{"id":"24561d27-d601-45aa-b2b3-ff028d91bd90","arxiv_id":"2608.12878","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"DFT calculations predict the α-MnTe(0001) surface orders with a ferromagnetic outermost Mn bilayer, 17 meV per surface lower in energy than the bulk antiferromagnetic stacking, driven by a ligand hole on Te.","lead":"A first-principles study predicts that the magnetic order at the surface of the altermagnet α-MnTe(0001) reconstructs so the outermost manganese layer pair aligns ferromagnetically, unlike the antiferromagnetic bulk. The driving cause is a single positively charged hole left on a tellurium atom at the clean surface, which flips the magnetic coupling across that bond.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 17.1 meV reconstruction energy is computed only with PBE+U; without a hybrid/SCAN cross-check the surface ground-state ordering is not quantitatively established.","rationale":"The paper makes a coherent, internally well-supported computational prediction: a direct enumeration of 32 collinear interlayer configurations, a layer-resolved exchange fit with small residual, robustness to U_surf and slab thickness, and a linear spin-wave calculation showing the reconstructed state is dynamically stable. These are real strengths. I agree with the reader that the weak point is the DFT approximation, and I do not see evidence that PBE+U with U_eff = 4 eV is accurate to the needed roughly 10 meV. The U_surf sweep changes the margin from -9.8 to -19.1 meV, which shows sensitivity within the same functional family but does not bound errors from missing exact exchange or from treating a charge-transfer insulator with a single U parameter on Mn only. The electron-doping control is a useful mechanistic check but is not a free-energy comparison and uses a charged-cell background at frozen geometry. The paper's own caveat that photoemission shape alone does not select the magnetic state reinforces the lack of experimental discrimination. Therefore the appropriate verdict remains CONDITIONAL: the prediction is plausible and well-structured, but a functional cross-check or a direct surface-sensitive experiment is needed before the 17.1 meV ordering can be treated as quantitative. My recommendation is UNCHANGED relative to the reader's verdict.","tokens_in":14549,"tokens_out":6874,"duration_ms":80703,"concrete_test":"Recompute the A-type versus reconstructed total-energy difference for the 6- and 10-Mn Te-terminated slabs with HSE06 at the PBE+U relaxed geometries, using a reduced but converged k-mesh (e.g., 4x4x1) and the same spin-orbit treatment; if the reconstructed state remains lower by more than about 5 meV per surface at both thicknesses, the PBE+U ordering is robust, whereas a sign change or a margin below numerical noise would show that the headline is functional-dependent.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is the total-energy ordering that places the reconstructed stacking 17.1 meV per 1x1 surface below the bulk-continued A-type state (\"Surface magnetic reconstruction\" section). This ordering is produced by a single approximation: PBE+U in the Dudarev scheme with U_eff = 4.0 eV on Mn 3d (Methods). Alpha-MnTe is a charge-transfer insulator, so errors in the Te 5p versus Mn 3d balance feed directly into the exchange constants. The robustness check in Supplemental Material S2 varies only the Hubbard U on the outermost Mn plane; the margin moves from -9.8 meV at U_surf = 2 eV to -19.1 meV at U_surf = 5 eV, a spread comparable to the quoted 17.1 meV effect, with the smallest margin only about 10 meV. No hybrid functional, SCAN, self-consistent U, or dispersion correction is tested. The electron-doping control (\"Layer-resolved exchange\") is a charged-cell calculation with a uniform compensating background at frozen geometry, so it supports the proposed mechanism but not the absolute ground-state ordering. The photoemission comparison is explicitly stated not to discriminate between the reconstructed and A-type surfaces (Supplemental Material S5: same petal structure, only azimuthal orientation differs, which the published maps do not fix). Thus the quantitative central claim rests on the assumption that PBE+U with U_eff = 4 eV is accurate to better than about 10 meV for a surface exchange energy; that assumption is currently unsupported.