{"id":"47c5700b-3768-458f-8bb8-a9a36997724f","arxiv_id":"2412.16816","paper_version":3,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Self-consistent Hubbard parameters or hybrid functionals are needed to predict the correct ground state of LiMnO2; standard DFT+U fails, and the failure is traced to Jahn-Teller ordering and electron localization.","lead":"This paper tests which computer models of materials correctly predict the most stable form of LiMnO2, a candidate battery cathode material, and finds that standard methods fail while more expensive methods with self-consistent corrections succeed. The findings help battery researchers choose reliable modeling tools for manganese-based cathodes.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Incomplete magnetic search may bias gamma-LiMnO2 upward: its HSE06 margin over ortho is only ~1 meV/atom, so a missed AFM ordering could flip the ground state.","rationale":"The reader's verdict (CONDITIONAL) already identifies the magnetic-order enumeration as a key weakness. I agree and single it out as the most load-bearing because the central claim is a 0 K phase-stability ranking validated against experiment. The finite-temperature phonon mixing (PBEsol+U phonons added to PBEsol+(U+V)sc energies) is a real but secondary concern: it affects the temperature-dependent trends (Figure 4), not the 0 K conclusion that HSE06/self-consistent U recover ortho. The magnetic ordering, by contrast, enters every single total-energy comparison at 0 K. The paper gives strong independent support elsewhere: the ortho AFM ground state matches neutron refinement, the self-consistent U values correlate linearly with Mn moments (R^2=0.94), the collinear JT versus zig-zag comparison is controlled, and the G0W0 band structure validates the HSE06 electronic structure. But none of these checks validates the completeness of the gamma magnetic search. If the gamma magnetic ground state is richer than the ~30 collinear orders tried, the conclusion that gamma is spurious is not yet established, and the HSE06 margin is small enough that even a modest correction could reverse it. The proposed test is therefore not a generic 'try harder' request; it targets the exact energy window (1-5 meV/atom) in which the paper's central distinction operates.","tokens_in":34645,"tokens_out":4252,"duration_ms":36970,"concrete_test":"Recompute gamma-LiMnO2 (and disordered layered) with an expanded magnetic search: (i) enumerate at least 100 collinear orderings on 2x2x2 and 4x2x2 supercells plus 4-6 noncollinear configurations using PBEsol+U, r2SCAN+U, and HSE06; (ii) relax the lowest-energy candidate from each method with HSE06 and PBEsol+Usc(+Vsc) to obtain consistent magnetic ground states; (iii) compare gamma-ortho energy differences. If gamma drops by >5 meV/atom relative to ortho or becomes the ground state in HSE06, the central claim is falsified. If gamma remains >5 meV/atom above ortho in HSE06 and >10 meV/atom in PBEsol+Usc, the current conclusion is supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that only HSE06 and self-consistent Hubbard U/V (PBEsol+Usc/Vsc) recover orthorhombic (Pmmn) LiMnO2 as the ground state, while empirical +U methods spuriously stabilize gamma-LiMnO2. All phase energies in Figures 2-4 are evaluated at the single collinear AFM ordering selected for each phase by enumerating ~30 collinear AFM orderings relaxed with PBEsol+U (U=3.9 eV; Section II, SI S6). The ortho ordering matches neutron data, but gamma-LiMnO2 and disordered layered have no experimental magnetic reference. If the true magnetic ground state of gamma has a different propagation vector (e.g., larger supercell, incommensurate, or noncollinear) not in the enumeration, its computed total energy is an upper bound. Because PBEsol+U is itself the functional the paper argues mis-orders the phases, using it to pre-screen magnetic order for all subsequent functionals risks propagating a systematic bias: the comparison functional is not being tested on each phase's actual magnetic ground state. The fragility is concrete: HSE06 places gamma only ~1 meV/atom above ortho (Figure 3). A missed ordering that stabilizes gamma by just 2-3 meV/atom would make gamma the HSE06 ground state, directly contradicting the abstract's claim that HSE06 recovers the experimental phase stability. PBEsol+Usc places gamma ~25 meV/atom above, a larger margin, but the same concern applies: the relative gap could shrink substantially if gamma's magnetic ordering is incomplete. The paper's own data (Figure 2c) show AFM order stabilizes phases by 5-25 meV/atom depending on ordering, so the energy scale of a missed magnetic configuration is comparable to the HSE06 gamma-ortho separation.