{"id":"84ec891b-1722-42f3-aea4-6b6cc50badf5","arxiv_id":"2608.07782","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"Metal-intercalated biphenylene bilayers (B/M/B) are predicted to be competitive ORR/OER catalysts, with Cu and Pt best for ORR and Fe best for OER, but the proposed electronic descriptors rely on fits with several systems excluded.","lead":"This paper uses density functional theory to screen 13 transition metals intercalated between biphenylene bilayers as catalysts for oxygen reduction and evolution, finding several with competitive overpotentials. It then proposes simple electron-counting descriptors, such as the metal d-orbital charge, as guides for future catalyst design.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Reported overpotentials and descriptor trends assume the intercalated metal stays at its clean square-site position while O*/OH*/OOH* bind at C468; the paper shows this fails for Nb, W, Os, and Ti, but never checks it for the headline catalysts Cu, Pt, Ru, Mn, and Fe.","rationale":"The reader's conditional verdict is justified, and the same premise is the weakest link. I considered making the descriptor overfitting the primary concern: the volcano fits in Fig. 8 remove the best ORR system (Cu) and other outliers to reach R2 = 0.99, and the d-orbital charge is nearly monotonic with atomic number, so its predictive value is limited. That is a real weakness, but it weakens the interpretive 'descriptor' layer rather than the computed overpotentials themselves. The metal-displacement/active-site premise is more load-bearing because it underlies every overpotential in Table III and every descriptor value in Table IV. The paper's own evidence that adsorption moves several intercalated metals means the assumption is not merely an untested idealization; it is demonstrated to fail in a subset of the same systems. Since a positive result on the proposed test would restore confidence in the numbers while a negative result would change both the activity ranking and the descriptor analysis, the appropriate disposition remains conditional rather than rejection.","tokens_in":30755,"tokens_out":7162,"duration_ms":66257,"concrete_test":"Compute the full ORR and OER free-energy diagrams for B/Cu/B and B/Fe/B (and, if resources allow, B/Pt/B, B/Ru/B, and B/Mn/B) under two conditions: (i) the intercalated metal constrained at its clean square-site position while O2*, O*, OH*, and OOH* relax on C468; (ii) the metal fully relaxed, starting from both the clean-site geometry and an off-site displaced geometry, with each intermediate also tried on C688. Track the metal displacement and relative energies. If the metal moves more than ~0.2 Å or any step Gibbs energy shifts by more than 0.1 eV, recompute U_onset, eta, and the d-orbital descriptor values from the relaxed structures; large shifts would falsify the reported activity ordering and descriptor volcano.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative results, including the headline overpotentials in Table III and the descriptor correlations in Fig. 8, are computed under the implicit premise that the catalyst geometry is the clean B/M/B cell with each ORR/OER intermediate adsorbed at the C468 site while the intercalated metal stays at its equilibrium square-site position. This premise is stated in Sec. III.B and justified only by O2 binding and PDOS at C468 (SM Fig. S.4); it is not tested for the intermediates O*, OH*, and OOH* on the proposed best catalysts. The paper itself documents failures of this premise: adsorption of O2 or of O*/OH*/OOH* displaces Nb, W, Os, and Ti from the square site, producing large energy changes (footnote 82; SM Figs. S.2 and S.11). No equivalent check is reported for Cu, Pt, Ru, Mn, or Fe, the systems that carry the central claim. If any of those metals relaxes off-site under an intermediate, the corresponding free energies, onset potentials, and clean-surface d-orbital populations no longer describe the operating active site.