{"id":"6af3c1cf-183b-4943-b003-af83d97667db","arxiv_id":"2608.08050","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"ORR on Pt nanoparticles in pH 13 alkaline media leaves OOH, OH, and O2 adsorbates that persist and show hysteresis for over an hour.","lead":"A Raman spectroscopy study of platinum nanoparticle catalysts in alkaline solution found that oxygen reduction reaction intermediates, including OOH and OH, remain on the surface for more than an hour after the reaction stops. The long-lived, history-dependent adsorbates challenge simple microkinetic models and may affect how fuel cell catalysts are understood and designed.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The species-level claim hinges on a three-Gaussian deconvolution of one unresolved band, with peak positions imported from acidic single-crystal studies and no isotope control on the NP surface; if that assignment fails, the long-lived OOHad/OHad/O2,ad narrative collapses.","rationale":"The reader identified the same weak point, and I agree. The paper's controls (Ar vs O2, electrode stability, Au/glassy-carbon checks) are good and support the existence of an O2-induced surface species. However, the novelty and mechanistic story depend on identifying which species are retained. The deconvolution of one unresolved band into three fixed components cannot bear that weight without isotope validation, especially because the 686 cm−1 peak overlaps the PtOx tail and because cation/defect environments at pH 13 may shift modes relative to the acidic single-crystal references. I do not think this warrants rejection: the experimental platform is sound, the O2 dependence is controlled, and the retention times are plausible; a conditional accept with isotope verification as a requirement is appropriate. I therefore leave the reader's verdict unchanged.","tokens_in":1007,"tokens_out":800,"duration_ms":68925,"concrete_test":"Repeat the EC-NERS experiment on the same Au/Pt NPs in O2-saturated 0.1 M KOH prepared in D2O, and separately with 18O2 gas. Confirm that the 686 and 875 cm−1 components shift by the expected ν(O–O) isotopic factor (~√(16/18) ≈ 5.7% for 18O2) and that the 778 cm−1 component shifts for D2O by the expected δ(Pt–O–H) factor, while the PtOx band stays put. If the components shift as predicted, the species assignments are verified; if not, the deconvolution is not identifying the claimed intermediates and the central claim needs revision.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is species-specific: ORR on Au/Pt NPs is said to produce OOHad, OHad, and O2,ad that are long-lived and hysteretic. The evidence for this is the O2-induced 550–950 cm−1 envelope, which is deconvolved into three Gaussians at 686, 778, and 875 cm−1 using peak positions taken from prior single-crystal and acidic SERS studies (refs 39, 41, 77–79). No isotope labeling (D2O or 18O2) is performed in this work, so the transfer of those assignments to pH-13, polycrystalline, defect-rich NP surfaces is untested. The 686 cm−1 component sits directly on the tail of the broad PtOx band (320–700 cm−1), and the deconvolution is not shown to be unique; small baseline or PtOx residuals could masquerade as retained intermediates. The long-lived retention reported in Figures S7a/b and the oxidation-hold test is quantified with this same deconvolved band, without a stated noise floor or fitting residuals. If the assignments are wrong, the species-level conclusions and the active/inactive partitioning in the proposed pathway (eqs 1–5) are unsupported; only the weaker claim of 'some O2-induced adsorbed species' would remain.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports an in situ electrochemical nanoparticle-enhanced Raman spectroscopy (EC-NERS) study of Au/Pt core-shell nanoparticles in 0.1 M KOH and LiOH (pH ~13) during the oxygen reduction reaction (ORR). By comparing Ar- vs O2-saturated electrolytes across multiple potential scans, the authors observe an O2-induced Raman band at ~550–950 cm−1, which they deconvolve into three components assigned to adsorbed OOH, OH, and O2. They report that these adsorbates accumulate during negative-going scans, persist for 81–112 minutes at open circuit and after an oxidation hold at 1.1 V, and show hysteresis between negative- and positive-going scans. The paper proposes an ORR pathway with coexisting kinetically active and inactive intermediates, modulated by electrolyte cations (K+ vs Li+) and by the defect-rich, heterogeneous surface of the Pt nanocatalysts.","tokens_in":21617,"tokens_out":3216,"duration_ms":37079,"significance":"If the species-level assignments and retention results hold, this