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents an ab initio study of the phase stability of LiMnO2 polymorphs, focusing on the interplay among electron localization, magnetic order, and Jahn-Teller (JT) distortions. Using PBEsol+U, r2SCAN+U, HSE06, and PBEsol with self-consistently determined Hubbard U and V parameters, the authors find that empirical Hubbard U methods spuriously predict a gamma-LiFeO2-like phase (gamma-LiMnO2) as the ground state, whereas HSE06 and PBEsol+Usc(+Vsc) correctly recover orthorhombic (Pmmn) LiMnO2. The self-consistent U values are shown to correlate linearly with Mn magnetic moments, and smaller U values are associated with collinear JT ordering, increased Mn-O covalency, and enhanced vibrational entropy. The paper also reports large HSE06 and G0W0 band gaps (>3 eV) for ortho-LiMnO2 and identifies a previously unreported low-energy epsilon-LiMnO2 phase.","tokens_in":34919,"tokens_out":7432,"duration_ms":63463,"significance":"If correct, the central claim is significant for the DFT community: it demonstrates that empirical Hubbard U parameters can be qualitatively wrong for phase stability in a Jahn-Teller active oxide, and that self-consistent Hubbard parameters (or hybrids) are necessary. The paper's strengths include the use of multiple independent electronic-structure methods (HSE06, PBEsol+Usc, PBEsol+(U+V)sc, and r2SCAN+Usc in the SI) that agree on the ground state, a clean demonstration of the correlation between self-consistent U and magnetic moment, and a physical mechanism (cooperative JT ordering) that rationalizes the energy differences. The paper also makes testable predictions, such as the near-degeneracy of ortho and epsilon phases and the strongly insulating character of ortho-LiMnO2. The reproducibility of the methodology is good, with code versions and calculation parameters documented in the text and SI.","major_comments":[{"comment":"The AFM ground state of each phase is selected by enumerating ~30 collinear AFM orderings relaxed with PBEsol+U (U=3.9 eV), and this ordering is then used for all subsequent functionals (HSE06, PBEsol+Usc, PBEsol+(U+V)sc). No convergence test is reported for this enumeration, and for gamma-LiMnO2 and disordered layered there is no experimental magnetic reference. The HSE06 energy of gamma is only ~1 meV/atom above ortho (Figure 3). If the true magnetic ground state of gamma involves a different propagation vector, a larger supercell, or noncollinear ordering not included in the enumeration, its energy could be lower by more than 1 meV/atom, which would overturn the HSE06 claim that ortho is the ground state. Because the abstract explicitly relies on the HSE06 result, this concern is load-bearing. Please demonstrate that the magnetic ordering search is converged (e.g., by testing a larger set of orderings or by re-ranking the low-lying magnetic states with HSE06 or self-consistent U), or qualify the HSE06 prediction as marginal and rest the main claim on the self-consistent Hubbard methods, which give a much larger energy gap (~25 meV/atom).","section":"Section II, Section III A, Figure 3"},{"comment":"The finite-temperature free energy is computed by adding harmonic phonon free energies obtained with PBEsol+U (U=3.9 eV) to PBEsol+(U+V)sc electronic energies. Since the paper argues that PBEsol+U mis-orders the phases, the vibrational free energies of gamma and disordered layered could be biased by the functional choice. The phonon entropy differences among phases are 0.3–0.5 kB/f.u. (Table III), which corresponds to ~3–9 meV/atom over the 300–900 K range. While the direction of the vibrational correction (destabilizing gamma) is consistent with the 0 K electronic result, a quantitative verification of at least one phase's phonons with a more accurate functional (e.g., PBEsol+(U+V)sc or HSE06 on a reduced supercell) would strengthen the statements in the abstract and Discussion about vibrational entropy stabilizing the collinear-JT phases.","section":"Section III A, Figure 4"}],"minor_comments":[{"comment":"The abstract uses 'g-LiMnO2' while the text uses 'gamma-LiMnO2'; please use a single notation throughout.","section":"Abstract and Introduction"},{"comment":"References [128] and [132] appear to be the same paper (Radin and Van der Ven, Chemistry of Materials 30, 607), but they are listed with different years (2018 and 2017). Please consolidate and correct the year.","section":"References"},{"comment":"The statement in the abstract and Section III B that U in gamma and disordered layered is 'by 0.5–0.6 eV' larger than in the observed phases is not uniformly accurate for disordered layered, where U spans 5.92–6.34 eV; some sites differ by only ~0.1 eV. Please rephrase to 'up to 0.5–0.6 eV' or refer to the specific Mn sites with noncollinear JT distortions.","section":"Table I and Section III B"},{"comment":"The caption uses 'E - Eortho' while the text uses Delta E; please align the notation.","section":"Section III A, Figure 2 caption"},{"comment":"For reproducibility, specify the number of AFM orderings actually enumerated for each phase, the size of the magnetic supercells, and the criteria used to define a distinct ordering.","section":"Section II"},{"comment":"The SI text refers to 'SI Table III' but the table is labeled Table S3; please correct the cross-reference.","section":"SI, Section S4"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the scope of the journal and addresses a question of broad interest in first-principles materials science. The central result is plausible and supported by multiple methods, but the magnetic-ordering enumeration is a load-bearing step that needs strengthening before the claims about HSE06 can be considered robust. The authors should either add convergence tests for the magnetic search or adjust the wording of the abstract and conclusions."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Worth a careful read. The paper does a genuinely useful thing: it benchmarks a wide set of DFT functionals on LiMnO2 and shows that empirical U (on GGA and r2SCAN) predicts a gamma-LiMnO2 ground state that is almost certainly wrong, while self-consistent U/U+V and HSE06 recover the experimental orthorhombic phase. The correlation between collinear JT ordering, smaller self-consistent U, and enhanced Mn-O covalency is convincing, and the prediction of the previously unreported epsilon phase is a nice bonus. The G0W0 and HSE06 band gaps (3-4 eV) give the community solid numbers to test against.\n\nThe biggest soft spot is the magnetic ordering search. The authors enumerate ~30 collinear AFM orderings for each phase and relax them with PBEsol+U (U=3.9 eV) — the exact functional they later argue is inadequate. The ortho ordering matches neutron data, but gamma has no experimental reference. The energy scale of AFM order is 5-25 meV/atom, and HSE06 puts gamma only ~1 meV/atom above ortho. So a missed gamma ordering could plausibly flip that specific HSE06 result. This doesn't sink the central claim, because PBEsol+Usc and (U+V)sc give gamma ~25 meV/atom higher, but it does mean the HSE06 leg of the argument is fragile. A referee should ask for a more aggressive magnetic search on gamma (larger cells, noncollinear options) before the HSE06 result is called definitive.\n\nSecond, the finite-temperature free energies mix PBEsol+U phonons with PBEsol+(U+V)sc electronic energies. They justify the phonon functional by prior benchmarks, which is reasonable, but it is still a mixed-level calculation. Also, there is no data or code deposit, which limits reproducibility for a methods-heavy paper.\n\nOverall, the main conclusion holds up. This is a solid contribution that deserves a serious refereeing. I would push the authors to strengthen the magnetic search and release input/output files, but I would not block on these. If I were working on Mn-rich cathodes or DFT+U methodology, I would cite it.","headline":"Clear and useful benchmark showing that empirical Hubbard U gets LiMnO2's ground state wrong and self-consistent U fixes it, though a thin HSE06 margin and a limited magnetic search keep it from being the last word.","tokens_in":35654,"tokens_out":3281,"would_cite":true,"duration_ms":27631,"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":"Standard density-functional theory with an empirical Hubbard U predicts a never-observed gamma-LiMnO2 phase as the ground state; self-consistent Hubbard parameters and hybrid functionals recover the experimentally known orthorhombic…","keywords":["LiMnO2","Hubbard U","Jahn-Teller distortion","antiferromagnetic order","phase stability","DFT+U","hybrid functional","manganese-rich cathodes"],"falsifier":"Enumerate and relax noncollinear and incommensurate antiferromagnetic orderings for the six phases (for example with spin-spiral DFT) and add their energies to the same free-energy comparison; if gamma-LiMnO2 or disordered