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports a DFT study (PBE+U, vdW-DF2, VASPsol, CHE model) of the ORR and OER on metal-intercalated biphenylene bilayers B/M/B for M = Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Nb, Ru, W, Os, Pt. The authors compute adsorption free energies of OOH*, O*, and OH* at the C468 carbon site and derive thermodynamic overpotentials, finding B/Cu/B, B/Pt/B, B/Ru/B, and B/Mn/B to be the best ORR systems (η = 0.42–0.56 V) and B/Fe/B the best OER system (η = 0.44 V). They then examine d-band centers, p_z-band centers, and orbital charge populations as activity descriptors, proposing volcano-type correlations between overpotentials and metal d-orbital or carbon p-orbital charge populations. Stability is assessed via intercalation energies and 400 K AIMD simulations.","tokens_in":30995,"tokens_out":4212,"duration_ms":37025,"significance":"If the central overpotential results are reliable, the paper identifies a genuinely new catalyst architecture in which the carbon framework of a metal-encapsulated biphenylene bilayer is the active site, with competitive ORR/OER activity and improved resistance to metal dissolution. The work uses a standard and internally consistent computational protocol: spin-polarized PBE+U with linear-response Hubbard U values, vdW-DF2 dispersion, VASPsol solvation, vibrational zero-point and entropy corrections, four magnetic configurations per system, and explicit intercalation-energy and AIMD stability checks. The simulated XPS fingerprints also provide useful experimentally testable predictions. However, the descriptor claims in the abstract and conclusions are stronger than the evidence supports: the 'volcano' correlations are post hoc fits to the same DFT dataset that produced the overpotentials, with exclusions tuned to maximize R², and one branch of each volcano has essentially zero correlation. The reliability of the headline overpotentials additionally depends on an untested structural premise for the best catalysts.","major_comments":[{"comment":"The computed free energies assume that the intercalated metal remains at its clean square-site position while O*, OH*, and OOH* bind at C468. The manuscript itself documents that this premise fails for Nb, W, Os, and Ti, where adsorption displaces the metal from the square site and strongly raises the system energy (footnote 82, SM Fig. S.11). No equivalent check is reported for Cu, Pt, Ru, Mn, or Fe, the systems that carry the central claims of Table III and Figs. 7–8. The AIMD simulations in Fig. 6 and SM Fig. S.12 cover only the clean systems and therefore cannot rule out adsorbate-induced metal displacement under reaction conditions. If any of the headline metals relaxes off-site upon binding an intermediate, the adsorption free energies, onset potentials, and the d-orbital populations used as descriptors no longer describe the operating active site. This premise must be tested for all systems that contribute to the main conclusions.","section":"Sec. III.B, footnote 82, SM Fig. S.11"},{"comment":"The volcano-descriptor claim is not supported by the fits as presented. In Fig. 7(a), Cu and Pt are excluded to obtain the right-branch R² = 0.45; in Fig. 8(c) the d-orbital ORR right branch has R² = 0.10, and in Fig. 8(e) the p-orbital ORR right branch has R² = 0.15. These right-branch values are equivalent to no correlation, so the 'volcanoes' consist of a correlated left branch plus a scatter of points that are removed to maximize R². The paper states that exclusions were made '[t]o maximize the R² values of the linear fits,' which makes the correlations post hoc rather than predictive. Since the descriptors are computed from the same DFT dataset that produced the overpotentials, the abstract's statement that the descriptors 'can predict catalytic behavior' is not justified; the manuscript should reframe these as trends and test them on held-out systems or with a pre-specified exclusion criterion.","section":"Sec. III.B.2, Figs. 7 and 8"},{"comment":"There is an internal inconsistency between the reported overpotentials and the descriptor-based ranking. Table III lists B/Cu/B (η_ORR = 0.42 V) and B/Pt/B (η_ORR = 0.44 V) as the best ORR catalysts, yet the Conclusions state that 'B/Mn/B lies closest to the ORR optimum' on the basis of the d-orbital charge population. This conclusion is reached only because Cu and Pt are excluded from the ORR fits in Fig. 8. The manuscript should either provide a physical reason for excluding these systems or soften the claim that the descriptor identifies the optimal catalyst; as