{"id":"cc9f023f-7da2-431c-a1ca-08b6e6b66440","arxiv_id":"2607.28000","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"PASS samples perturbed space-group cells under 10 atoms to train a transferable ACE Fe–O potential that reproduces bulk/surface/interface tests and forms FeO-like oxide in large-scale MD.","lead":"The authors introduce PASS, a one-shot way to build small-cell training sets for machine-learned Fe–O potentials, and show an ACE potential that forms FeO-like oxide in large oxidation MD. It matters because iron oxidation underpins corrosion, energy storage, and materials durability, and better reactive potentials are still scarce.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.5","headline":"Fixed-FM small-cell training understates oxide thermodynamics, so the MD “FeO-like” claim may not be a true phase identification.","rationale":"The reader correctly isolates the load-bearing premise: fixed-FM, short-cutoff, small-cell LAEs are assumed sufficient for magnetically complex Fe oxides and high-T growth. The manuscript supplies extensive OOD structural/mechanical tests and a visually coherent oxidation run, so the contribution is real as a sampling method plus a practical potential; it is not internally empty. The soft spot is quantitative and application-facing: Table 3 and Fig. 4 already document oxide thermodynamic and vacancy-energy errors of the size that can select the wrong local stoichiometry/structure in reactive MD, and Fig. 5’s FeO-like reading is only short-range-order evidence under that biased PES. That does not require REJECT—methods novelty and pure-Fe/surface transferability still stand—but it keeps the verdict CONDITIONAL and argues for toning “first-of-its-kind transferable … reactive complexity of oxide growth” until DFT(+U) re-ranking of the MD product (or spin-aware/U-consistent labeling) closes the gap. Agreement with the reader is full on the weakest assumption; the stress test only sharpens the concrete falsifier around the oxidation showcase rather than magnetism in the abstract.","tokens_in":19150,"tokens_out":852,"duration_ms":18809,"concrete_test":"Re-relax the extracted region-I oxide slab (and bulk FeO, Fe2O3, Fe3O4 references) with the same DFT(+U) protocol used for Table 3 / Fig. 4, and recompute ACE vs DFT(+U) formation energies and relative phase stability at the observed composition/defect level; if ACE still favors FeO-like order by ≳0.2–0.3 eV/atom while DFT(+U) does not (or ranks another oxide lower), the MD phase assignment and the transferability claim for oxide growth do not hold.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim requires that PASS small-cell LAEs (max 10 atoms, 6 Å ACE cutoff), labeled only under fixed ferromagnetic DFT without U or spin degrees of freedom, suffice for transferable oxide thermodynamics and kinetics so that a large-scale oxidation MD can be read as forming FeO-like oxide. The paper’s own numbers undercut that premise for the oxides that matter: Table 3 formation enthalpies are systematically too exothermic relative to DFT(+U) (FeO −1.57 vs −0.91/−1.43; Fe2O3 −2.00 vs −1.30/−1.81; Fe3O4 −1.93 vs −1.29/−1.75 eV/atom), while bulk moduli and some lattice constants also deviate; Fig. 4(d) Fe vacancy formation energies in FeO/Fe3O4 are underestimated even though trends are preserved; Bain-path (Fig. 3c) only qualitatively tracks the AFM-favored FCC-like region. Discussion explicitly limits the potential where magnetism and spin disorder matter. The oxidation showcase (Fig. 5) then identifies “FeO-like” structure from one 873 K trajectory via ~50% composition, broadened RDF peak positions, and first-shell CN—without energy ranking against Fe2O3/Fe3O4, without magnetic order, and without a control that the biased ΔHf does not preferentially stabilize rock-salt FeO-like short-range order. If those thermodynamic biases drive the observed layer, PASS has not been shown to capture reactive oxide-growth complexity beyond a structurally plausible but energetically unvalidated morphology.