work is significant because it challenges the common assumption that ORR intermediates exist only transiently and at negligible coverages under high overpotential, and it highlights a gap between single-crystal model studies and realistic nanoparticle catalysts. The paper is also commendable for its systematic controls: O2/Ar comparisons, repeated potential scans, open-circuit retention tests, oxidation-hold tests, and control experiments ruling out Au-core exposure and glassy-carbon degradation. The 3D-FDTD enhancement simulation and electrochemical benchmarking against Pt/C further strengthen the platform characterization. The proposed active/inactive intermediate distinction, if properly supported, would have broad implications for interpreting operando vibrational spectra on practical electrocatalysts.","major_comments":[{"comment":"The central species-level claim rests on a Gaussian deconvolution of an unresolved 550–950 cm−1 envelope into three components at ~686, ~778, and ~875 cm−1, with peak positions imported from prior acidic single-crystal and SERS studies (refs 39, 41, 77–79). No isotope labeling (D2O or 18O2) is performed on the Au/Pt NP surfaces in this work, and the 686 cm−1 component lies directly on the tail of the broad PtOx band (320–700 cm−1). The uniqueness of the three-component fit is not demonstrated: the authors do not report fitting residuals, fixed-versus-floating center tests, or a noise floor for the intermediate band. If the assignments are not transferable to pH-13 polycrystalline NP surfaces, the species-specific conclusions (OOHad, OHad, O2,ad) and the active/inactive partitioning in eqs 1–5 are not supported. Please provide an in-situ isotope control, a rigorous stability/uniqueness analysis of the deconvolution, or explicitly reframe the claims to 'oxygenated adsorbates' without species identification.","section":"Intermediate Deconvolution and Reaction Pathway (Figure 4a–d; Methods 3.7)"},{"comment":"The long-lived retention and hysteretic coverage claims are quantified using uncalibrated Raman band areas of the same deconvolved components. For the retention measurements after 81–112 minutes at open circuit and after the oxidation hold at 1.1 V, the manuscript does not state the detection limit of the intermediate band, the fitting residuals, or the replicate-to-replicate variance. Without a stated noise floor, the finite residual band could partly arise from baseline drift or the PtOx tail rather than from retained ORR intermediates. Please show representative raw spectra with fits at the retention time points and report a quantitative criterion for what constitutes a detectable intermediate band above the noise.","section":"EC-NERS in O2-Saturated Solution (Figures 3e–g, 4e, S7a/b)"},{"comment":"The conclusion that LiOH produces a larger fraction of kinetically active OOHad (i.e., a greater x2 in eq 2) is based on the sharp decay of the OOHad component area below ~0.3 V, while OHad and O2,ad remain roughly constant. Since the Raman band area is a convolution of coverage, scattering cross-section, and local enhancement, and no calibration is provided, the quantitative x_i values and the 'fraction of active species' language are not justified. At minimum, the manuscript should state explicitly that the x_i values are qualitative descriptors of the potential-dependent trends, not measured fractions.","section":"Electrolyte Effect (Figure 5g–i; eqs 1–5)"}],"minor_comments":[{"comment":"The abstract states that the paper 'determined three key intermediates' (OOHad, OHad, O2,ad), but the deconvolution-based assignment is not validated by isotope labeling in this work. Consider tempering the wording to 'assigned to' rather than 'determined' in the abstract and conclusions.","section":"Abstract and Introduction"},{"comment":"The Gaussian component colors in Figure 4 are not consistent with the color scheme used in Figure 4e (orange for OOHad, magenta-purple for OHad, green for O2,ad). Please clarify the correspondence in the caption, as the current text describes the shaded areas without a direct color legend.","section":"Figure 4 and Supporting Information"},{"comment":"Equation (3) includes an adsorbed oxygen atom Oad that is not directly observed in the Raman spectra. The text acknowledges that the exact molecular pathway cannot be identified, but the inclusion of Oad in the central mechanistic scheme could be more explicitly flagged as an assumed intermediate rather than a spectroscopically detected species.","section":"Reaction pathway (eqs 1–5)"},{"comment":"The paper repeatedly states that statistics were derived from technical replicates, but no error bars or replicate numbers are visible in Figures 3–5. Please indicate how many replicate measurements were averaged and whether the error bars are smaller than the plotted symbols.