{"id":"8bdfcf03-bf65-4dca-8cbc-4c5aa8ad51bb","arxiv_id":"2501.03165","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"A computational and experimental survey of the Na8-xAxP2O9 framework reports a new phase, Na4SnP2O9, and identifies tunable chemistry, while machine-learned potential predictions of fast room-temperature conduction are not confirmed by experiment.","lead":"This paper surveys a little-studied sodium ion conductor family, NAP, and reports a new member, Na4SnP2O9, made by solid-state synthesis. It predicts several substituted variants could conduct sodium quickly at room temperature, but those predictions clash with the measured conductivity of the materials that were tested.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"MLMD conductivity predictions for unsynthesized NAP candidates are unsupported: the same fine-tuned CHGNet pipeline underestimates measured NTP/NSP activation barriers by ~0.5 eV and misses the conductivity transition.","rationale":"The reader's weakest_assumption correctly identifies the reliability of fine-tuned CHGNet MLMD diffusivities as load-bearing. My independent read of Sections II.B.2, III.A, III.B, III.E and Figures 3 and 8 reaches the same conclusion: the computational pipeline that produces the headline high predicted conductivities for unsynthesized candidates underpredicts the measured activation energies of the two experimentally characterized NAP phases by roughly 0.5 eV and does not reproduce the coupled structural-conductivity transition. Because all candidate prioritization in Figure 4b is drawn from this pipeline, the central abstract claim about high predicted ionic conductivities is not currently supported by an independent calculation or by experiment. The paper is transparent about this limitation, explicitly acknowledging the discrepancy, the incomplete NSP structure solution, and the possibility that the structures investigated or the AIMD mechanism are incomplete. This transparency is a strength, but it does not supply the missing validation. A focused DFT NEB study on LT-NTP is the cleanest test because LT-NTP's structure is well established, unlike NSP, so any discrepancy can be attributed to the potential rather than to an unresolved crystal structure. If the NEB barrier reproduces ~0.8 eV, the MLMD-based room-temperature conductivity numbers should be labeled as qualitative rankings at best, and the verdict remains conditional on that revalidation rather than being accepted as a quantitative prediction.","tokens_in":44275,"tokens_out":3688,"duration_ms":40054,"concrete_test":"Perform DFT NEB calculations for a Na-vacancy hop in the experimentally determined monoclinic LT-NTP structure (ICSD 39901) along the layer-ii pathway, using the Section II.B.1 DFT settings. If the NEB barrier is approximately 0.8 eV, matching the measured EIS activation energy of 0.836 eV, rather than ~0.3 eV from MLMD, then the CHGNet-based Figure 4b room-temperature conductivities for all unsynthesized NAP candidates are unsupported and require revalidation before being reported as predictive.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The abstract's claim that screening identifies candidates with 'high predicted ionic conductivities' rests on the fine-tuned CHGNet MLMD pipeline described in Section II.B.2. That pipeline is contradicted by the paper's own experiments. For NTP, MLMD gives 0.294 eV (LT) and 0.280 eV (HT), versus EIS values of 0.836 eV and 0.449 eV, and it predicts no conductivity transition (Figure 3). For NSP, MLMD gives 0.27 eV versus measured values of 0.916 and 0.867 eV in the two low-temperature regimes (Figure 8). The extrapolated room-temperature conductivities of 0.01-12.4 mS/cm for unsynthesized candidates in Figure 4b therefore inherit an unvalidated barrier model, and the reported fine-tuning force MAEs of 46-75 meV/Å (Table S24) do not guarantee accurate Na migration barriers. The paper honestly documents this discrepancy, but the central forward-looking claim still depends on it.