written, the descriptor does not rank-order the actual best-performing systems.","section":"Table III, Fig. 8, Conclusions"}],"minor_comments":[{"comment":"Equation (13) integrates over the d-band up to the vacuum level Evac, while the text states that eigenvalues are referenced to the Fermi level; please clarify the energy reference used in the band-center definition.","section":"Sec. II, Eq. (13)"},{"comment":"The phrase 'C-1score-level' is a typo; it should read 'C-1s core-level.'","section":"Sec. II"},{"comment":"In the definition of the net charge density difference, 'B e B' should read 'B/B'; the sentence is also missing a closing phrase after 'ρ_M and ρ_B/B'.","section":"SM, Eq. (E.2)"},{"comment":"The word 'painel' appears twice and should be 'panel.'","section":"Fig. 6 caption"},{"comment":"The caption states that systems with η_ORR < 3.5 eV and η_OER < 2.0 eV 'are indicated by blue squares,' but the phrasing is ambiguous about whether all shown points are blue squares; please reword for clarity.","section":"Fig. 7 caption"},{"comment":"The Supplemental Material link is a placeholder '[http://site]'; please provide the actual URL or DOI.","section":"Reference 66"}],"recommendation":"major_revision","confidential_remarks":"The core computational protocol is sound and the work is within the journal's scope, but the two load-bearing issues must be addressed: (i) verifying the fixed-metal premise for the headline catalysts under intermediate adsorption, and (ii) reframing the descriptor analysis as post hoc trends rather than predictions, given the tuned exclusions and the near-zero right-branch R² values. If the geometry check passes for Cu, Pt, Ru, Mn, and Fe, and the descriptor language is moderated, the manuscript would be suitable for publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThis paper is a first-principles screening of metal-intercalated biphenylene bilayers (B/M/B) for ORR/OER, with the carbon framework as the active surface. The particular architecture—metal sandwiched between two biphenylene layers, with C468 as the catalytic site—is new for these reactions; previous work looked at pristine, doped, or decorated monolayers. The overpotential calculations follow the standard CHE recipe with PBE+U, vdW, solvation, and ZPE corrections, and the numbers are plausible. Cu and Pt are predicted to beat Pt(111) for ORR, and Fe beats IrO2 for OER. If even half of that survives more careful checks, it is a useful addition to the search for stable non-noble catalysts.\n\nWhat the paper does well: the structural characterization (intercalation energies, magnetic phases, XPS core-level shifts) is thorough, the AIMD stability checks are a good faith effort, and the authors are transparent about which systems deviate from their descriptor trends. The descriptor analysis is honestly labeled as qualitative.\n\nThe soft spots are real, though. The central premise—that the metal stays at the square site and all intermediates bind at C468—is only tested for O2, and only for the systems where it fails. The stress-test note is right: for Nb, W, Os, and Ti, adsorbing O*, OH*, or OOH* displaces the metal and wrecks the energetics. The authors never report that check for Cu, Pt, Ru, Mn, or Fe, which carry the headline overpotentials. If any of those metals relax off-site under an intermediate, the computed free energies and the d-orbital populations no longer describe the operating catalyst. Given that the paper itself shows this happens for a third of the metals, the omission is not a minor gap—it is a check the central claim depends on.\n\nThe descriptor volcano plots are also weaker than the abstract implies. Systems are excluded to maximize R2, and in one case the excluded point (Cu) is the best catalyst. The paper acknowledges this, but the claim that d-orbital charge 'predicts' activity is post hoc, not predictive. The band-center analysis is appropriately hedged, but the charge-population analysis inherits the same problem.\n\nBottom line: this is a competent computational study of a new material class, and it deserves a serious referee. The referee should insist on geometry stability checks for the promising systems with each intermediate, and should treat the descriptor trends as suggestive rather than established. A revision addressing those points would make the paper genuinely solid.