layered drops below orthorhombic once its true magnetic order is included, the paper's central stability conclusion fails.","tokens_in":34333,"feed_emoji":"🔋","tokens_out":7883,"duration_ms":63529,"temperature":0.7,"pith_summary":"The paper asks which density-functional approximation correctly orders the energies of the LiMnO2 polymorphs, the reference compound for manganese-rich battery cathodes. It shows that standard GGA and meta-GGA functionals with an empirically fitted Hubbard U – a corrective parameter that controls how localized the manganese d electrons are – all predict a never-observed gamma-LiMnO2 phase to be more stable than the orthorhombic phase known from experiment. When the Hubbard U is instead computed self-consistently for each structure, or when a hybrid functional with exact exchange is used, orthorhombic LiMnO2 becomes the ground state and the layered and spinel phases sit close in energy, matching experiment and explaining why they form as impurities. The physical reason is that the self-consistent U is 0.5–0.6 eV smaller in phases whose Jahn-Teller distortions are collinearly aligned, an alignment that increases Mn–O covalency, stabilizes the antiferromagnetic state, and raises vibrational entropy. A sympathetic reader would take away that electron localization is not a fixed property of a transition-metal element but varies with local orbital ordering, and phase stability predictions hinge on capturing that variation.","feed_headline":"DFT with fixed Hubbard U picks the wrong LiMnO2 ground state","feed_subtitle":"Self-consistent Hubbard parameters and hybrids restore orthorhombic LiMnO2 by capturing Jahn-Teller orbital order.","key_machinery":"The load-bearing object is the self-consistent on-site Hubbard parameter U, computed from linear-response theory for each polymorph, together with the ordering pattern of Jahn-Teller axes. A Jahn-Teller distortion is the elongation of two Mn-O bonds in an MnO6 octahedron that lowers symmetry when the Mn3+ eg orbital is singly occupied; 'collinear' means all elongated axes point the same way, 'noncollinear' means they do not. The paper correlates U with the Mn magnetic moment (linear fit, R2 ~ 94%) and the inter-site V with Mn-O bond length, and shows that Mn sites in 180-degree Mn-O-Mn units with one long and one short bond carry the large U values. The machinery works by showing that empirical or averaged U erases these site-to-site differences, producing a spurious gamma ground state, while self-consistent U encodes the local orbital ordering and restores the experimental stability order.","core_discovery":"On the paper's own terms, the central discovery is that the phase stability of LiMnO2 is controlled by the interplay of antiferromagnetic order, Jahn-Teller distortion direction, and the degree of electron localization, and that this interplay is only rendered correctly by methods that let the Hubbard interaction strength respond to the local environment. Empirically fixed U over-stabilizes gamma-LiMnO2, the ordering with the lowest electrostatic energy, because it applies the same localization penalty to every phase; self-consistently computed U is smaller in the phases with collinearly ordered JT axes (orthorhombic, layered, spinel, and the newly identified epsilon phase) and larger in gamma and disordered layered phases that have noncollinear JT arrangements. This variation of about 0.6 eV in U flips the ground state. The paper further shows, through charge-density differences and projected densities of states, that antiferromagnetic order increases Mn-O covalency along the JT axis, that the band gap is large (HSE06 3.1 eV, G0W0 3.8 eV), and that the collinear JT phases have higher phonon entropy, so vibrational free energy reinforces the stability of the experimental phases up to 1000 K.","pith_inferences":["If this mechanism generalizes, other cathode oxides with Jahn-Teller-active ions (Ni3+, Cu2+) may also need self-consistent Hubbard parameters: an empirical U fitted to average properties will not see orbital-ordering-dependent localization.","The paper's finite-temperature free energies use harmonic phonons from an empirical-U calculation; recomputing phonons with self-consistent U, or including anharmonicity, would test whether the vibrational stabilization of collinear JT phases survives beyond the harmonic approximation.","The predicted epsilon-LiMnO2 phase has an XRD pattern close to orthorhombic but with one missing peak near 37 degrees; searching for that fingerprint in existing LiMnO2 samples