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript reports a combined computational and experimental survey of the Na8−xAxP2O9 (NAP) family of sodium ion conductors. The authors analyze the parent Na4TiP2O9, attribute its low-temperature distortion to unstable phonons, perform high-throughput DFT and machine-learned potential screening of 24 A-site substitutions, and attempt synthesis of the promising candidates in an automated laboratory. They report successful synthesis of one new member, Na4SnP2O9, along with structural, microscopic, and impedance characterization, and propose a conduction pathway model for the NAP framework.","tokens_in":44553,"tokens_out":4350,"duration_ms":37789,"significance":"If the screening results were reliable, the NAP framework would be a valuable new tunable platform for sodium solid electrolytes, and the phonon-based explanation of the conductivity transition would be a useful design rule. The experimental work also provides a rare detailed account of the difficulties of solid-state synthesis of sodium phosphates, including a quantitative comparison of mixing methods, which is a useful contribution to synthesis science. The paper is exemplary in documenting its own limitations: the authors explicitly state that the NSP structure solution is not definitive, that the MLMD barriers disagree with EIS, and that more characterization is needed. The machine-checkable aspects, however, are limited: the DFT hull and phonon calculations are standard and reproducible, but the MLMD conductivity predictions are not validated against experiment.","major_comments":[{"comment":"The MLMD pipeline used for all predicted room-temperature conductivities is quantitatively inconsistent with the EIS measurements reported in the same paper. For NTP, the computed activation energies are 0.294 eV (LT) and 0.280 eV (HT) versus measured 0.836 eV and 0.449 eV, and the computation shows no conductivity transition (Fig. 3). For NSP, the computed 0.27 eV barrier contrasts with measured 0.916 eV and 0.867 eV in the two low-temperature regimes (Fig. 8). Since the same pipeline yields the 0.01–12.4 mS/cm predictions for unsynthesized candidates in Fig. 4b, the central claim of 'high predicted ionic conductivities' in the abstract is not supported by the evidence presented. The authors acknowledge the discrepancy but do not provide a resolution within the manuscript; this needs to be addressed by either validating the MLMD approach on related measured frameworks or substantially tempering the forward-looking claims.","section":"II.B.2, III.A, III.E (Figs. 3, 8)"},{"comment":"The structural assignment of the new phase Na4SnP2O9 is explicitly acknowledged as 'not definitive.' The best refinement uses the orthorhombic Bmem prototype but has Rwp = 13.22% with unmatched peaks near 15°, 31.5°, and beyond 50° 2θ, and intensity mismatches on major peaks. Since the paper's central experimental claim is the discovery of this new NAP phase, and subsequent conductivity measurements and computational comparisons depend on this structure, a more rigorous structure solution (e.g., quantitative comparison of alternative ordered/distorted models, or additional diffraction data) is required before the synthesis claim can be taken as established.","section":"III.E, Fig. 7a"},{"comment":"The abstract and Section III.B describe the screened candidates as 'likely synthesizable' based on Ehull < 30 meV/atom and negative computed reaction energies. However, the automated synthesis trials in Section III.C succeeded for only one of the approximately ten candidates satisfying these criteria (Table S21); most reactions yielded only precursor phases or competing oxides. The authors offer plausible kinetic explanations, but the phrase 'likely synthesizable' is contradicted by the paper's own experimental outcomes. The screening claim should be reframed as 'thermodynamically promising but kinetically challenging' or supported by a quantitative metric that accounts for the observed failures.","section":"III.B, III.C"}],"minor_comments":[{"comment":"Typo: 'pseduo-Jahn Teller' should be 'pseudo-Jahn Teller'.","section":"Abstract"},{"comment":"References to 'Figure 8a', 'Figure 8b', and 'Figure 8c' should refer to Figure 9 (the pathway schematic); Figure 8 is the NSP conductivity plot.","section":"III.F"},{"comment":"In the Supporting Information, reference [25] (Dara) and reference [28] (AutoSEMEDS) are listed as 'tbd tbd' and should be completed before publication.","section":"SI references"},{"comment":"The phrase 'Reaction network [50,51] was performed with the NSP phase set as the target' is missing an article; also there is inconsistent use of 'Reaction Network' versus 'reaction network'.","section":"II.B.3"},{"comment":"There are several spelling errors: 'Brillion' zone (Section III.A), 'Suppplemental' (Section III.D.4), and 'Reitveld' (SI Section 7.1).","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"This manuscript is from a well-known group and contains an unusually large and honest experimental dataset. The main risk is that the forward-looking screening claims are not supported by the internal validation. I recommend major revision, with the option for the authors to reframe the paper as a characterization and synthesis study rather than a validated screening study, which would be a more accurate and still valuable contribution. The NSP structure solution and the MLMD validation are the two points that need the most work."