{"id":"516f4693-d317-4fe7-ba83-9ad35b7f49ba","arxiv_id":"2607.27095","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Band-avoiding occupation-constrained DFT yields converged InAs point-defect charge-transition levels despite a zero PBE gap by forcing occupation of localized defect states instead of the spurious Γ conduction-band edge.","lead":"A DFT trick that skips the collapsed conduction-band edge lets researchers compute atomic defect levels in InAs even when the calculated band gap is zero. That matters for radiation-hard infrared detectors and other narrow-gap electronics where standard DFT and hybrid methods struggle.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.5","headline":"Central claim still hinges on correct Γ-state character under forced non-Aufbau occupation; internal diagnostics help but do not fully close the gap.","rationale":"The reader correctly isolated the load-bearing assumption (Γ character / avoided-crossing occupation hypothesis). The manuscript’s numerical coherence (supercell flattening, bond-length charge diagnostics, LMCC electrostatics, failure of plain Aufbau at Γ) makes the method plausible and internally consistent with the authors’ prior Si/GaAs LMCC work; it does not, however, replace a direct character proof or a cross-check against a calculation in which the gap is open and constraints are unnecessary. External validation remains thin (loose DLTS associations, qualitative shallow-shoulder reading), so “rigorous” is still ahead of the evidence. No stronger internal contradiction or hidden circularity appears. The appropriate stance remains CONDITIONAL on clearer orbital-character evidence and public artifacts; the stress test does not move the verdict.","tokens_in":10908,"tokens_out":641,"duration_ms":43859,"concrete_test":"In the 512- or 1000-atom AsIn(0) and AsIn(1+) cells, project the Γ eigenstates (Aufbau vs ba-occ) onto antisite-centered and bulk-like orbitals (or construct defect Wannier functions). Require the ba-occ occupied state to be >~90% localized on the AsIn dangling-bond manifold and the emptied state to be delocalized CBE-like. Separately, recompute the (0/+) total-energy difference with a minimal gap-opening perturbation (scissor or affordable HSE on the 216-atom cell) without occupation constraints; if the level moves by more than ~0.1 eV vs ba-occ-PBE, band-edge error is not cleanly separated.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The strongest claim (ba-occ-DFT yields uncorrupted localized-defect total energies, hence rigorous levels despite a collapsed PBE gap) rests on the Fig. 1(d) hypothesis: that the physical defect orbital continues through Γ above the spuriously lowered CBE, and that emptying the CBE while occupying that orbital recovers the true defect ground-state energy. Internal support is real—defect-band flattening with supercell size, avoided-crossing language, As–As bond lengths that track charge only under ba-occ (Γ-only Aufbau collapses all charges to the 2+ geometry), and near-agreement of Γ-(2×2×2) vs off-Γ formation energies—but none of these is a direct proof of orbital character or of adiabatic equivalence between the constrained KS state and the physical gapped defect. If the Γ states remain hybridized, or if the constraint energy is not transferable once the experimental gap is restored by hand (VBE from InAs(2+/1+), CBE from experiment), every level in Fig. 4 shifts by an uncontrolled amount. That is the single point on which the “no band gap, no problem” claim is least secure.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.5","summary":"The manuscript introduces band-avoiding occupation-constrained DFT (ba-occ-DFT): at Γ, non-Aufbau occupations empty the spuriously collapsed conduction-band-edge (CBE) state and occupy the putative localized defect orbital, while off-Γ points use standard Aufbau. Combined with LMCC charged-supercell boundary conditions and PBE, the method is applied to intrinsic point defects in InAs (antisites, vacancies, divacancy, interstitials) in supercells up to 1000 atoms. The authors report converged defect transition levels despite a zero PBE gap, an effective defect band gap bounded by clean defect states, and qualitative consistency with sparse DLTS data (including a shallow-donor interpretation of a broad emission shoulder). HSE06 band structures in small cells are shown not to remove defect-band dispersion or Γ hybridization.","tokens_in":11200,"tokens_out":1080,"duration_ms":45471,"significance":"If the constrained total energies are physically faithful, the work supplies a practical route to defect levels in narrow-gap III–Vs where standard DFT gaps collapse and hybrid+jellium supercell calculations remain costly. The internal convergence campaign (64→1000 atoms, defect-band flattening, Γ-only vs off-Γ and Aufbau vs ba-occ bond-length diagnostics for AsIn) is careful and falsifiable. Predictions that primary displacement defects terminate as shallow donors offer a concrete experimental target. The approach extends a previously benchmarked total-energy/LMCC framework (Si, GaAs) rather than introducing an uncontrolled free parameter for the gap, which is a genuine methodological strength.","major_comments":[{"comment":"Central hypothesis (Fig. 1d and text): the claim that ba-occ-DFT yields uncorrupted localized-defect ground-state energies rests on inverted CBE/defect character at Γ and on adiabatic equivalence of the constrained KS state