{"id":"a9f46ccb-b4ca-4bd4-ba01-caa2fe18a21c","arxiv_id":"2608.10910","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"A site-resolved database of simulated O K-edge XAS spectra for NMC811 at three states of charge, validated against experiment and released as a community resource.","lead":"The paper presents a database of 1,127 simulated oxygen K-edge X-ray absorption spectra for the battery cathode NMC811 and six benchmark transition-metal oxides, with each spectrum resolved to an individual oxygen site. If reliable, the database gives researchers a free, citable reference for interpreting X-ray measurements and for training machine learning models in battery materials science.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"R2SCAN's documented spurious LiCoO2 pre-edge doublet (Sec. 4.1) may corrupt site-resolved fingerprints for Co-coordinated O environments in NMC811; the paper never checks whether this artifact appears in those sites.","rationale":"The paper is a substantial data descriptor: it provides an automated XCH/R2SCAN workflow, benchmarks on six binary oxides, validates NMC811 total spectra at three states of charge, and releases code. The central scientific claim is that the site-resolved spectra constitute a reliable reference for fingerprinting and ML training. The reader's weakest assumption identified systematic spectral deviations and the use of only three static configurations per state of charge; I agree that environment-dependent errors would be damaging. The most concrete and load-bearing manifestation is the spurious LiCoO2 pre-edge doublet that the paper itself reports as robust under R2SCAN. Since NMC811 contains Co, and some O sites have Co neighbors, this known qualitative failure could directly bias a subset of the site-resolved fingerprints. The total-spectrum validation against experiment cannot rule this out because Co is a minority component and the artifact may be diluted in the average. The proposed test—grouping database site-resolved spectra by Co presence and inspecting the pre-edge lineshape—would settle whether the concern lands. If it does, the database remains valuable but the site-resolved fingerprint mapping for Co environments would require explicit caveats or reanalysis with a different functional. Hence CONDITIONAL, consistent with the reader's verdict, with the specific concern sharpened.","tokens_in":10702,"tokens_out":3204,"duration_ms":31383,"concrete_test":"Group the site-resolved spectra in the database by first-shell composition, comparing O sites whose six nearest cations include at least one Co against sites with only Ni/Mn/Li neighbors within the same NMC811 structure. Compute the average pre-edge (528-532 eV) lineshape for each group and check whether the Co-containing group exhibits the doublet feature seen in the R2SCAN LiCoO2 calculation (Fig. 2). If the doublet appears in Co-containing O sites but not in Co-free sites, the site-resolved fingerprint mapping for Co environments is affected by the documented R2SCAN artifact and the central claim needs qualification.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim is that the site-resolved O K-edge spectra reliably encode local environments for fingerprinting and ML training. The paper itself documents (Sec. 4.1) that under the adopted R2SCAN protocol LiCoO2 exhibits a spurious pre-edge doublet absent in experiment and in LDA/PBE calculations, and that this splitting is robust to magnetic initialization. NMC811 contains 6 Co atoms per 60 transition-metal sites, so many O sites have Co in their first cation shell; Sec. 4.5 explicitly lists Co-containing environment classes. If the doublet arises from R2SCAN's description of Co-O hybridization, then the site-resolved spectra for Co-coordinated O sites in the NMC811 database likely carry the same artifact. The paper never checks whether Co-coordinated O sites show a doublet. The only validation (Sec. 4.3) is on the configuration-averaged total spectrum, where the 10% Co fraction and the modest counts per environment class (roughly 10-30 O atoms per class in Fig. 6) can dilute the artifact into a subtle shoulder. Because the database's distinguishing asset is site-resolved fingerprints, an environment-dependent spectral error in Co-containing environments directly undermines the mapping from local structure to spectrum, and hence the database's value for ML training and fingerprinting. The other acknowledged deviations (pre-edge intensity, 1-2 eV main-edge shift) are less problematic if roughly uniform, but the LiCoO2 doublet is a qualitative, material-specific failure that demands a site-resolved check.