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper uses DFT+U (PBE, Dudarev scheme, U_eff = 4 eV) to study the Te-terminated (0001) surface of altermagnetic α-MnTe. It reports that the surface magnetic ground state is not the bulk-continued A-type stacking but a reconstructed stacking in which the outermost Mn bilayer at each surface is ferromagnetic, lying 17.1 meV per 1×1 surface below the A-type state. The mechanism is attributed to the ligand hole in the Te dangling bonds: the hole is localized on the Te that mediates the exchange between the two outermost Mn planes, and it reverses that bond ferromagnetically. Electron doping, modeled by adding one electron per surface with a uniform background at frozen geometry, restores the A-type ordering. The authors also compute constant-energy photoemission contours and show consistency with published surface-sensitive maps, while explicitly noting that those maps do not discriminate between the reconstructed and A-type stackings. Additional checks reported include slab-thickness dependence, a sweep of the surface Hubbard U, a layer-resolved exchange fit, a spin-cluster expansion stability test, and one-face passivation calculations.","tokens_in":14757,"tokens_out":7655,"duration_ms":84019,"significance":"If the central ordering survives closer scrutiny, this is an important and somewhat surprising result for altermagnet surfaces: the surface magnetic order differs from the bulk, with direct consequences for surface-dominated transport, spectroscopy, and magnetometry, and it produces an uncompensated surface moment exchange-coupled to the Néel vector. The paper is methodologically transparent in several respects: it enumerates all 32 inequivalent interlayer configurations, reports a layer-model fit with sub-meV residual, tests slab thickness and surface U_eff, provides a spin-wave stability check, and states honestly that the photoemission comparison is non-discriminating. These are genuine strengths. The main weakness is that the central 17 meV-scale energy difference is computed with a single exchange-correlation functional, and the only causal control (electron doping) uses a charged-cell setup whose robustness is not established.","major_comments":[{"comment":"The central claim, that the reconstructed stacking lies 17.1 meV per 1×1 surface below the bulk-continued A-type state, is computed only with PBE+U at U_eff = 4 eV. The robustness check in SM S2 varies U_eff only on the outermost Mn plane and changes the margin from -9.8 to -19.1 meV per surface, a spread comparable to the quoted effect. No hybrid functional, SCAN, self-consistent U, or dispersion correction is tested. Because α-MnTe is a charge-transfer insulator and the hole is mostly on Te, the relative Te 5p / Mn 3d energetics is exactly the part of the functional most in doubt for this ordering. Please add at least one independent functional cross-check (for example SCAN or a hybrid calculation on a representative slab) or provide an explicit uncertainty analysis that justifies quoting a 10 meV-scale surface energy difference.","section":"Methods; Surface magnetic reconstruction; SM S2"},{"comment":"The electron-doping control that fills the ligand hole is performed in a charged slab with a uniform compensating background at frozen geometry, and it is the only calculation that reverses the energy ordering (+43.8 meV per surface) and shifts J_12 by -41.2 meV. Total energies of charged supercells with jellium background are not thermodynamically well-defined and can depend on cell shape and vacuum size, and the frozen-geometry approximation omits the relaxation channel that is part of the clean-surface result. The related one-face-passivated calculation in SM S4 reports 23.0 meV per surface for the same surface reversal, whereas the clean symmetric slab gives 17.1 meV per surface; this discrepancy is not discussed. Please validate the electron-doping control with a local charge-compensation scheme or with varying slab/vacuum sizes, allow relaxation, and clarify the origin of the 23.0 versus 17.1 meV difference.","section":"Layer-resolved exchange; SM S4"}],"minor_comments":[{"comment":"The abstract and conclusion state that the computed constant-energy contours are 'consistent with photoemission maps'; the text itself correctly notes that the reconstructed and A-type petal patterns differ only in azimuthal orientation and that the published maps do not fix it. Please make this non-discriminating character explicit in the abstract and conclusion so that the wording is not read as validation of the reconstruction.","section":"Abstract; Constant-energy contours; SM S5"},{"comment":"The horizontal axis label 'Number of Mn layer' should be 'Number of Mn layers'.","section":"Fig. 1(b)"},{"comment":"The sentence 'The residual is 0.8 meV on a configuration set spanning 238 meV' should specify whether this is an RMSE or a maximum error; SM S1 reports an RMSE of 0.82 meV, so the main text should match that definition.","section":"Layer-resolved exchange; SM S1"},{"comment":"The statement that data are available from the first author upon reasonable request would be strengthened by depositing slab structures, input files, and convergence data in a public repository, especially since the paper makes quantitative surface-energy claims that others may want to reproduce.","section":"Data availability"}],"recommendation":"major_revision","confidential_remarks":"The paper is within scope and the authors are unusually candid about the limits of the photoemission comparison. My recommendation of