to the physical gapped defect. Main-text support is indirect (supercell flattening, avoided-crossing language, As–As bond lengths that track charge only under ba-occ, near-equality of Γ-(2×2×2) and off-Γ formation energies). A direct main-text diagnostic—e.g., real-space localization or projection of the occupied Γ orbital onto bulk Bloch/CBE character under both Aufbau and ba-occ—should be added (or the SM result elevated and quantified). Without it, absolute levels in Fig. 4 carry an unquantified systematic risk if residual hybridization remains.","section":"Fig. 1(d); AsIn discussion and SM cross-ref"},{"comment":"Fig. 4 level diagram: VBE is bounded by InAs(2+/1+) and CBE is placed at the experimental gap, while an “effective defect band gap” is defined from the highest clean defect state. Charge-transition total-energy differences are method-internal; absolute placement relative to the true edges is not. The text should state explicitly which reported numbers are independent of the experimental-gap alignment, quote the numerical uncertainty from the shallow-bound choice of VBE, and avoid language (“rigorous defect level predictions”) that blurs this distinction.","section":"Fig. 4 and surrounding paragraphs"}],"minor_comments":[{"comment":"Several run-on or missing-space tokens appear in the compiled text (e.g., “extremeinnarrow-gapsemiconductors”, “bandgapproblem”, “END MA TTER”). A full proofreading pass is needed.","section":"passim; End Matter heading"},{"comment":"Fig. 2–3 band-structure panels would be clearer with an explicit legend for occupied vs empty states under ba-occ and a horizontal marker for the constrained defect level at Γ.","section":"Figs. 2 and 3"},{"comment":"The experimental associations (Salman; Murawski) are appropriately hedged as non-conclusive; consider adding one sentence on what measurement (e.g., stress splitting, annealing kinetics, or ODMR) would discriminate the predicted AsIn second donor near the VBE.","section":"Discussion of experimental comparison"},{"comment":"End Matter: briefly state whether spin-orbit coupling was tested for bulk InAs edges or any defect, even if only to justify its omission for the present total-energy differences.","section":"End Matter"}],"recommendation":"minor_revision","confidential_remarks":"Fit and novelty are appropriate for a condensed-matter/materials journal Letter. Self-citation of the authors’ LMCC/Si/GaAs corpus is methodologically necessary, not decorative. The two major points are clarification and evidence-elevation, not a rewrite of the method; I would not block acceptance once they are addressed. I have not seen the SM in full; if the SM already contains quantitative Γ projections, elevating a short summary into the main text would resolve the first major comment with minimal effort."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"The one thing to know: they show you can still get usable charged-defect total energies in PBE InAs when the gap has collapsed, by skipping the spurious CBE at Γ and occupying the flatter defect-like state (ba-occ-DFT), then finishing with their usual LMCC electrostatics. That is a real methods extension of their Si/GaAs LMCC line, not just a rebrand.\n\nWhat is new is the Γ-specific band-avoiding constraint for a zero DFT gap, the side-by-side that HSE06 in 64/216-atom cells still hybridizes and disperses the “defect” band badly, and a full primary-defect survey (antisites, vacancies, divacancy, interstitials) with a coherent shallow-donor reading of the broad DLTS shoulder. The internal work is the strong part: 64→1000 atom cells, defect bands flattening, As–As bond lengths that only track charge under the constraint, Γ-only Aufbau collapsing everything to AsIn(2+), and off-Γ vs constrained Γ-(2×2×2) energies within tens of meV. They cleanly separate defect total-energy differences from where you later paint the experimental CBE. Circularity is low; the gap is mostly a drawing aid in Fig. 4.\n\nSoft spots in proportion: the load-bearing hypothesis is still that the physical localized orbital continues through Γ above the collapsed CBE and that the constrained KS energy is the right ground-state energy. Flattening, avoided-crossing language, and bond-length diagnostics support that picture but do not prove orbital character or adiabatic equivalence once the real gap is restored by hand. Experimental anchors are loose (two near-CBE peaks called consistent with AsIn(0/+), plus a qualitative shoulder). “Rigorous” is ahead of the data. Code is unpublished SeqQuest; SM tables are cited but not in the text we have. None of that sinks the Letter; it just caps how hard you should lean on absolute level positions.