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript reports a database of simulated oxygen K-edge X-ray absorption spectra for the layered cathode material NMC811, computed with the excited-electron/core-hole (XCH) method and the R2SCAN meta-GGA functional as implemented in VASP. The database contains spectra for six benchmark binary oxides and for nine NMC811 supercells at three states of charge, with 1,127 total oxygen-site-resolved spectra. The authors validate the computational protocol against their own measurements on the binary oxides and against published experimental spectra for NMC811, and they document the diversity of oxygen coordination environments in the supercells. They argue that the level of agreement supports the reliability of the protocol, making the site-resolved spectra suitable for spectral fingerprinting, direct experiment comparison, and machine-learning training. The data, workflow, and post-processing scripts are made openly available.","tokens_in":102,"tokens_out":4444,"duration_ms":110353,"significance":"If the central claim holds, the database fills a clear gap: there is no first-principles, site-resolved O K-edge XAS database for a realistic disordered Ni-rich layered oxide across states of charge. The automated XCH workflow, the inclusion of configurational spread at each delithiation level, and the benchmark against independently measured binary-oxide spectra are concrete strengths that make the resource potentially useful to the battery and spectroscopy communities. The site-resolved nature of the data is the most distinctive asset, as it allows structure-to-spectrum mapping. However, the significance is tempered by known systematic errors that the paper itself documents: the R2SCAN LiCoO2 pre-edge doublet, underestimated pre-edge intensity, and 1-2 eV main-edge offsets. Because these errors may be environment-dependent, they directly affect the fingerprinting and machine-learning use case that motivates the database.","major_comments":[{"comment":"The R2SCAN calculation for LiCoO2 produces a spurious pre-edge doublet that is absent in experiment and in LDA/PBE, and the paper states that this splitting is robust to magnetic initialization. NMC811 contains Co in many oxygen first-shell environments, as shown in the environment classification of Fig. 6, so Co-coordinated O sites in the NMC811 database may carry the same artifact. The paper never checks this. Because the database's distinctive value is its site-resolved spectra, an environment-dependent spectral error in Co-containing environments would directly corrupt the structure-to-spectrum mapping that the abstract and Sec. 4.5 claim to support. I ask for a specific test: group the NMC811 site-resolved spectra by whether Co appears in the first cation shell and look for doublet features in those groups, or recompute a few representative Co-coordinated O sites with LDA/PBE as a cross-check. Without such a check, the central claim of site-resolved reliability is not fully established for Co-containing environments.","section":"Sec. 4.1, Fig. 2; Sec. 4.5, Fig. 6"},{"comment":"The caption of Fig. 4 states that the calculated and experimental NMC811 spectra are 'aligned on the dominant pre-edge peak.' The text then says that 'the pre-edge peak position is well captured by the calculation.' Aligning on the feature that is being validated makes this claim circular. The paper should either report the absolute excitation onsets obtained from the total-energy-difference procedure described in Sec. 2.1, or provide an unaligned comparison, so that readers can judge the accuracy of the energy scale independently. If the absolute alignment is poor, that should be stated explicitly, since the database's energy axis is part of its usability for fingerprinting and machine learning.","section":"Sec. 4.3, Fig. 4"},{"comment":"The NMC811 database relies on only three low-energy static configurations per state of charge. The authors acknowledge in Sec. 4.3 that thermal disorder is not captured, and Fig. 6 shows that the environment populations vary substantially among the three configurations at fixed delithiation. For a database intended to train machine-learning models that map local structure to spectral response, it is important to know whether three configurations are sufficient to represent the local environment diversity of disordered NMC811. I ask the authors to quantify this, for example by reporting the number of distinct environment classes as a function of the number of configurations, the convergence of the configuration-averaged spectrum, or an explicit statement that the database samples only low-energy configurations and therefore does not cover the full thermal disorder distribution.","section":"Sec. 2.2 and Sec. 4.3"}],"minor_comments":[{"comment":"The Materials