{"id":"d40be0ab-402b-4368-a84f-5ee1880df1ec","arxiv_id":"2607.15066","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"A range-separated hybrid van der Waals functional with a molecularly tuned screening parameter predicts the H2/Cu(111) dissociation barrier within about 1 kcal/mol of the reference value.","lead":"This paper tests a family of range-separated hybrid van der Waals density functionals on the energy barrier for H2 breaking apart on a copper surface, and proposes tuning the functional's screening parameter from molecular properties. If the recipe holds, it gives a relatively cheap non-empirical way to predict reaction barriers on metal catalysts that previously required very expensive calculations.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"H2 optimal tuning at γ* is incomplete; barrier agreement may reflect size-boundary constraint rather than QP-guided selection.","rationale":"The reader identified the incomplete H2 OT as the weakest assumption, and I agree. This is load-bearing because the paper's core methodological claim—that AHBR(γ*) is a non-empirical, QP-guided descriptor—depends on the OT selection being meaningful. If the OT criterion is not satisfied at γ*, the selection is governed by the γ'' size limit, not by QP alignment. The paper is transparent about this limitation, but it does not reconcile it with the 'best-possible non-empirical' framing. The proposed test directly settles this: locating the true OT γ and computing the barrier there would show whether the near-chemical-accuracy result is tied to the QP criterion or to the boundary. If the barrier at γ_OT is similar, the claim is strengthened; if not, the paper's central claim is reduced to an empirical observation at a constrained boundary. The reader's CONDITIONAL verdict is appropriate; no further adjustment is needed.","tokens_in":34209,"tokens_out":4971,"duration_ms":53328,"concrete_test":"Perform AHBR-mRSH(γ) tuning for isolated H2 at γ values above 0.5 a0^-1 (e.g., 0.6, 0.7, 0.8, 1.0) to find the γ_OT where –εH equals the functional's own adiabatic IP (≈16.96 eV). Then compute the H2+Cu(111) dissociation barrier with AHBR(γ_OT) using the same convergence protocol (k=12, q=4, 2×2 cell). If the barrier at γ_OT differs from 0.678 eV by more than 1 kcal/mol (0.043 eV), the close agreement at γ*=0.5 is not a robust consequence of QP-OT selection. Also evaluate piecewise linearity and derivative discontinuity at γ_OT to assess whether the QP-consistency argument holds.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central non-empirical justification for AHBR(γ*) is the constrained optimal tuning (OT) of AHBR-mRSH(γ) to align the H2 HOMO level with the adiabatic IP. Table VII shows this alignment is not achieved at the selected γ* = 0.5 a0^-1: –εH(0.5) = 16.30 eV versus IP(0.5) = 16.96 eV, a mismatch of 0.66 eV. The paper itself acknowledges in Appendix A that 'alignment... does require one to go to a larger γ'' value.' Thus γ* is not a converged OT value but the upper boundary of the allowable molecular-size constraint (1/γ'' = 2a0). The barrier prediction E_AHBR(γ*) = 0.678 eV is close to chemical accuracy, but this agreement could be coincidental: the barrier varies strongly with γ in this regime (Table III: 0.725 eV at γ=0.106, 0.689 eV at 0.450, 0.678 at 0.500). If the true OT γ lies beyond 0.5, the corresponding barrier may deviate significantly from the SRP reference, undermining the claim that the near-chemical-accuracy result is a consequence of QP-guided, non-empirical selection rather than of picking the size-limit boundary.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper tests the range-separated hybrid van der Waals density functional AHBR and its gamma-tuned variants for the classical dissociative chemisorption barrier of H2 on Cu(111) and for CO adsorption-site preferences on Cu, Ag, Au, and Pt(111). The central proposal is a 'constrained optimal tuning' procedure: first tune the molecule-optimized AHBR-mRSH(gamma) functional so that the H2 HOMO level matches the adiabatic IP, while respecting a size-based upper limit gamma'' = 0.5 a0^-1; then use the corresponding metal-compatible AHBR(gamma*) for the surface calculation. The paper reports AHBR(gamma* = 0.5 a0^-1) barrier of 0.678 eV, compared with the SRP-DFT reference of 0.628 eV (elsewhere 0.635 eV), and qualitatively correct CO top-site preferences.","tokens_in":34629,"tokens_out":3757,"duration_ms":42271,"significance":"If the non-empirical justification were fully established, the paper would offer a practical route to near-chemical-accuracy surface-reaction barriers