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.5","summary":"The manuscript introduces PASS, a one-shot workflow that generates compact Fe and Fe–O training sets from space-group prototypes (≤10 atoms), stepwise relaxation, controlled rattle/stretch perturbations, SOAP+PCA embedding, and farthest-point sampling, then trains an ACE MLIP under a fixed ferromagnetic DFT labeling protocol. The potential is validated on pure-Fe EOS, thermal expansion, Bain path, cleavage/T–S and GSF energetics, point defects and surfaces, and on Fe–O bulk oxides, surfaces, Fe/FeO interfaces, O adsorption/dissolution, O diffusivity versus experiment, and vacancy formation/migration barriers, before a 1 ns BCC Fe(100) oxidation MD at 873 K that produces a near-equiatomic surface layer whose RDF and first-shell coordination are interpreted as FeO-like. The central claim is that small-cell LAE sampling alone yields a transferable Fe–O MLIP capable of reactive oxide-growth complexity without iterative active learning or explicit large oxidation structures in training.","tokens_in":19439,"tokens_out":1155,"duration_ms":21003,"significance":"If the transferability claim holds, PASS would be a practical, symmetry-informed alternative to manual enumeration, RSS, and multi-cycle active learning for chemically complex reactive systems, and a first demonstration that small-cell space-group sampling can support large-scale Fe oxidation MD. Strengths include the systematic out-of-distribution validation suite (defects, GSF, adsorption, CINEB, experimental O diffusion, interfaces) and the explicit comparison of SOAP embeddings against application-specific Fe and Fe–O datasets. The work is of clear interest for MLIP dataset design and high-temperature oxidation modeling, provided thermodynamic biases and the magnetic limitation are quantified rather than only acknowledged.","major_comments":[{"comment":"Table 3 and Fig. 4(d): formation enthalpies of FeO, Fe2O3 and Fe3O4 are systematically more exothermic than DFT(+U) references (e.g. FeO −1.57 vs −0.91/−1.43 eV/atom), and Fe vacancy formation energies in FeO/Fe3O4 are underestimated even though trends are preserved. These are load-bearing for the claim that the potential captures oxide thermodynamics. The manuscript should quantify how these biases affect relative oxide stability (with and without U, and versus O chemical potential) and state clearly which thermodynamic conclusions remain reliable.","section":"Table 3; Fig. 4(d)"},{"comment":"Fig. 5 oxidation showcase: the “FeO-like” identification rests on ~50% composition, broadened RDF peak positions, and first-shell CN from a single 873 K trajectory, without energy ranking against Fe2O3/Fe3O4-like order, without magnetic order, and without a control that the biased ΔHf does not preferentially stabilize rock-salt short-range order. Given the Discussion’s own limits on magnetism and the Table 3 offsets, either add comparative structural/thermodynamic diagnostics or soften the claim from phase-like identification to short-range Fe–O order under the simulated conditions.","section":"Fig. 5; Application to large-scale oxidation simulation"},{"comment":"Methods and Discussion: all training labels use fixed ferromagnetic ordering without U or explicit spin degrees of freedom, while target oxides are AFM/ferrimagnetic and high-T oxidation involves spin disorder. Bain-path (Fig. 3c) and oxide tests only partially probe this gap. A concrete statement of the domain of applicability—and, if feasible, a limited spin-ordered or +U comparison on key oxide/interface energies—is needed so readers can judge when the structural-transferability argument is sufficient.","section":"Methods (DFT calculations); Discussion; Fig. 3(c)"}],"minor_comments":[{"comment":"Training errors (31.89 meV/atom, 120.52 meV/Å) are given without a held-out test split or learning-curve context; a brief test-set or bootstrap estimate would help interpret absolute accuracy.","section":"Systematic validation of the ACE MLIP for pure Fe"},{"comment":"Supplementary Table 1 (perturbation grid) is cited as central to PASS reproducibility but is not in the main text; ensure the grid and FPS size are fully specified for reuse.","section":"The PASS workflow"},{"comment":"Fig. 2(c,d) PCA overlays are illustrative; state explicitly that coverage is necessary but not sufficient evidence of transferability (the property tests carry that burden).","section":"Fig. 2"},{"comment":"Minor typographical issues appear in the reference list and elsewhere (e.g. oxida3on-style character substitutions, “V olmin”, “plaroorm”); a full proofread is needed.","section":"References"},{"comment":"ACE cutoff (6 Å) and max cell size (<10 atoms) imply truncated long-range electrostatics in ionic oxides; a short note on implications for charged defects and polar surfaces would help.","section":"Discussion; Methods"}],"recommendation":"major_revision","confidential_remarks":"The methodological contribution (PASS) is real and the validation breadth is above average for an Fe–O MLIP paper. The skeptic concern on fixed-FM thermodynamics and the FeO-like MD reading is substantive and should be fixed in revision; it does not look like an unfixable flaw if the authors quantify biases and tone the phase claim. Fit to a materials/MLIP journal is appropriate."