Cloud DOI is given as a placeholder 'xx-xx' and the GitHub release is listed as 'vJ.H.'; the manuscript should contain the permanent DOI and version once available, since these are the access points for the database.","section":"Sec. 3"},{"comment":"The benchmark validation is described only qualitatively. Reporting a simple quantitative metric, such as the energy separation between the pre-edge and main-edge maxima for each compound, would make the strength of the benchmark easier to evaluate.","section":"Sec. 4.2, Fig. 3"},{"comment":"The Lorentzian broadening slope a=0.1 is a fitting parameter. Because the post-processing choice affects the displayed line shapes, the database should clearly state whether raw unbroadened spectra are included or only the post-processed spectra with this broadening, so that users are not misled about which data are parameter-free.","section":"Sec. 2.1"},{"comment":"The sentence 'R2SCAN was nevertheless adopted for the database because of its improved description of the ground-state electronic structure' is a reasonable choice, but the reader needs a comment on whether the LiCoO2 doublet is expected to transfer to Co-substituted NMC811 or whether disorder or dilution suppresses it; this links to the major comment on Co-coordinated sites.","section":"Sec. 4.1"},{"comment":"The selection of the three lowest-energy configurations from the earlier DFT dataset is not described in detail; a sentence explaining how many configurations were considered and how the energy ranking was made would improve reproducibility.","section":"Sec. 2.2"}],"recommendation":"major_revision","confidential_remarks":"The paper is a solid data-descriptor submission with a useful open database and a transparent account of its limitations. The main technical risk is the unexamined transfer of the R2SCAN LiCoO2 artifact to Co-coordinated oxygen sites in NMC811; that is a load-bearing point for the site-resolved fingerprinting claim and should be addressed with a specific test before publication. The alignment issue in Fig. 4 is also important because it affects how readers interpret the validation. I would recommend major revision rather than rejection, because the database itself is valuable and the concerns are addressable with additional analysis."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague, this one is worth your time if you work on XAS interpretation or ML for cathode materials. The authors present the first first-principles, site-resolved O K-edge XAS database for NMC811 across three states of charge — 1,127 spectra from an automated XCH/R2SCAN workflow, plus benchmarks on six binary Ti and Mn oxides. The benchmarks reproduce the main spectral trends and are compared against their own measured NEXAFS data, which is more than most database papers do. The NMC811 spectra track the experimental evolution of the pre-edge with delithiation. That is a genuinely new and useful artifact.\n\nThe soft spots are real but fixable. Most immediately, the Materials Cloud DOI is a placeholder, '10.24435/materialscloud:xx-xx'. That blocks any external use and makes the central claim of the paper unverifiable. Second, the paper honestly reports that R2SCAN yields a spurious LiCoO2 pre-edge doublet, robust to magnetic initialization. The stress-test concern that this artifact may appear in Co-coordinated O sites in NMC811 is legitimate. The paper does not check site-resolved spectra for Co-containing environments, even though such environments are present in the database and the site-resolution is the database's differentiating asset. The configurational average can dilute the artifact to a shoulder, but fingerprinting and ML training on site-resolved spectra would be directly affected. This needs a targeted analysis before the database is used as reference. Third, the Fig. 4 comparison aligns calculated and experimental spectra on the pre-edge peak, which partially masks edge-onset errors; the fitted Lorentzian slope a=0.1 is a minor post-processing parameter.