major revision rests on two load-bearing methodological gaps: the single-functional dependence of the central 17 meV energy difference and the unvalidated charged-cell control for the hole-filling experiment. Neither seems fatal; if the authors add an independent functional cross-check and a more robust treatment of the electron-doping setup, I would support publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe paper you'll want to know about: Tanaka and Gohda predict that the (0001) surface of the altermagnet α-MnTe reconstructs its magnetic order. The outermost Mn bilayer couples ferromagnetically, inverting the bulk A-type stacking, by about 17 meV per 1×1 surface. The driver is a stoichiometric ligand hole in the Te dangling bonds, which sits on the Te that mediates the interlayer exchange and flips that bond's sign. The result is a surface with an uncompensated moment coupled to the Néel vector, which matters if you interpret surface-sensitive transport or spectroscopy on MnTe films.\n\nWhat's good: The central energy ordering is a direct total-energy difference, not a fitted quantity. They enumerate all 32 collinear interlayer configurations of a 6-Mn slab, relax each in its own geometry, and the reconstruction wins. It survives slab thickness from 6 to 10 Mn layers and surface U_eff from 2 to 5 eV. The electron-doping control — adding one electron per surface at frozen geometry — reverses the ordering and moves J12 by −41 meV while other bonds barely shift, which is a clean causal test of the ligand-hole mechanism. They also extend the check beyond collinear order: a spin-cluster expansion fitted to constrained noncollinear calculations gives a spin-wave spectrum with no negative modes, so the reconstruction is a local minimum against finite-wavevector deviations, not just the best of the 32 stackings. The paper is unusually honest about what its photoemission comparison can and cannot show: the computed constant-energy contours match the observed pockets, but the reconstructed and A-type surfaces differ only in azimuthal orientation, which published maps do not fix. The surface grand potential treatment that avoids an elemental Mn reference is also careful and sensible.\n\nSoft spots: The quantitative margin is computed entirely with PBE+U at U_eff=4 eV. The paper's own sweep changes the surface U_eff from 2 to 5 eV and the energy moves from −9.8 to −19.1 meV per surface; the spread is comparable to the quoted 17 meV. A hybrid-functional or SCAN cross-check would make the ordering claim much more solid. The electron-doping control uses a charged cell with a uniform compensating background, so it supports the mechanism but not the absolute ground state. There is no archived code or data artifact, just 'available upon request.' None of these are fatal: the sign of the ordering is stable across the tested range, and the mechanism is physically plausible. But the exact energy is not nailed down, and a real referee should push for a functional cross-check before publication.\n\nMy read: the paper is a solid, well-scoped first-principles prediction, honestly presented, with a plausible mechanism and a testable experimental consequence (photon-energy scan, magnetometry on uncapped films). It deserves serious peer review. I'd send it out, and I'd ask for the functional check in revision. I'd bring it to the reading group — it's a good example of a surface effect that could complicate altermagnet transport experiments, and I'd cite it if I worked on MnTe surfaces.","headline":"Solid first-principles prediction of a surface magnetic reconstruction in α-MnTe(0001), with a clean ligand-hole mechanism and honest caveats; the quantitative margin needs a functional cross-check.","tokens_in":15375,"tokens_out":2836,"would_cite":true,"duration_ms":27741,"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":"The clean (0001) surface of the altermagnet α-MnTe reverses its outermost Mn bilayer to ferromagnetic order, driven by a single stoichiometric ligand hole on the mediating Te.","keywords":["altermagnetism","alpha-MnTe (0001) surface","magnetic surface reconstruction","ligand hole","ferromagnetic Mn bilayer","A-type antiferromagnet","density functional theory","photoemission"],"falsifier":"Two measurements would settle it. Scanning-probe magnetometry on a clean, uncapped Te-terminated α-MnTe(0001) film should find an in-plane uncompensated moment of order 8.7 $\\mu_B$ per surface cell, present only on the hole-bearing termination; and a photon-energy-dependent ARPES scan across the Γ pockets should show no $k_z$ dispersion if they are the surface states the reconstruction predicts, with the surface petal sheets rotated relative to the bulk A-type stacking.","tokens_in":14256,"feed_emoji":"🧲","tokens_out":17217,"duration_ms":147610,"temperature":0.7,"pith_summary":"This paper predicts from first principles that the clean tellurium-terminated (0001) surface of the altermagnet α-MnTe does not continue