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within scope for a physical chemistry/electrochemistry journal and the central observation of persistent, hysteretic O2-induced adsorbates on Pt nanoparticles is timely. The main risk is that the species-specific narrative (OOHad/OHad/O2,ad) is more strongly worded than the spectroscopic evidence supports, given the absence of isotope controls and the unvalidated deconvolution on a PtOx tail. If the authors can supply either an isotope control or an explicit deconvolution robustness analysis, or alternatively soften the species-level claims, the paper would be a solid contribution. I would not recommend rejection, as the core empirical observation of long-lived O2-induced species appears reproducible from the O2/Ar comparisons and retention tests; the required changes are substantial but local."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this paper reports a real and useful observation—O2-saturated 0.1 M KOH/LiOH leaves adsorbates on Au/Pt nanoparticles that linger for over an hour and show scan-direction hysteresis. The controls are decent: Ar vs O2 comparisons, repeated scans, open-circuit retention, Au-core and carbon-support checks. If you work on alkaline ORR or operando spectroscopy of NP catalysts, this changes the picture of what 'surface state' means after the reaction stops.\n\nWhat's new: prior Raman work on Pt in acid saw OOH/OH on single crystals and NPs, and weakly alkaline studies on smooth electrodes saw superoxide. This is the first pH-13 NP dataset, and the multi-hour persistence is a genuinely new empirical claim. The paper is careful to separate the PtOx band from the O2-induced envelope, and the potential-dependent trends are shown across multiple scans and two cations.\n\nThe soft spots are concentrated in the species-level assignment. The 550-950 cm-1 envelope is deconvolved into three Gaussians at 686, 778, 875 cm-1, assigned to OOHad, OHad, O2,ad using peak positions from acidic/single-crystal studies and one D2O-shift reference from a different system. No isotope experiment is done here, and the 686 cm-1 component sits on the PtOx tail. The fit is not shown to be unique; no residuals or error bars are given. That means the quantitative coverage trends for each intermediate are less certain than the main band-area trends. The active/inactive partitioning in the reaction scheme is plausible but not directly measured.\n\nNone of this kills the central claim. The O2-induced band is real, it persists, and it is hysteretic. What is shaky is the detailed inventory of which molecular species are involved and how much of each. The authors acknowledge some of this—they say the O2,ad signal is weak and the precise adsorption configuration of OOH is unresolved—but they do not flag the transfer-of-assignment risk on their own surfaces.\n\nWho is this for: electrochemists and spectroscopists studying ORR or adsorbate dynamics on practical catalysts. It deserves a serious referee: the experiment is hard, the controls are thoughtful, and the main observation is a step change from the single-crystal picture. Recommended for peer review with a request for isotope labeling or at least a clear statement of the deconvolution's non-uniqueness, raw spectra, and fit parameters.","headline":"A solid in situ Raman study showing persistent, hysteretic O2-induced adsorbates on Pt nanocatalysts at pH 13; the species-specific assignment rests on imported peak positions without in-house isotope verification.","tokens_in":22217,"tokens_out":1994,"would_cite":true,"duration_ms":21672,"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":"Oxygen reduction on platinum nanoparticles leaves OOH, OH, and O2 on the surface for over an hour after the reaction stops.","keywords":["oxygen reduction reaction","platinum nanocatalyst","in situ Raman spectroscopy","adsorbed OOH","adsorbed OH","adsorbed O2","electrochemical hysteresis","alkaline electrolyte"],"falsifier":"A D2O or 18O2 isotope-labeling experiment on the same Au/Pt nanoparticles in 0.1 M KOH or LiOH would settle the central claim: the 686, 778, and 875 cm−1 components must shift by the expected vibrational isotope factors if they are OOHad, OHad, and O2,ad. A complementary check is a time-resolved desorption measurement after the reaction is stopped: if the intermediate band decays with a short lifetime under conditions where the paper claims hour-long persistence, the hysteresis and trapping conclusions would