or nanostructured rear reflectors to see whether light trapping pushes efficiencies beyond the planar-stack values reported here.","One inserted reference note in the manuscript appears to be an internal editorial comment rather than a normal citation; it does not affect the scientific claims but suggests the reference list was assembled with some automation."],"forward_implications":["If correct, MgZrS3 and MgHfS3 are credible lead-free absorber candidates with efficiencies competitive with current perovskite devices.","The Sb2S3 back-surface-field layer provides a concrete design lever: it raises quantum efficiency by roughly 13-15 percentage points at representative wavelengths.","Doping and defect-density trends give practical guidance: moderate acceptor doping improves voltage and fill factor, while defect densities above 10^14 cm^-3 quickly erode efficiency.","The predicted millikelvin temperature rise suggests these ultrathin stacks would not suffer severe self-heating under normal operation.","The 3D simulation approach, if validated, could replace 1D tools for diagnosing lateral and interfacial losses in thin-film photovoltaics."],"fun_headline_variants":["MgZrS3 solar model hits 28.18% efficiency, lead-free","3D model: lead-free MgXS3 perovskites reach 28.18% PCE","Simulated MgZrS3 cell: 28.18% efficiency without lead","MgZrS3 and MgHfS3 near 28% in lead-free solar simulation","Nontoxic MgXS3 perovskites: 28.18% top efficiency in model"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The optical generation profile is computed with a one-dimensional Beer-Lambert law and a Tauc-type absorption coefficient rather than with the actual wavelength-dependent absorption spectra of MgXS3, so if the real absorption shape differs, the simulated short-circuit current and efficiency will change materially.","fun_headline_variants_meta":{"raw":{"variants":["MgZrS3 solar model hits 28.18% efficiency, lead-free","3D model: lead-free MgXS3 perovskites reach 28.18% PCE","Simulated MgZrS3 cell: 28.18% efficiency without lead","MgZrS3 and MgHfS3 near 28% in lead-free solar simulation","Nontoxic MgXS3 perovskites: 28.18% top efficiency in model"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000705,"raw_usage":{"total_tokens":3129,"prompt_tokens":974,"completion_tokens":2155,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":718,"completion_tokens_details":{"reasoning_tokens":2041}},"tokens_in":718,"tokens_out":2155,"duration_ms":13341,"temperature":1.0,"reasoning_tokens":2041,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T11:43:03.033730+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the absorption coefficient of MgZrS3 thin films and run the same device simulation with that measured spectrum; if the resulting J_SC deviates by more than a few mA/cm2 from the 34.46 mA/cm2 predicted here, the headline efficiency claim is not robust.","supporting_citations":[],"review_version":1}