would test the prediction without new synthesis.","A broader practical implication: Hubbard U should be treated as a local material descriptor, not a transferable constant, so high-throughput databases built with fixed U may need to be reexamined for Mn-rich compounds."],"forward_implications":["Phase-stability calculations for manganese-rich cathodes should not rely on a single empirical U applied to every polymorph; self-consistent U or a hybrid functional is required to find the correct ground state.","Collinear ordering of Jahn-Teller axes carries concrete physical consequences: larger Jahn-Teller bond ratio, increased Mn-O covalency, stronger antiferromagnetic stabilization, and higher phonon entropy.","The layered-to-disorder anti-site defect formation energy is substantial (112 meV/defect at HSE06 and about 400 meV/defect with self-consistent U+V), so cation disorder in layered LiMnO2 is energetically costly.","The previously unreported epsilon-LiMnO2 ordering is predicted to be within about 2 meV/atom of the orthorhombic ground state and more stable than layered or spinel, making it a plausible hidden low-energy phase.","Orthorhombic LiMnO2 is predicted to be a strongly insulating cathode with a band gap above 3 eV, which has direct implications for its electronic conductivity and electrochemical kinetics."],"supporting_citations":[{"why":"Earlier calculation showing that antiferromagnetic order and accurate functionals are needed for LiMnO2 phase stability.","marker":"[11]"},{"why":"Experimental structural study establishing orthorhombic (Pmmn) LiMnO2 as the ground state.","marker":"[12]"},{"why":"Provides the neutron-diffraction-refined antiferromagnetic structure used as the experimental reference for the orthorhombic phase.","marker":"[17]"},{"why":"Gives the empirical U = 3.9 eV that, when applied uniformly, produces the spurious gamma ground state.","marker":"[32]"},{"why":"Supplies the linear-response formalism on which self-consistent Hubbard parameter calculations rest.","marker":"[34]"},{"why":"Supplies the density-functional perturbation theory method used to compute self-consistent U and V.","marker":"[46]"},{"why":"Is the screened hybrid functional HSE06 that recovers the correct ground state.","marker":"[69]"}],"fun_headline_variants":["Fixed Hubbard U predicts wrong LiMnO2 ground state","Self-consistent Hubbard U recovers LiMnO2 phase stability","Cooperative Jahn-Teller order stabilizes LiMnO2 phases","Hybrids and adaptive U fix LiMnO2 ground state","Magnetic order and JT distortion control LiMnO2 stability"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The magnetic ground state of each phase is assumed to be one of the roughly thirty collinear antiferromagnetic spin arrangements that were enumerated and relaxed with a single empirical Hubbard U; if any phase's true magnetic ordering is a different pattern, including one with non-collinear spins, the relative energies could shift enough to change the predicted ground state.","fun_headline_variants_meta":{"raw":{"variants":["Fixed Hubbard U predicts wrong LiMnO2 ground state","Self-consistent Hubbard U recovers LiMnO2 phase stability","Cooperative Jahn-Teller order stabilizes LiMnO2 phases","Hybrids and adaptive U fix LiMnO2 ground state","Magnetic order and JT distortion control LiMnO2 stability"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000524,"raw_usage":{"total_tokens":2647,"prompt_tokens":1177,"completion_tokens":1470,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":793,"completion_tokens_details":{"reasoning_tokens":1386}},"tokens_in":793,"tokens_out":1470,"duration_ms":10791,"temperature":1.0,"reasoning_tokens":1386,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T10:14:56.871950+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Enumerate and relax noncollinear and incommensurate antiferromagnetic orderings for the six phases (for example with spin-spiral DFT) and add their energies to the same free-energy comparison; if gamma-LiMnO2 or disordered layered drops below orthorhombic once its true magnetic order is included, the paper's central stability conclusion fails.","supporting_citations":[{"cited_title":"Hubbard parameters from density-functional perturbation theory","cited_arxiv_id":"1805.01805","evidence_quote":"Supplies the density-functional perturbation theory method used to compute self-consistent U and V."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Is the screened hybrid functional HSE06 that recovers the correct ground state."}],"review_version":1}