need revision.","tokens_in":21151,"feed_emoji":"🧪","tokens_out":11594,"duration_ms":106407,"temperature":0.7,"pith_summary":"The paper sets out to show that the oxygen reduction reaction on realistic platinum nanoparticle catalysts leaves behind surface adsorbates that are far longer-lived than standard models assume. Using in situ electrochemical nanoparticle-enhanced Raman spectroscopy on Au-core/Pt-shell nanoparticles in strongly alkaline solution (pH about 13), the authors identify three oxygenated intermediates — adsorbed OOH, OH, and O2 — that appear only when oxygen is present, accumulate during negative-going potential scans, and persist for 81 to 112 minutes after the reaction has been turned off. The retention is hysteretic: when the potential is scanned back positive, the adsorbate coverage does not return to its pre-reaction level. The authors attribute the persistence to a population of kinetically inactive intermediates trapped at defect sites on the multi-faced nanoparticle surface, and show that the trapped fraction depends modestly on cations (K+ versus Li+). If right, this means practical catalysts carry a surface memory of prior operation that single-crystal studies and many microkinetic models omit.","feed_headline":"ORR leftovers cling to Pt nanoparticles for over an hour","feed_subtitle":"In situ Raman shows reaction intermediates trapped on Pt surfaces for over an hour — a memory single-crystal models miss","key_machinery":"The load-bearing platform is in situ electrochemical nanoparticle-enhanced Raman spectroscopy (EC-NERS) using Au-core/Pt-shell nanoparticles: the Pt shell acts as the oxygen reduction catalyst while the Au core provides plasmonic enhancement, with a calculated Raman enhancement factor up to $4.7 \\times 10^{7}$ at interparticle junctions. The argument turns on a difference measurement — a band at roughly 550–950 cm−1 appears only in O2-saturated electrolyte, not in Ar-saturated electrolyte, so it is assigned to oxygen reduction products. Gaussian deconvolution of that unresolved band into components at roughly 686, 778, and 875 cm−1 is the step that converts one broad feature into species-resolved coverage of OOHad, OHad, and O2,ad, using peak positions from prior single-crystal and acidic surface-enhanced Raman studies. A companion band at 320–700 cm−1, treated as amorphous PtOx, provides a simultaneous measure of oxide coverage and reveals a competitive accumulation effect between oxide and intermediates. The potential- and history-dependent areas of these bands are the data that support the long-lived, hysteretic, trap-related conclusions.","core_discovery":"The central discovery is that oxygen reduction on Pt nanocatalysts in alkaline media produces oxygenated adsorbates — OOHad, OHad, and O2,ad — that are highly long-lived and hysteretic, persisting even after the reaction has ended. In O2-saturated 0.1 M KOH or LiOH, a Raman band at roughly 550–950 cm−1 appears only in the presence of oxygen and grows as the potential is lowered; Gaussian deconvolution splits it into components near 686, 778, and 875 cm−1, assigned to the O–O stretch of adsorbed OOH, the Pt–O–H bend of adsorbed OH, and the O–O stretch of bridge-adsorbed O2. The intermediate band remains detectable after 81 minutes at open circuit in KOH and after 112 minutes in LiOH, and it survives a deliberate oxidation hold at 1.1 V. The paper interprets this persistence as the coexistence of kinetically active and kinetically inactive intermediates, proposes an associative oxygen reduction pathway in which a fraction of each intermediate branches into trapped states, and reports that Li+ yields a larger active fraction below 0.3 V while K+ gives flatter, more retained coverage, attributed to cation blocking of defect sites.","pith_inferences":["A natural next experiment the paper does not report is isotope labeling: repeating the measurements in D2O or with 18O2 would test the 686/778/875 cm−1 assignments on the authors' own nanoparticle surfaces, and would either confirm or redirect the whole interpretation.","If trapped adsorbates really are the oxygen reduction intermediates, then pulsed-potential or alternating-current operation should selectively drain the active fraction and expose the desorption kinetics of the trapped fraction, giving a direct measure of the trapping energy that the current work only infers.","The cation-blocking explanation predicts a systematic trend across the alkali series — smaller cations should block defect sites more effectively and lower the trapped fraction — which could be tested with Cs+, Na+, and Rb+ without changing any other condition.","A practical