\n\nOverall, the approach is sound, the validation is honest, and the limitations are stated. The resource is likely to be cited and used. I would send it to peer review, but with a request for the Co-site doublet check and a live DOI. The paper deserves a serious referee, and with those fixes it would be a solid database contribution.","headline":"A genuinely new site-resolved O K-edge XAS database for NMC811 with honest validation, but the unaddressed R2SCAN Co-site artifact and the placeholder DOI need fixing before this is fully usable.","tokens_in":11558,"tokens_out":2159,"would_cite":true,"duration_ms":19977,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A database of 1,127 site-resolved oxygen K-edge spectra lets researchers read NMC811's local structure from its X-ray absorption.","keywords":["oxygen K-edge X-ray absorption spectroscopy","NMC811","layered cathode","core-hole XCH method","R2SCAN","site-resolved spectra","spectral fingerprinting","machine learning training data"],"falsifier":"Take a measured NMC811 O K-edge spectrum, fit it as a weighted sum of the database's site-resolved spectra, and inspect the residual; if the residual is structured or if the fitted weights disagree with the known state of charge, the site-to-spectrum fingerprint mapping is not capturing the real material.","tokens_in":10491,"feed_emoji":"🔋","tokens_out":8938,"duration_ms":75866,"temperature":0.7,"pith_summary":"The paper sets out to make the oxygen K-edge X-ray absorption spectrum of NMC811, a nickel-rich cathode for lithium-ion batteries, interpretable at the level of individual oxygen atoms. It computes 1,127 spectra from first principles—47 for benchmark oxides and 1,080 for nine NMC811 supercells at three states of charge—and validates the protocol against measured reference spectra. The central claim is that the simulated spectra reproduce the key experimental features and their evolution with lithium removal, so the database can serve as a reference for spectrum fingerprinting and as training data for machine-learning models. A sympathetic reader would care because interpreting O K-edge spectra of disordered cathodes has so far relied on matching to simple reference compounds, which fails when no suitable reference exists.","feed_headline":"1,127 simulated spectra map NMC811's oxygen environments","feed_subtitle":"Site-resolved O K-edge spectra at three states of charge give a fingerprint reference and ML training data.","key_machinery":"The load-bearing object is the excited-electron-and-core-hole (XCH) calculation: a core electron from an O 1s level is promoted into unoccupied states while the remaining electrons relax self-consistently in the presence of the core hole, and the absorption spectrum is read from the dielectric response. The workflow identifies symmetry-inequivalent oxygen sites automatically, runs one XCH calculation per site, aligns the site spectra by their total-energy excitation onsets, and averages them with site multiplicities. This is what converts a static atomic structure into a library of environment-specific spectral fingerprints, and it is why the data can be used to map spectral features back to local coordination.","core_discovery":"On the paper's own terms, the discovery is a validated computational protocol and an open database that ties every symmetry-distinct oxygen site in NMC811 to its calculated O K-edge signature. The spectra are computed with the excited-electron-and-core-hole method using the R2SCAN meta-GGA functional, with a full core hole, and each spectrum is aligned by its computed excitation onset and combined into a multiplicity-weighted site average. For six binary titanium and manganese oxides the computed spectra reproduce the observed oxidation-state trends, and for NMC811 at three states of charge the calculated spectra match the measured pre-edge position, the broad main edge, and the growth of pre-edge intensity upon delithiation. The authors therefore assert that the protocol captures the essential physics of the O K-edge in NMC811 and that the resulting site-resolved spectra are dependable enough for spectral fingerprinting and machine-learning training.","pith_inferences":["A natural next step the paper does not demonstrate is inverting the database: fitting a measured spectrum as a linear combination of site-resolved spectra to estimate the population of Ni oxidation states and local environments.","If the fingerprint assumption holds, the database could be extended to other cathode compositions by interpolation, because similar oxygen environments recur in related layered oxides.","One testable extension is to train a model on the NMC811 site spectra and predict the spectrum of a configuration not in the training set, then compare with a fresh XCH calculation; this would expose environment-dependent errors that the static sampling might hide."],"forward_implications":["Experimental O K-edge spectra of NMC811 at any state of charge can be decomposed into contributions from specific oxygen environments, since the database records each site's spectrum separately.","Machine-learning models that map local atomic structure to X-ray absorption response can be trained on the 1,080 site-resolved NMC811 spectra, which cover the compositional disorder of the real cathode.","The benchmark results imply the same XCH/R2SCAN protocol can be applied to other layered oxide cathodes to generate reference spectra