within a single-parameter RSH vdW-DF framework, with the parameter set by molecular QP physics rather than by fitting to the target barrier. The strength of the paper lies in its transparent convergence documentation (Tables I and II), the independent CO chemisorption cross-check, and the explicit acknowledgment that the H2 OT criterion is not completed at the selected gamma*. These features make the work reproducible and the central limitation clearly visible. However, the main claim is conditional on the transfer of a not-fully-converged molecular tuning parameter to the metal-surface calculation, which is currently the weakest link.","major_comments":[{"comment":"The optimal-tuning criterion is not satisfied at the selected gamma* = 0.5 a0^-1: Table VII gives -epsilon_H = 16.30 eV versus IP = 16.96 eV, a mismatch of 0.66 eV, and the text itself states that alignment requires a larger gamma value. Thus gamma* is not a converged OT value but the imposed upper boundary gamma''. Since E_B varies from 0.689 eV at gamma = 0.450 to 0.678 eV at gamma = 0.500 (Table III), the near-agreement with SRP could arise from choosing the boundary rather than from QP physics. The paper should either show that the barrier is insensitive to gamma in the range where H2 OT would actually converge, or provide an independent test of AHBR(gamma*) that does not rely on this boundary pick.","section":"Appendix A, Table VII"},{"comment":"The barrier is a monotonically decreasing function of gamma in the range 0.106-0.500 a0^-1, with a slope of roughly 0.1 eV per 1 a0^-1 at the upper end. The choice gamma* = gamma'' is therefore not a stationary point. Without an estimate of the uncertainty associated with the incomplete OT, the claim of 'near chemical accuracy' is not robust. A sensitivity analysis or an error bar derived from the spread of molecule-specific OT gamma values would be needed to support the central conclusion.","section":"Table III / Fig. 3"},{"comment":"The CO site-preference cross-check is performed with the default AHBR (gamma = 0.106 a0^-1), not with the proposed AHBR(gamma* = 0.5 a0^-1). The paper asserts that molecular-energy differences have only a soft gamma dependence, but no CO adsorption data at gamma* are given. Consequently, the CO results do not validate the specific descriptor AHBR(gamma*); they only validate the default AHBR. Please report at least one CO site-preference value at gamma* or explain why the transferability of the gamma* choice to CO is established.","section":"Table II and Sec. IV"},{"comment":"The SRP-DFT reference barrier is quoted inconsistently as 0.628 eV in the Introduction and Table III and as 0.635 eV in Sec. II.A. Since the central claim is that AHBR(gamma*) is within chemical accuracy (1 kcal/mol = 0.043 eV) of this target, the discrepancy matters: 0.678 eV is 0.050 eV above 0.628 eV but 0.043 eV above 0.635 eV. The paper must state which value is the benchmark and discuss whether the result is within or slightly outside chemical accuracy.","section":"Sec. II.A, Table III, abstract"}],"minor_comments":[{"comment":"Typographical errors include 'slap' for 'slab' (Sec. III), 'inverse-scatting' for 'inverse-scattering' (Sec. II.A), and 'tricker' for 'trigger' (Introduction). These should be corrected.","section":"General"},{"comment":"The convergence table is incomplete: for k=6, q=5 and q=6 entries are blank, and for k=10 only the q=4 result is shown. Please provide the full grid or indicate which entries were not computed.","section":"Table I"},{"comment":"The row label '∆BEAHBR H2-Cu' appears to mix notation; it should be consistent with E_B^{AHBR} used in the text.","section":"Table III header"},{"comment":"The equivalence argument between AHBR(gamma) and AHBR-mRSH(gamma) would benefit from an explicit statement that it applies to total exchange energies, not to individual orbital levels. Table VII shows HOMO energies differing by more than 5 eV at gamma=0.5, which could confuse readers.","section":"Sec. II.B, Fig. 2"}],"recommendation":"major_revision","confidential_remarks":"The paper is in scope and the central idea is interesting, but the non-empirical justification is incomplete as presented. The author is encouraged to address the OT convergence issue and the gamma-boundary concern directly. I see no grounds for rejection, but the current version overstates the strength of the QP-guided selection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Per Hyldgaard reports new numerical data on a range-separated hybrid vdW-DF: the gamma-dependence of the H2/Cu(111) dissociation barrier, carefully converged CO site-preference energies on four metals, and molecular quasiparticle tables. The convergence documentation in Table I is solid, and the molecular QP results are genuinely useful. The functional is implemented in QE, so the numbers are reproducible. The paper is also honest: it says 'approach chemical accuracy' rather than claiming full chemical accuracy.