\n\nRecommendation: send it to review. Not a desk reject.","headline":"Solid computational screening of a new catalyst architecture; descriptor claims are overfit and the active-site geometry is unverified for the headline systems.","tokens_in":31537,"tokens_out":2043,"would_cite":false,"duration_ms":18749,"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":"Metal atoms sandwiched between biphenylene layers turn the carbon framework into a competitive oxygen electrocatalyst, with overpotentials as low as 0.42 V for ORR and 0.44 V for OER.","keywords":["biphenylene bilayer","metal intercalation","oxygen reduction reaction","oxygen evolution reaction","computational hydrogen electrode","d-band center","orbital charge population","density functional theory"],"falsifier":"Compute the adsorption free energies of O*, OH*, and OOH* on the carbon site bridging hexagonal and octagonal rings and on the intercalated metal atom for B/Cu/B and B/Fe/B; if any intermediate binds more strongly at a non-C468 site, the reported overpotentials do not describe the full reaction path.","tokens_in":30571,"feed_emoji":"⚡","tokens_out":8151,"duration_ms":66764,"temperature":0.7,"pith_summary":"This paper claims that transition metals intercalated between two biphenylene carbon layers turn the carbon framework itself into a competitive oxygen electrocatalyst, without exposing the metal to the electrolyte. Using density functional theory and the computational hydrogen electrode, it predicts ORR overpotentials as low as 0.42 V for B/Cu/B and 0.44 V for B/Pt/B, beating the Pt(111) benchmark, and an OER overpotential of 0.44 V for B/Fe/B, below IrO2. The paper further claims that the d-orbital charge population of the encapsulated metal is a predictive descriptor: ORR activity peaks near 5.31 e− and OER activity near 6.51 e−, giving volcano plots that single out Mn for ORR and Fe for OER. If correct, this identifies a design principle where cheap carbon hosts, not noble metals, carry the catalysis.","feed_headline":"Copper tucked in carbon bilayer rivals platinum for oxygen reduction","feed_subtitle":"DFT predicts sandwiched copper and platinum beat Pt(111); iron best for water oxidation.","key_machinery":"The central mechanism is the computational hydrogen electrode (CHE), which converts DFT adsorption free energies of OOH*, O*, and OH* on the C468 carbon site into onset potentials and overpotentials for ORR and OER. The explanatory engine is the descriptor analysis: the d-band center, pz-band center, and their difference, plus orbital charge populations obtained from projected wave functions. The descriptor that carries the paper's conclusion is the metal d-orbital charge population, which forms volcano plots peaking around 5.31 e− for ORR and 6.51 e− for OER, identifying Mn and Fe as optimum systems. This links a simple electronic count to catalytic activity in a system where the metal is not the active site.","core_discovery":"In the paper's own terms, the discovery is that metal-intercalated biphenylene bilayers B/M/B are a new family of ORR/OER electrocatalysts in which the active site is a carbon atom, C468, at the square-planar sublattice, while the intercalated metal d-orbital charge population acts as a descriptor of activity. The best predicted systems are B/Cu/B and B/Pt/B for ORR (η = 0.42 and 0.44 V) and B/Fe/B for OER (η = 0.44 V). The paper argues that band centers alone do not correlate simply with overpotentials but that (εpz − εd) and especially the metal d-orbital charge population produce volcano-type plots consistent with the Sabatier principle. It also predicts distinct C-1s core-level shifts in the biphenylene bilayer upon intercalation, giving an XPS fingerprint that experiments could use to verify metal placement.","pith_inferences":["The paper does not compute kinetic barriers, so we infer that the thermodynamic overpotentials could understate the real ones if any O*→OH* step has a large activation barrier.","The paper reports metal displacement upon adsorption for Nb, W, Os, and Ti; we infer that the same check is still open for the best catalysts Cu, Pt, Ru, Mn, and Fe.","We infer that the d-orbital-charge descriptor may transfer to other carbon allotropes