consequence the paper leaves implicit is that fuel-cell shutdown may leave a chemically active adsorbate layer on the cathode; on restart, that layer could alter the first-cycle activity or participate in degradation, making the hysteresis relevant beyond spectroscopy."],"forward_implications":["Oxygen reduction intermediates on realistic Pt/C-like catalysts can be detected long after the reaction stops, so delayed or ex-situ vibrational measurements may still report reaction products rather than the resting surface.","Microkinetic models that predict near-zero intermediate coverage at high overpotential are missing a trapped, inactive adsorbate population on defect-rich nanoparticles, which may explain why measured spectra and kinetic models diverge.","Cation identity affects not only oxygen reduction activity but also surface speciation: K+ and Li+ produce different fractions of active versus trapped intermediates, so electrolyte choice changes the surface state that subsequent reactions see.","The first potential scan on an oxide-covered surface differs from later scans: intermediates appear only after PtOx is reduced, whereas previously accumulated intermediates make later scans accumulate intermediates at higher potentials.","Vibrational band area should not be directly equated with kinetic current, because a substantial part of the spectroscopic signal may come from kinetically inactive adsorbates."],"supporting_citations":[{"why":"This single-crystal Raman study supplies the peak assignments for OOHad and OHad that anchor the 686 and 778 cm−1 components.","marker":"39"},{"why":"This bimetallic-nanocatalyst study provides the D2O-isotope-confirmed assignments for OOHad and O2,ad that this work transfers to alkaline Au/Pt surfaces.","marker":"77"},{"why":"This DFT study computes the O–O stretch of bridge-adsorbed O2 on Pt(111), the basis for assigning the 875 cm−1 band to O2,ad rather than end-on O2.","marker":"78"},{"why":"This work assigns the Pt–O–H bending mode of adsorbed OH, the identification used for the 778 cm−1 component.","marker":"79"},{"why":"This in-situ IR study observed adsorbed intermediates on Pt nanoparticles, a result this Raman work complements and extends to strong alkaline pH.","marker":"26"},{"why":"This weakly alkaline single-crystal study identified superoxide as the oxygen reduction intermediate, providing the contrast against which this paper's different species list is defined.","marker":"9"},{"why":"This mechanistic study supplies the associative alkaline oxygen reduction pathway on Pt(111) that the proposed active/inactive branching scheme builds on.","marker":"1"}],"fun_headline_variants":["ORR intermediates cling to Pt nanoparticles for over an hour","Pt nanoparticles retain oxygen intermediates long after ORR ends","Persistent oxygenated adsorbates on Pt nanocatalysts defy expectations","Raman shows ORR intermediates stick to Pt surfaces for over an hour"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the three fitted Raman peaks at 686, 778, and 875 cm−1 really are adsorbed OOH, OH, and O2 on these platinum nanoparticles; the authors did not verify the assignment on their own surfaces with isotope labeling, so if those peak identifications are wrong, the claim about which species are long-lived collapses.","fun_headline_variants_meta":{"raw":{"variants":["ORR intermediates cling to Pt nanoparticles for over an hour","Pt nanoparticles retain oxygen intermediates long after ORR ends","Persistent oxygenated adsorbates on Pt nanocatalysts defy expectations","Raman shows ORR intermediates stick to Pt surfaces for over an hour"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000792,"raw_usage":{"total_tokens":3524,"prompt_tokens":1018,"completion_tokens":2506,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":634,"completion_tokens_details":{"reasoning_tokens":2435}},"tokens_in":634,"tokens_out":2506,"duration_ms":18603,"temperature":1.0,"reasoning_tokens":2435,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T00:29:59.355823+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A D2O or 18O2 isotope-labeling experiment on the same Au/Pt nanoparticles in 0.1 M KOH or LiOH would settle the central claim: the 686, 778, and 875 cm−1 components must shift by the expected vibrational isotope factors if they are OOHad, OHad, and O2,ad. A complementary check is a time-resolved desorption measurement after the reaction is stopped: if the intermediate band decays with a short lifetime under conditions where the paper claims hour-long persistence, the hysteresis and trapping conclusions would need revision.","supporting_citations":[],"review_version":1}