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Plainly: this is an honest, useful survey of an understudied sodium-ion framework, and it contains one genuinely new phase, but the headline computational claim—that cation substitution yields several fast room-temperature sodium conductors—is not supported by the paper's own measurements. The same fine-tuned CHGNet pipeline that predicts 0.27–0.29 eV barriers for NTP and NSP gives measured barriers of 0.45–0.92 eV and misses the conductivity transition entirely. So the room-temperature conductivities in Figure 4b, and the \"promise\" of Na3VP2O9 and Na3TaP2O9, rest on an unvalidated model.\n\nWhat is actually new and good: the pseudo-Jahn-Teller phonon explanation for the NTP structural/conductivity transition is plausible and backed by DFT phonons; the 24-cation stability and diffusivity screen is a reasonable first pass; and the synthesis of Na4SnP2O9, with the detailed optimization of precursor, dwell temperature, calcination, and mixing, is a real experimental contribution. The paper also deserves credit for reporting negative synthesis results, which are rare, and for plainly acknowledging the computational/experimental conductivity discrepancy rather than burying it.\n\nThe soft spots are exactly where the reader's report puts them. First, the MLMD predictions: fine-tuning force MAEs of 46–75 meV/Å (Table S24) do not establish accuracy for migration barriers, and the EIS comparison is a direct falsification for the two phases where data exist. The paper's suggestion that the discrepancy might come from inaccurate structural models or grain boundaries is reasonable but unproven; either way, the predicted room-temperature conductivities for unsynthesized candidates are unsupported. Second, the NSP structure solution is not definitive—Rwp = 13.22% with unmatched peaks, and the authors themselves call for further refinement. Since the computed NSP conductivity depends on that structure, it inherits the uncertainty. Neither of these flaws is hidden; the paper is transparent about both, which is why I would not call it a takedown.\n\nThe DFT hull and phonon work stands on its own. The synthesis study stands on its own. The MLMD-based conductivity ranking does not. A serious referee should send this to major revision: either revalidate the MLMD pipeline against a measured phase, or sharply soften the forward-looking conductivity claims and present Figure 4b as a provisional, unvalidated screen. The NSP structure should be flagged as tentative until better diffraction data or a higher-symmetry solution emerges.\n\nThis paper is worth a reading group discussion and warrants a serious referee. I would accept it for peer review, but with the expectation of heavy revision.","headline":"Honest and useful survey of a neglected sodium-ion framework with a new phase and a plausible phonon mechanism, but the MLMD conductivity predictions over-promise relative to the paper's own EIS data.","tokens_in":45125,"tokens_out":2496,"would_cite":true,"duration_ms":22975,"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 NAP sodium phosphate framework is a tunable solid-electrolyte family whose high-temperature conduction is gated by a phonon-driven structural transition, and whose substituted members screen as fast sodium conductors.","keywords":["sodium solid-state electrolyte","NAP framework","phonon-driven conductivity transition","cation substitution screening","machine-learned potential molecular dynamics","Na4SnP2O9","solid-state synthesis"],"falsifier":"Measure EIS on a phase-pure Na4SnP2O9 pellet with a fully solved low-temperature structure across 25–400 °C and compare with MLMD trajectories on that same structure; if the measured activation energy remains near 0.9 eV rather than the predicted 0.27 eV and no high-conductivity transition appears in the simulation, the MLMD-based room-temperature conductivities for the unsynthesized NAP candidates are unsupported.","tokens_in":44113,"feed_emoji":"🔋","tokens_out":8450,"duration_ms":78019,"temperature":0.7,"pith_summary":"This paper argues that the NAP family of sodium phosphates, formula $\\mathrm{Na}_{8-x}A^{x}\\mathrm{P}_2\\mathrm{O}_9$, is a tunable platform for solid-state sodium electrolytes that has been overlooked because its known members stop conducting when cooled. The parent phase Na4TiP2O9 is shown to be dynamically unstable in its high-temperature form, with phonon modes that distort titanium octahedra and sodium cages and raise the energy barrier for sodium