the bulk antiferromagnetic stacking: the outermost Mn bilayer reverses to ferromagnetic order, 17.1 meV per $1\\times1$ surface below the bulk-continued A-type state. The driver is a single hole per surface cell left in the Te dangling bonds by the polar termination, a carrier density fixed by stoichiometry rather than by doping, and in this charge-transfer insulator the hole sits on the Te that mediates the exchange joining the two surface Mn planes. A half-filled Mn $d^5$ shell can only fill that hole with a majority-spin electron, so both Mn neighbors of the hole-bearing Te must point parallel, making the bond ferromagnetic; filling the hole electronically restores the bulk order. The computed constant-energy contours match published photoemission pockets with no adjustable quantity. Transport and spectroscopy on films therefore read a surface magnetic order with an uncompensated in-plane moment, not the bulk altermagnetic order.","feed_headline":"One missing electron flips MnTe's surface to ferromagnetic","feed_subtitle":"One hole per surface cell on a mediating Te forces the outer Mn pair parallel; films read this order, not the bulk.","key_machinery":"The central object is the ligand hole created by the polar termination: one hole per $1\\times1$ surface, fixed by stoichiometry, residing on the Te $5p$ shell because α-MnTe is a charge-transfer insulator. The argument is carried by two pieces of machinery. The first is the layer-resolved spin model $E = E_0 - \\sum_{l<l'} J_{ll'}\\sigma_l\\sigma_{l'}$, fitted to the complete enumeration of 32 collinear interlayer configurations, which isolates $J_{12}=+17.1$ meV as the only ferromagnetic interlayer bond against antiferromagnetic subsurface bonds. The second is the hole-mediated exchange principle: a half-filled $d^5$ Mn fills the ligand hole with a majority-spin electron, so both Mn neighbors of the hole-bearing Te must have parallel moments, and filling the anion shell closes that channel, returning the bond to the antiferromagnetic coupling of the bulk — which is why the electron-doped slab reverts to A-type order.","core_discovery":"Enumerating all 32 inequivalent collinear interlayer spin configurations of a six-Mn-layer Te-terminated slab, the ground state is not the bulk-continued A-type stacking but the stacking obtained by reversing the outermost Mn plane at each surface, so the two outermost Mn planes couple ferromagnetically while the interior keeps the A-type order; relaxed in its own geometry the reconstructed state lies 17.1 meV per $1\\times1$ surface below the A-type slab, and the ordering survives slabs of 8 and 10 Mn layers and surface $U_{\\rm eff}$ from 2 to 5 eV. Fitting the enumeration to the layer-resolved model $E = E_0 - \\sum_{l<l'} J_{ll'}\\sigma_l\\sigma_{l'}$ gives a ferromagnetic $J_{12}=+17.1$ meV against antiferromagnetic subsurface bonds ($J_{23}=-26.8$, $J_{34}=-31.0$ meV), with a residual of 0.8 meV over a 238 meV ladder. The reversal is traced to the ligand hole: electron counting fixes one hole per surface cell on the dangling Te bonds, the near-Fermi states carry 83% of their weight on Te, and the Te mediating the $J_{12}$ bond is the most hole-rich mediator in the slab. Refilling the hole electronically turns the reconstruction energy positive (+43.8 meV per surface) and moves $J_{12}$ by $-41.2$ meV while every other exchange constant changes by at most 8.9 meV, so the carrier acts on the one bond whose mediating Te carries it.","pith_inferences":["A reversible control knob suggests itself: gating, adsorbates, or photoexcitation that fills or depletes the surface hole should toggle the surface between ferromagnetic bilayer and bulk A-type order, switching the sign of the surface spin splitting without disturbing the bulk Néel vector.","The mechanism is not specific to MnTe: any polar chalcogenide termination whose dangling bonds leave a half-filled anion shell on the mediator of a magnetic bond could reconstruct magnetically, so surface magnetic order should be enumerated rather than assumed to continue the bulk.","The two surface magnon branches expelled above the bulk band (up to 62 meV against a measured bulk top near 36 meV) are a spectroscopic fingerprint; surface-sensitive neutron or electron scattering on films could observe surface excitations above the bulk continuum.","If the 17 meV ordering survives a functional without +U tuning, the reconstructed surface offers a controlled platform for studying altermagnetic symmetry breaking at an interface, since it breaks the bulk sublattice relation while preserving the lattice periodicity."],"forward_implications":["Surface- and interface-derived transport in α-MnTe films — conduction that beyond a few unit cells belongs to surface states — must be analyzed with the reconstructed surface order rather than the bulk A-type order continued to the surface.","The compensated altermagnet acquires an uncompensated in-plane moment of about 8.68 $\\mu_B$ per $1\\times1$ surface, exchange-coupled to the Néel vector and confined to the termination, giving surface-sensitive probes a direct handle on an order parameter no magnetic field couples to.","Across the reversed outermost bond the two spin sublattices are parallel rather than antiparallel, so the rotation relation that defines altermagnetism fails at the surface.","The computed constant-energy contours reproduce the surface-sensitive photoemission pockets labeled α and α₁ with no adjustable quantity, and a photon-energy scan across the Γ pockets would test their surface character directly.","Experimental support already exists: Li doping, which makes the crystal more hole-rich, weakens the antiferromagnetic interlayer exchange by 8% in inelastic neutron scattering."],"supporting_citations":[{"why":"Supplies the surface-sensitive photoemission maps (the α and α₁ pockets) whose constant-energy contours the reconstruction reproduces, and the evidence that film conduction beyond a few unit cells belongs to surface states.","marker":"[13]"},{"why":"The prior bulk exchange study of α-MnTe whose $U_{\\rm eff} = 4$ eV setting the calculation carries over for the Mn 3d shell.","marker":"[9]"},{"why":"Provides the bulk-sensitive six-petaled ARPES maps at $k_z = 0$ that the altermagnetic bulk and the reconstructed surface must both be consistent with.","marker":"[10]"},{"why":"Establishes α-MnTe as a charge-transfer insulator, so the surface hole resides on the Te ligand rather than changing the Mn valence.","marker":"[17]"},{"why":"Supplies the hole-mediated ferromagnetic exchange mechanism from p-type dilute magnetic semiconductors that drives the $J_{12}$ reversal.","marker":"[22]"},{"why":"Identifies the antiferromagnetic interlayer exchange of bulk MnTe that the bond returns to once the ligand shell is filled.","marker":"[41]"},{"why":"The Li-doping neutron-scattering result that hole doping weakens the antiferromagnetic interlayer exchange by 8%, quantitative agreement with the mechanism.","marker":"[44]"},{"why":"The electron-counting model that fixes one hole per $1\\times1$ surface on the dangling Te bonds.","marker":"[15]"}],"fun_headline_variants":["One Te ligand hole rewrites MnTe surface magnetic order","Ligand hole drives MnTe surface to ferromagnetic stacking","MnTe surface magnetism flips via a single Te dangling hole","Ferromagnetic MnTe surface: one hole on Te does it"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The ordering that decides the ground state — 17.1 meV per surface cell — is computed with a single approximation, PBE+U at $U_{\\rm eff} = 4$ eV, whose error for surface energy differences of this size can be tens of meV, and no other functional or dispersion correction is tested.","fun_headline_variants_meta":{"raw":{"variants":["One Te ligand hole rewrites MnTe surface magnetic order","Ligand hole drives MnTe surface to ferromagnetic stacking","MnTe surface magnetism flips via a single Te dangling hole","Ferromagnetic MnTe surface: one hole on Te does it"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000578,"raw_usage":{"total_tokens":2749,"prompt_tokens":992,"completion_tokens":1757,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":608,"completion_tokens_details":{"reasoning_tokens":1687}},"tokens_in":608,"tokens_out":1757,"duration_ms":13925,"temperature":1.0,"reasoning_tokens":1687,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T21:28:08.185718+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Two measurements would settle it. Scanning-probe magnetometry on a clean, uncapped Te-terminated α-MnTe(0001) film should find an in-plane uncompensated moment of order 8.7 $\\mu_B$ per surface cell, present only on the hole-bearing termination; and a photon-energy-dependent ARPES scan across the Γ pockets should show no $k_z$ dispersion if they are the surface states the reconstruction predicts, with the surface petal sheets rotated relative to the bulk A-type stacking.","supporting_citations":[{"cited_title":"Alaei, P","cited_arxiv_id":null,"evidence_quote":"The prior bulk exchange study of α-MnTe whose $U_{\\rm eff} = 4$ eV setting the calculation carries over for the Mn 3d shell."},{"cited_title":"Krempask´ y, L","cited_arxiv_id":null,"evidence_quote":"Provides the bulk-sensitive six-petaled ARPES maps at $k_z = 0$ that the altermagnetic bulk and the reconstructed surface must both be consistent with."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes α-MnTe as a charge-transfer insulator, so the surface hole resides on the Te ligand rather than changing the Mn valence."},{"cited_title":"Ferrand, J","cited_arxiv_id":null,"evidence_quote":"Supplies the hole-mediated ferromagnetic exchange mechanism from p-type dilute magnetic semiconductors that drives the $J_{12}$ reversal."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The electron-counting model that fixes one hole per $1\\times1$ surface on the dangling Te bonds."}],"review_version":1}