with square-planar carbon sites, but the paper does not test this.","A direct experimental test would be to synthesize B/Cu/B and B/Fe/B films and compare measured onset potentials with the predicted 0.42 V and 0.44 V, though support and pH effects would complicate the comparison."],"forward_implications":["If the predictions hold, B/Cu/B and B/Pt/B beat the Pt(111) benchmark for ORR with overpotentials of 0.42 V and 0.44 V, respectively.","B/Fe/B becomes a candidate OER catalyst with an overpotential of 0.44 V, below the IrO2 benchmark of 0.65 V.","The metal d-orbital charge population could serve as a fast screening descriptor for new intercalated bilayers, before full reaction-energy calculations.","The predicted C-1s core-level shifts give an experimental XPS fingerprint for verifying where the metal sits and how much charge it transfers.","Encapsulation should suppress metal dissolution, directly addressing the stability limitation of single-atom catalysts."],"supporting_citations":[{"why":"Supplies the computational hydrogen electrode framework that converts adsorption free energies into onset potentials and overpotentials.","marker":"[71]"},{"why":"Provides the universal OER activity descriptor ΔGO*−ΔGOH* and the IrO2 benchmark used for comparison.","marker":"[13]"},{"why":"Establishes the optimal d-band-center range for oxygen electrocatalysis adopted in the descriptor analysis.","marker":"[12]"},{"why":"Identifies the square-planar carbon sites of biphenylene as electron-rich and reactive, justifying C468 as the active site.","marker":"[45]"},{"why":"Gives the Pt(111) ORR benchmark overpotential used to judge competitiveness.","marker":"[86]"},{"why":"Provides the linear-response Hubbard U method used to set U values for the transition metals.","marker":"[65]"}],"fun_headline_variants":["Metal-stuffed carbon bilayers rival Pt for ORR, Fe leads OER","Biphenylene bilayers with intercalated metals: ORR and OER catalysts","Cu in carbon bilayer matches Pt for oxygen reduction, says DFT","Fe intercalation makes carbon bilayer efficient OER catalyst","Metal d-population in carbon bilayers predicts ORR and OER activity"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The results rest on the assumption that the square-coordinated carbon site is the only active site, that the intercalated metal never leaves its equilibrium position during the reaction, and that thermodynamic free energies, not kinetic barriers, determine the overpotential.","fun_headline_variants_meta":{"raw":{"variants":["Metal-stuffed carbon bilayers rival Pt for ORR, Fe leads OER","Biphenylene bilayers with intercalated metals: ORR and OER catalysts","Cu in carbon bilayer matches Pt for oxygen reduction, says DFT","Fe intercalation makes carbon bilayer efficient OER catalyst","Metal d-population in carbon bilayers predicts ORR and OER activity"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000777,"raw_usage":{"total_tokens":3519,"prompt_tokens":1112,"completion_tokens":2407,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":728,"completion_tokens_details":{"reasoning_tokens":2309}},"tokens_in":728,"tokens_out":2407,"duration_ms":17026,"temperature":1.0,"reasoning_tokens":2309,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T04:12:52.985371+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the adsorption free energies of O*, OH*, and OOH* on the carbon site bridging hexagonal and octagonal rings and on the intercalated metal atom for B/Cu/B and B/Fe/B; if any intermediate binds more strongly at a non-C468 site, the reported overpotentials do not describe the full reaction path.","supporting_citations":[{"cited_title":"and Sumpter, Bobby G","cited_arxiv_id":null,"evidence_quote":"Supplies the computational hydrogen electrode framework that converts adsorption free energies into onset potentials and overpotentials."},{"cited_title":"Journal of the American Chemical Society , year=","cited_arxiv_id":null,"evidence_quote":"Provides the universal OER activity descriptor ΔGO*−ΔGOH* and the IrO2 benchmark used for comparison."},{"cited_title":"Angewandte Chemie International Edition , volume =","cited_arxiv_id":null,"evidence_quote":"Provides the linear-response Hubbard U method used to set U values for the transition metals."}],"review_version":1}