motion below roughly 300 °C. Using convex-hull stability, reaction-energy filters, and machine-learned-potential molecular dynamics, the authors identify fifteen cation-substituted NAP phases that are thermodynamically plausible, several of which are predicted to be fast sodium conductors; the 5+ candidates Na3VP2O9 and Na3TaP2O9 are extrapolated to room-temperature conductivities of about 11–12 mS/cm. One new member, Na4SnP2O9, was synthesized near phase purity, but it conducts only at high temperature, and its measured activation energy (0.916 eV) is far above the predicted 0.27 eV—the gap exposing the machine-learned-potential diffusivity step as the fragile link. The wider point is that NAP deserves a place in sodium solid-state electrolyte design despite, or because of, its temperature-driven conductivity transition.","feed_headline":"New sodium superionic family: 15 stable phases, one made","feed_subtitle":"An overlooked phosphate framework screens well for solid-state batteries, and a phonon instability explains why it only conducts when hot.","key_machinery":"The central object is the NAP framework itself: an orthorhombic (Bmem) sodium phosphate built from one-dimensional chains of corner-sharing AO6 octahedra, phosphate tetrahedra, and two kinds of sodium polyhedra (square-pyramidal and 8-coordinate cubic cages) that share faces and form percolation channels in the a-c plane. The argument is carried by two mechanisms. First, phonon calculations on the parent Na4TiP2O9 show unstable modes—pseudo-Jahn-Teller elongation/compression of TiO6 octahedra, octahedral rotation coupled to Na-Na dimerization, and collective oxygen wobbling—which together transform the fast-conducting orthorhombic phase into the distorted monoclinic phase with a much steeper sodium site-energy landscape. Second, a screening funnel combines convex-hull energy, precursor reaction energy, and machine-learned-potential molecular dynamics diffusivities to rank $\\mathrm{Na}_{8-x}A^{x}\\mathrm{P}_2\\mathrm{O}_9$ candidates, with sodium vacancies (created by 4+ and 5+ substitution) as the mobile-carrier variable and a radius-dependent activation energy attributed to a pillaring effect.","core_discovery":"The paper's central claim is that the NAP framework—$\\mathrm{Na}_{8-x}A^{x}\\mathrm{P}_2\\mathrm{O}_9$, built from one-dimensional AO6 octahedral chains and large sodium cages—is a chemically flexible sodium superionic platform whose poor reputation comes from a removable structural problem. The high-temperature orthorhombic parent Na4TiP2O9 is dynamically unstable: phonon calculations find pseudo-Jahn-Teller TiO6 elongation/compression modes, octahedral rotations, and oxygen wobbling that, on cooling, distort the sodium polyhedra and steepen the energy landscape for alternative sodium occupations, producing the measured order-of-magnitude conductivity drop near 300 °C. Substituting 3+, 4+, and 5+ cations changes sodium content and vacancy patterns; fifteen of twenty-five candidate phases are computed to lie within 30 meV/atom of their convex hull, and almost all have negative reaction energies from common sodium carbonate, ammonium phosphate, and oxide precursors. Molecular-dynamics conductivity screening then singles out 5+ substitutions (Na3VP2O9 and Na3TaP2O9) as room-temperature conductors near 10 mS/cm, with 4+ substitution showing a radius-dependent activation-energy trend attributed to a pillaring effect. Experimentally, one new phase, Na4SnP2O9, was synthesized and optimized to near phase purity, but its measured ambient conductivity is about $10^{-7}$ S/cm with a 0.916 eV activation energy—far higher than the predicted 0.27 eV—which the authors attribute to an incomplete structural model, grain-boundary resistance, or the tin-rich particle surface rather than to failure of the framework concept.","pith_inferences":["If the phonon-driven transition is the gate, then partial substitution or sodium-vacancy engineering that suppresses the soft modes could stabilize the fast orthorhombic phase at room temperature; the paper lists stabilization of high-symmetry polymorphs as future work but does not demonstrate it.","The systematic MLMD/EIS gap, with predicted barriers roughly one-third of measured values, suggests either that the literature structures used for simulation are not the real transport geometries or that interface and grain-boundary resistance dominate the pellets; a single-particle or single-crystal conductivity measurement would separate the two and is a direct next test.","The calcination effect implies that the molecular unit of the phosphate precursor—isolated PO4 versus condensed polyphosphate—is a synthesis variable