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"The thing worth knowing is that PASS is a practical one-shot recipe—space-group small cells, a mesh of rattle/stretch, SOAP–PCA–FPS—and they actually ship a Fe–O ACE with a broad out-of-distribution checklist, not just another training-set story. That combination is what makes the paper usable.\n\nWhat is new is not ACE or space-group sampling; it is the controlled perturbation layer plus descriptor selection applied to a chemically nasty binary, then pushed all the way to a nanosecond oxidation run. They do the hard work: pure-Fe EOS, Bain path, cleavage/T–S, GSF, vacancies/SIAs/surfaces; oxide lattices and ΔHf trends; O adsorption/dissolution; diffusivity vs experiment; CINEB vacancy barriers; Fe/FeO interfaces; and an MD case with composition, RDF, and coordination. Many of those classes were not hand-planted in the set. That is real transfer evidence, not circular PCA theater.\n\nSoft spots are real but bounded. Training is fixed-FM, no U, 6 Å cutoff, cells ≤10 atoms. Table 3 formation enthalpies are too exothermic vs DFT(+U); vacancy formation energies are low; Bain only qualitatively tracks the AFM FCC-like window. The Discussion owns the magnetism limit. So the Fig. 5 “FeO-like” layer—~50% composition, broadened RDF peaks, CN—is a short-range-order observation from one 873 K trajectory, not a demonstrated free-energy ranking against Fe2O3/Fe3O4. If the ΔHf bias prefers rock-salt local order, the showcase overclaims “reactive complexity.” Tone those sentences down and release the potential/data.\n\nCitations look fair (ASSYST, ACE, ReaxFF failures, Bienvenu). Methods are detailed enough to reproduce the pipeline even while artifacts are still closed.\n\nThis is for people building reactive MLIPs and oxidation/corrosion MD, not for spin-resolved oxide thermodynamics. I would send it to referees. Engage if you care about dataset design or Fe–O MD; treat the phase ID as provisional until magnetism and energies are tighter.","headline":"Useful ASSYST extension plus a real Fe–O ACE validation suite; the oxidation “FeO-like” read is structural, not a clean phase proof under fixed-FM thermodynamics.","tokens_in":20224,"tokens_out":555,"would_cite":true,"duration_ms":21504,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.5","headline":"Small-cell space-group sampling plus controlled perturbations trains a transferable Fe–O potential that grows FeO-like oxide in large simulations.","keywords":["Iron oxidation","Space groups","Machine learning interatomic potentials","Atomic cluster expansion","Molecular dynamics","PASS sampling","Fe-O system"],"falsifier":"Run the same large-scale BCC Fe oxidation MD with a spin-aware or DFT+U-referenced potential (or compare directly to in-situ structural probes) and check whether the surface layer still reaches near-equal Fe/O composition with FeO-like RDF peak positions and coordination numbers; systematic mismatch would falsify the claimed transferability.","tokens_in":19863,"feed_emoji":"⚛️","tokens_out":956,"duration_ms":17122,"temperature":0.7,"pith_summary":"Iron oxidation is hard to model atomistically because the Fe–O system mixes many crystal structures, stoichiometries, defects, and magnetic orderings, so building a trustworthy training set for a machine-learning potential has been slow and incomplete. This paper introduces PASS: generate compact crystals from space groups (cells under 10 atoms), relax them, then systematically rattle atoms and stretch cells to create non-equilibrium local environments, and keep only the most representative ones via SOAP descriptors and farthest-point sampling. An atomic-cluster-expansion potential trained on that one-shot set is then shown to match bulk, surface, defect, adsorption, diffusion, and interface energetics for pure iron and common oxides—even though those large or defective configurations were never put in the training data. In a nanosecond-scale molecular-dynamics run of oxygen on BCC iron at 873 K, the same potential forms