\n\nThis is for people who already do supercell defect levels in III–Vs or radiation effects in IR materials. Worth a serious referee. I would engage: read the SM when it appears, try the occupation trick in another code if you work narrow-gap hosts, and treat the InAs diagram as a well-converged PBE+LMCC baseline rather than final spectroscopy assignments.","headline":"Practical occupation fix for collapsed-gap defect totals in InAs; internal numerics are careful, external ID of levels is still thin.","tokens_in":11860,"tokens_out":586,"would_cite":true,"duration_ms":20485,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.5","headline":"A constrained occupation scheme lets standard DFT predict defect levels in InAs even when the computed band gap collapses to zero.","keywords":["density functional theory","band gap problem","point defects","InAs","occupation-constrained DFT","charge transition levels","narrow-gap semiconductors"],"falsifier":"A direct experimental identification of the As-antisite (0/+) and (+/2+) levels in InAs (or a large-supercell hybrid-functional calculation that cleanly resolves the same Γ-point state ordering) that disagrees with the predicted 0.35 eV and 0.12 eV positions would falsify the method.","tokens_in":11739,"feed_emoji":"⚛️","tokens_out":887,"duration_ms":20676,"temperature":0.7,"pith_summary":"Standard density functional theory badly underestimates semiconductor band gaps, and in narrow-gap materials such as indium arsenide the gap can vanish entirely. That collapse has long been thought to block reliable first-principles predictions of the charge-transition energies of atomic defects. This paper argues that the difficulty is not fatal: the localized defect orbital still exists above the spuriously lowered conduction-band edge, and one can simply force the calculation to occupy that defect state while leaving the fake band-edge empty. The resulting band-avoiding occupation-constrained DFT yields clean total energies and defect levels for the common intrinsic defects in InAs. The levels are consistent with the sparse experimental clues that exist and explain why irradiated InAs shows a broad shallow-donor signature. If the method holds, quantitative defect chemistry becomes possible in a whole class of infrared and high-mobility materials that were previously off-limits to ordinary DFT.","feed_headline":"DFT predicts InAs defect levels even with a zero band gap","feed_subtitle":"Forcing occupation of the localized defect orbital, not the collapsed band edge, restores clean total energies","key_machinery":"ba-occ-DFT: a Γ-specific non-Aufbau occupation constraint that skips the collapsed conduction-band-edge state and forces occupation of the flat, localized defect state, used together with local-moment countercharge boundary conditions and adequate k-point sampling.","core_discovery":"Band-avoiding occupation-constrained DFT (ba-occ-DFT) separates band-edge errors from defect energetics: by emptying the spuriously collapsed conduction-band edge at Γ and occupying the localized defect orbital instead, one obtains uncorrupted total energies and therefore rigorous charge-transition levels for defects in InAs despite a zero DFT gap.","pith_inferences":["The same Γ-point occupation switch should transfer directly to InSb, InAsSb, and related type-II superlattice detector materials where DFT gaps also collapse.","Once formation energies are trustworthy, multiscale kinetic models of radiation damage evolution (already demonstrated for GaAs) become feasible for InAs-based devices.","Codes that already support constrained occupations can implement ba-occ-DFT with only a k-point-dependent occupation mask, lowering the barrier to adoption."],"forward_implications":["Intrinsic primary defects in InAs terminate at stable (1+) charge states and therefore all act as shallow donors, explaining the broad DLTS shoulder seen after irradiation.","The As antisite remains a low-formation-energy double donor that can appear in as-grown material, analogous to EL2 in GaAs.","Ordinary PBE plus the occupation constraint and proper charged-supercell boundary conditions can replace hybrid functionals for defect levels in other zero-gap or narrow-gap III–V alloys.","The same occupation switch is unnecessary for valence-band-edge crossings; the dominant error is over-stabilization of the delocalized conduction edge, not defect delocalization."],"fun_headline_variants":["ba-occ-DFT yields InAs defect levels despite zero gap","Occupation constraints separate band errors from InAs defects","Band-avoiding DFT restores clean defect energies in InAs","Zero DFT gap no barrier to InAs charge-transition levels","Emptying collapsed edge enables rigorous InAs defect DFT"],"cache_read_input_tokens":128,"weakest_assumption_plain":"The true localized defect orbital really does continue through the zone center above the fake conduction-band edge, so that forcing its occupation yields the physical defect ground-state energy rather than an artifact.","fun_headline_variants_meta":{"raw":{"variants":["ba-occ-DFT yields InAs defect levels despite zero gap","Occupation constraints separate band errors from InAs defects","Band-avoiding DFT restores clean defect energies in InAs","Zero DFT gap no barrier to InAs charge-transition levels","Emptying collapsed edge enables rigorous InAs defect DFT"]},"model":"grok-4.5","effort":"low","cost_usd":0.005135,"raw_usage":{"total_tokens":1387,"prompt_tokens":688,"num_sources_used":0,"completion_tokens":65,"cost_in_usd_ticks":51348000,"prompt_tokens_details":{"text_tokens":688,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":634,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":688,"tokens_out":65,"duration_ms":11775,"temperature":1.0,"reasoning_tokens":634,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-30T11:16:52.517226+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"A direct experimental identification of the As-antisite (0/+) and (+/2+) levels in InAs (or a large-supercell hybrid-functional calculation that cleanly resolves the same Γ-point state ordering) that disagrees with the predicted 0.35 eV and 0.12 eV positions would falsify the method.","supporting_citations":[],"review_version":1}