\n\nWhere it gets soft is the gamma* story. The abstract's 'converged OT' phrasing oversells. Table VII shows that at gamma*=0.5 a0^-1 the H2 HOMO is -16.30 eV while the functional's own adiabatic IP is 16.96 eV—a 0.66 eV gap. Appendix A explicitly says alignment requires a larger gamma''. So gamma* is the upper boundary of the molecule-size constraint, not a converged optimal-tuning value. The barrier at gamma* is 0.678 eV, compared to the SRP reference 0.628 eV (or 0.635 eV elsewhere in the paper—the inconsistency is a minor but real blemish). The agreement is within about 1 kcal/mol, but it's hard to say it's a consequence of QP-guided selection rather than of taking the size-limit boundary, since the barrier still varies with gamma in this regime (0.689 eV at 0.450, 0.678 at 0.500).\n\nTwo lesser issues: the CO site-preference cross-check is run at default gamma=0.106, not at gamma*, so it doesn't test the proposed functional directly. And the barrier geometry comes from SRP-DFT, so the benchmark comparison is not fully independent. Neither is disqualifying.\n\nBottom line: this is a solid numerical contribution, but the central methodological claim is not as clean as the abstract implies. The audience is DFT method developers and surface-science groups working on barriers and site preferences. The paper is worth a serious referee. The referee should ask for either a completed OT within the allowed gamma range, a direct CO test at gamma*, or an explicit reclassification of gamma* as a physically constrained parameter. With that revision, it's a good methods paper.","headline":"Real numbers and honest hedging, but gamma* is the size-limit boundary, not a completed optimal tuning.","tokens_in":35043,"tokens_out":4941,"would_cite":true,"duration_ms":51509,"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 range-separated hybrid van der Waals density functional predicts the H2 dissociative chemisorption barrier on Cu(111) to within about 1 kcal/mol of the fitted reference.","keywords":["dissociative chemisorption","van der Waals density functional","range-separated hybrid","H2 on Cu(111)","quasi-particle","optimal tuning","CO adsorption site preference","barrier height"],"falsifier":"Re-run the H2+Cu(111) barrier calculation at a γ value slightly larger than 0.5 a0^-1 where AHBR-mRSH(γ) actually brings the H2 HOMO level into alignment with the adiabatic IP (i.e., beyond the size-limit cutoff). If the barrier shifts by more than 1 kcal/mol away from 0.678 eV, the near-chemical-accuracy result is a consequence of the hard cutoff rather than a converged non-empirical choice.","tokens_in":34095,"feed_emoji":"⚛️","tokens_out":8015,"duration_ms":75085,"temperature":0.7,"pith_summary":"The paper claims that a particular range-separated hybrid van der Waals density functional, AHBR(γ*), can describe the classical barrier for H2 dissociating on Cu(111) with near-chemical accuracy: a predicted barrier of 0.678 eV against a 0.628 eV benchmark from fitted SRP-DFT, a deviation of about 1 kcal/mol. The functional achieves this while also giving qualitatively correct CO adsorption site preferences on four transition-metal surfaces, an independent check that its charge-transfer behavior is plausible. The central idea is to select the functional's range-separation parameter γ non-empirically by tuning it so that isolated-molecule quasi-particle levels match adiabatic ionization potentials, then importing that value into the metal-surface calculation. This matters because dissociative chemisorption is a rate-limiting step in catalysis, and getting barrier heights reliably from first principles without system-specific fitting has been an open challenge.","feed_headline":"H2 dissociation barrier on Cu(111) predicted to 1 kcal/mol","feed_subtitle":"A non-empirical hybrid van der Waals functional reproduces the fitted reference barrier and gets CO adsorption sites right.","key_machinery":"The key object is AHBR(γ), a range-separated hybrid van der Waals density functional that mixes a fraction α=0.25 of short-range Fock exchange into the B86R semilocal exchange, with an inverse length scale γ controlling the crossover to screened exchange at large electron-hole separations. The paper uses the molecule-optimized sibling AHBR-mRSH(γ) to