a surface layer whose composition and short-range order are FeO-like. The practical claim is that carefully sampled small cells can transfer to chemically complex oxide growth without iterative active learning or hand-built oxidation pathways.","feed_headline":"Small cells train an Fe–O potential that grows FeO-like oxide","feed_subtitle":"Space-group sampling plus perturbations transfers from under-10-atom cells to large oxidation MD","key_machinery":"PASS (Perturbation Augmented Space group structure Sampling): space-group prototypes are volume- and fully relaxed, then extensively rattled and stretched on a grid of strain parameters; SOAP descriptors plus PCA and farthest-point sampling retain a diverse, non-redundant subset of local atomic environments for one-shot DFT labelling and ACE training.","core_discovery":"PASS produces a compact, representative Fe–O training set of small-cell structures such that an ACE machine-learning interatomic potential trained on it is accurate and transferable enough to reproduce bulk, surface, and interface thermodynamics and kinetics and to capture the reactive complexity of large-scale Fe oxidation, including formation of an FeO-like oxide layer.","pith_inferences":["Because magnetism is fixed and long-range interactions are cut off, regimes where AFM/ferrimagnetic order or spin disorder dominate (Curie/Néel crossings, certain vacancy barriers) remain the natural next stress tests.","PASS-style one-shot sampling could be paired later with targeted active learning only for rare defective interfaces rather than for the whole configurational space.","If the FeO-like layer result holds under broader oxygen pressures and temperatures, the potential becomes a practical tool for corrosion and iron-powder energy-storage modelling."],"forward_implications":["A transferable Fe–O ACE potential can be built from small cells alone, cutting the cost of high-fidelity DFT labelling relative to large hand-crafted or actively learned sets.","Large-scale MD of early oxide growth becomes practical and can spontaneously form FeO-like short-range order under high-temperature conditions.","The same sampling logic can be reused for other chemically complex reactive systems where manual pathway enumeration or many active-learning cycles are currently required.","Defect, adsorption, diffusion, and interface properties of Fe and Fe oxides can be screened without putting those configurations into the original training set."],"fun_headline_variants":["PASS small cells train transferable Fe–O ACE potential","Under-10-atom PASS set builds Fe–O MLIP for oxidation MD","Perturbed space-group sampling yields FeO-like growth potential","Small-cell Fe–O dataset transfers to bulk surface and oxide MD","ACE MLIP from PASS captures Fe oxidation to FeO-like layers"],"cache_read_input_tokens":16512,"weakest_assumption_plain":"Local environments taken from heavily perturbed cells of at most ten atoms, labelled under one fixed ferromagnetic DFT setting and without explicit magnetism or long-range electrostatics, are assumed to be enough for the energetics and kinetics of real iron oxides, interfaces, and high-temperature oxidation.","fun_headline_variants_meta":{"raw":{"variants":["PASS small cells train transferable Fe–O ACE potential","Under-10-atom PASS set builds Fe–O MLIP for oxidation MD","Perturbed space-group sampling yields FeO-like growth potential","Small-cell Fe–O dataset transfers to bulk surface and oxide MD","ACE MLIP from PASS captures Fe oxidation to FeO-like layers"]},"model":"grok-4.5","effort":"low","cost_usd":0.004819,"raw_usage":{"total_tokens":1375,"prompt_tokens":749,"num_sources_used":0,"completion_tokens":76,"cost_in_usd_ticks":48188000,"prompt_tokens_details":{"text_tokens":749,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":550,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":749,"tokens_out":76,"duration_ms":9344,"temperature":1.0,"reasoning_tokens":550,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-31T20:50:31.735621+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"Run the same large-scale BCC Fe oxidation MD with a spin-aware or DFT+U-referenced potential (or compare directly to in-situ structural probes) and check whether the surface layer still reaches near-equal Fe/O composition with FeO-like RDF peak positions and coordination numbers; systematic mismatch would falsify the claimed transferability.","supporting_citations":[],"review_version":1}