where no experimental reference exists.","The calculated spectra confirm that nickel is the primary redox-active center in NMC811 and that delithiation increases the intensity of the pre-edge by creating more unoccupied O 2p–metal 3d states."],"supporting_citations":[{"why":"Supplies the nine NMC811 supercell structures and cation configurations on which all NMC811 spectra are computed.","marker":"[1]"},{"why":"Provides the standard interpretation of O K-edge pre-edge features as O 2p–transition metal 3d hybridization used throughout the analysis.","marker":"[7]"},{"why":"Supplies the excited-electron-and-core-hole method that every spectrum in the database is computed with.","marker":"[9]"},{"why":"Supplies the total-energy-difference method used to fix each absorption edge's absolute energy and to align site-resolved spectra before averaging.","marker":"[17]"},{"why":"Provides the experimental layered-oxide spectra against which LDA, PBE, and R2SCAN are benchmarked in the functional test.","marker":"[21]"},{"why":"Supplies the R2SCAN meta-GGA functional used for all production calculations.","marker":"[23]"},{"why":"Provides the comparison showing R2SCAN describes transition-metal oxide electronic structure more reliably, the basis for the functional choice.","marker":"[24]"},{"why":"Provides the experimental NMC811 spectra at three states of charge used to validate the calculated spectra.","marker":"[30]"},{"why":"Provides the precedent of a simulated X-ray absorption database built from XCH calculations, which this work extends to O K-edge NMC811.","marker":"[19]"}],"fun_headline_variants":["O K-edge spectra database links NMC811 oxygen sites to signals","NMC811 oxygen edges decoded: 1,127 simulated spectra","Open database matches O K-edge to every NMC811 oxygen site","Site-resolved O K-edge spectra for NMC811: ML-ready training set"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole database rests on the assumption that three static, low-energy atomic configurations per state of charge—plus the single-electron core-hole approximation—capture the range of oxygen environments in the real disordered material closely enough that a computed site spectrum can stand in for the measured response.","fun_headline_variants_meta":{"raw":{"variants":["O K-edge spectra database links NMC811 oxygen sites to signals","NMC811 oxygen edges decoded: 1,127 simulated spectra","Open database matches O K-edge to every NMC811 oxygen site","Site-resolved O K-edge spectra for NMC811: ML-ready training set"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000767,"raw_usage":{"total_tokens":3393,"prompt_tokens":934,"completion_tokens":2459,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":550,"completion_tokens_details":{"reasoning_tokens":2381}},"tokens_in":550,"tokens_out":2459,"duration_ms":16443,"temperature":1.0,"reasoning_tokens":2381,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T14:28:12.126395+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a measured NMC811 O K-edge spectrum, fit it as a weighted sum of the database's site-resolved spectra, and inspect the residual; if the residual is structured or if the fitted weights disagree with the known state of charge, the site-to-spectrum fingerprint mapping is not capturing the real material.","supporting_citations":[{"cited_title":"Direct Simulation of LiNi0.8Mn0.1Co0.1O2 Transport Properties Using an Efficient and Accurate Machine Learning Potential","cited_arxiv_id":"2605.19747","evidence_quote":"Supplies the nine NMC811 supercell structures and cation configurations on which all NMC811 spectra are computed."},{"cited_title":"& Mauri, F","cited_arxiv_id":null,"evidence_quote":"Supplies the excited-electron-and-core-hole method that every spectrum in the database is computed with."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the total-energy-difference method used to fix each absorption edge's absolute energy and to align site-resolved spectra before averaging."},{"cited_title":"URLhttps://chemrxiv.org/engage/chemrxiv/ article-details/667f17abc9c6a5c07ab375aa","cited_arxiv_id":null,"evidence_quote":"Provides the experimental layered-oxide spectra against which LDA, PBE, and R2SCAN are benchmarked in the functional test."},{"cited_title":"URLhttps://doi.org/10.1149/ 1945-7111/ac3c21","cited_arxiv_id":null,"evidence_quote":"Provides the experimental NMC811 spectra at three states of charge used to validate the calculated spectra."},{"cited_title":"URLhttps://www.nature.com/articles/ s41597-023-02262-4","cited_arxiv_id":null,"evidence_quote":"Provides the precedent of a simulated X-ray absorption database built from XCH calculations, which this work extends to O K-edge NMC811."}],"review_version":1}