for the whole phosphate family, so precursor ranking algorithms should incorporate pathway intermediates rather than only final reaction energies."],"forward_implications":["The NAP framework is compositionally flexible: fifteen of the twenty-five tested A-site substitutions sit within 30 meV/atom of the convex hull, so the sodium phosphate family is not restricted to the titanium parent.","Five-plus cation substitutions with lower sodium content, particularly Na3VP2O9 and Na3TaP2O9, are predicted to be fast room-temperature conductors on the order of 10 mS/cm with three-dimensional diffusion at high temperature, making them the most promising targets for further synthesis.","The conductivity transition in NTP is caused by unstable phonon modes that distort the sodium site energy landscape, so suppressing those modes is the route to room-temperature conduction in NAP materials.","A new NAP phase, Na4SnP2O9, is synthesizable by conventional solid-state routes, but only the poorly conducting low-temperature form was obtained; ball milling with nanoparticle SnO2 and a roughly 950 °C dwell gives near phase purity.","Thermodynamic reaction energy alone is a poor predictor of solid-state synthesis success: calcination temperature and precursor mixing change target yield by tens of percent through the reaction pathway, so synthesis planning must include intermediate phases."],"supporting_citations":[{"why":"Supplies the low- and high-temperature Na4TiP2O9 structure models and the measured conductivity transition used to anchor the phonon and molecular-dynamics analysis.","marker":"[14]"},{"why":"Describes the pretrained machine-learned potential that is fine-tuned per NAP structure and used for all diffusivity and conductivity predictions.","marker":"[46]"},{"why":"Provides the open database of density-functional-theory energies used to build convex hulls and assess thermodynamic stability of substituted NAP candidates.","marker":"[55]"},{"why":"Describes the automated solid-state synthesis system used for high-throughput experimental screening of the candidate phases.","marker":"[23]"},{"why":"Supplies the reaction-network model used to propose precursor sets and reaction energies for forming NAP targets.","marker":"[50]"},{"why":"Gives the synthesis-recipe recommendation model used to generate precursor sets for NSP and other targets.","marker":"[52]"},{"why":"Provides the synthesis planning algorithm whose predicted route was tested and found low-yield, motivating the precursor-pathway analysis.","marker":"[53]"},{"why":"Reports partial sodium disorder on Na2, Na3, and Na4 sites in NTP, supporting the diffusion pathway picture.","marker":"[21]"}],"fun_headline_variants":["Sodium superionic family: 15 stable phases, one synthesized","Overlooked NAP framework yields 15 stable sodium phases","Lattice instability explains why this sodium conductor needs heat","High-throughput search finds new sodium conductor candidates"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The screening's promise rests on fine-tuned machine-learned-potential molecular dynamics giving quantitatively reliable sodium diffusion barriers, yet the measured activation energies for both synthesized NAP phases are about three times the predicted ones.","fun_headline_variants_meta":{"raw":{"variants":["Sodium superionic family: 15 stable phases, one synthesized","Overlooked NAP framework yields 15 stable sodium phases","Lattice instability explains why this sodium conductor needs heat","High-throughput search finds new sodium conductor candidates"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000395,"raw_usage":{"total_tokens":2227,"prompt_tokens":1253,"completion_tokens":974,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":869,"completion_tokens_details":{"reasoning_tokens":908}},"tokens_in":869,"tokens_out":974,"duration_ms":9878,"temperature":1.0,"reasoning_tokens":908,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T21:52:56.892948+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure EIS on a phase-pure Na4SnP2O9 pellet with a fully solved low-temperature structure across 25–400 °C and compare with MLMD trajectories on that same structure; if the measured activation energy remains near 0.9 eV rather than the predicted 0.27 eV and no high-conductivity transition appears in the simulation, the MLMD-based room-temperature conductivities for the unsynthesized NAP candidates are unsupported.","supporting_citations":[{"cited_title":"Duan \\ and\\ author R","cited_arxiv_id":null,"evidence_quote":"Reports partial sodium disorder on Na2, Na3, and Na4 sites in NTP, supporting the diffusion pathway picture."}],"review_version":1}