perform optimal tuning: adjusting γ until the HOMO level of a small molecule aligns with its adiabatic ionization potential. That tuned γ is then imported into the metal-compatible AHBR(γ) form for the surface calculation, justified by the claim that for the compact H2 molecule the two functionals give equivalent exchange descriptions when γ ≤ 0","core_discovery":"The AHBR(γ*) descriptor, with γ* = 0.5 a0^-1, predicts the H2+Cu(111) dissociative chemisorption barrier at 0.678 eV, within about 1 kcal/mol of the SRP-DFT reference value of 0.628 eV (0.635 eV elsewhere in the paper). The same functional family, at its default setting, predicts the experimentally observed top-site preference for CO on Cu(111), Ag(111), Au(111), and Pt(111), which standard semilocal functionals get wrong. The paper argues that the choice of γ* is guided by optimal tuning of molecular quasi-particle levels in the molecule-optimized AHBR-mRSH(γ) generalization, and that the resulting AHBR(γ*) retains a derivative discontinuity that suppresses spurious charge transfer, making","pith_inferences":["Because the chosen γ* sits at the size-limit boundary rather than at a converged optimal-tuning point, a natural next step is to compute the barrier for slightly larger γ (e.g., 0.55 a0^-1) and see whether the near-chemical-accuracy result is robust or an artifact of the cutoff; the paper's own data show the barrier falls toward 0.550 eV as γ→∞.","The same tuning recipe could be tested on other benchmarked dissociative chemisorption systems (e.g., O2 or N2 on metal surfaces) where SRP-DFT barriers exist, to see whether the non-empirical protocol generalizes.","If the derivative-discontinuity argument is right, the functional should also improve descriptions of charge transfer in other molecule-surface systems where delocalization errors plague GGA, such as CO2 or NO adsorption."],"forward_implications":["If accurate, AHBR(γ*) gives a computationally cheaper alternative to RPA-level treatments for the H2+Cu(111) barrier, with near-chemical accuracy.","The correct CO site preferences suggest the functional captures the forward- and back-donation balance that GGA and many vdW-DFs miss.","The QP-tuning protocol is transferable in principle: the paper finds the tuned γ decreases with adsorbate size, suggesting extensions to other small-molecule/metal reactions.","The barrier prediction can be fed into dynamical simulations of sticking probability and compared with molecular-beam experiments, providing a direct experimental test.","The paper notes a practical 45% AHBR / 55% vdW-DF2-b86r merger that exactly matches the SRP-DFT barrier, but emphasizes it is empirical rather than non-empirical."],"fun_headline_variants":["H2 barrier on Cu(111) to 1 kcal/mol from new hybrid","Hybrid vdW functional nails H2 dissociation on Cu(111)","AHBR(γ*) hits H2 barrier on Cu(111) and CO adsorption sites","Non-empirical hybrid predicts H2/Cu(111) barrier within 1 kcal/mol"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The protocol assumes that a range-separation parameter tuned against isolated-molecule ionization potentials can be imported into the metal-surface calculation, even though at the chosen γ* = 0.5 a0^-1 the H2 HOMO level (-16.30 eV) does not match the functional's own adiabatic IP (16.96 eV), so the optimal-tuning criterion is not actually satisfied; the choice is the size-limit boundary, not a converged OT value.","fun_headline_variants_meta":{"raw":{"variants":["H2 barrier on Cu(111) to 1 kcal/mol from new hybrid","Hybrid vdW functional nails H2 dissociation on Cu(111)","AHBR(γ*) hits H2 barrier on Cu(111) and CO adsorption sites","Non-empirical hybrid predicts H2/Cu(111) barrier within 1 kcal/mol"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.0012,"raw_usage":{"total_tokens":4840,"prompt_tokens":858,"completion_tokens":3982,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":602,"completion_tokens_details":{"reasoning_tokens":3892}},"tokens_in":602,"tokens_out":3982,"duration_ms":32583,"temperature":1.0,"reasoning_tokens":3892,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T00:13:57.242493+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-run the H2+Cu(111) barrier calculation at a γ value slightly larger than 0.5 a0^-1 where AHBR-mRSH(γ) actually brings the H2 HOMO level into alignment with the adiabatic IP (i.e., beyond the size-limit cutoff). If the barrier shifts by more than 1 kcal/mol away from 0.678 eV, the near-chemical-accuracy result is a consequence of the hard cutoff rather than a converged non-empirical choice.","supporting_citations":[],"review_version":1}