{"id":"aed3ff75-64f4-4864-90ff-0449603cddf7","arxiv_id":"2608.10432","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A thickness criterion based on surface atomic density change predicts when ultrathin metal sheets switch surface reconstruction, unifying the suppressed (1x2) on 5d (110) sheets and the induced quasi-hexagonal reconstruction on 4d (001) sheets.","lead":"The authors derive a simple thickness rule: when a metal nanosheet gets thinner, it prefers the surface reconstruction that packs more atoms per unit area. The rule explains two opposite behaviors in DFT calculations of noble metal sheets and gives a critical thickness from bulk surface energies alone.","discovery_kind":"unification","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Per-atom convention is not fatal—fixed-N gives the same l_c in the linear model—but the DFT validation never actually tests the fixed-N ensemble that governs a real nanosheet.","rationale":"I re-derived Eq. (4) and the fixed-N energy difference and found that they share the same zero condition in the idealized linear surface-energy model. The reader's assertion that a fixed-N ensemble would not give the same rule is therefore not correct for the model as written. However, the paper's DFT dots are per-atom energies of separately relaxed supercells with different atom counts and different equilibrium areas; they are neither the fixed-area model nor the physical fixed-N comparison. The surface-stress deviations for Pt and Au (Figs. 2e and 2f) show that the linearity assumption is imperfect, and the grand-canonical ensemble would remove the thickness dependence entirely. Because the paper never states the thermodynamic boundary conditions and never performs a fixed-N comparison, the central claim that Eq. (6) predicts the physical critical thickness of an isolated nanosheet remains conditional. The derivation is algebraically sound, the DFT is internally consistent, and the paper is transparent about its limitations, but the missing fixed-N test is the most load-bearing gap. I therefore keep the reader's CONDITIONAL verdict rather than moving to ACCEPT or REJECT.","tokens_in":6962,"tokens_out":25118,"duration_ms":257600,"concrete_test":"For Pt(110) and Rh(001), construct fixed-N pairs near the predicted l_c: choose a total atom count N = l m for a given integer l (for example, l = 10 for Pt and l = 5 for Rh), build the unreconstructed slab with that N, and build the reconstructed slab with the same N by changing the in-plane supercell area so that the surface layer density is n/m while keeping the same number of layers; fully relax both structures and compare total energies directly. Repeat for l one layer above and one layer below the Table I critical values. If the sign of E_RE(N) - E_UN(N) switches at l approximately equal to l_c from Eq. (6), the physical fixed-N prediction is confirmed; if not, the per-atom DFT validation has masked a fixed-N correction. Record the resulting lateral area difference, since the fixed-N equivalence assumes strain-free area adjustment.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The reader's weakest-assumption identifies a real issue, but it can be partially answered. Eqs. (1)-(6) compare E/N for slabs with different total atom counts at fixed lateral area, while an isolated nanosheet has fixed N. However, in the same linear surface-energy model, a fixed-N comparison (same number of layers l, lateral area adjusted so total N is equal) gives E_UN(N) = N E_bulk + 2N E_S-UN/(l m) and E_RE(N) = N E_bulk + 2N E_S-RE/((l-2)m + 2n). Setting their difference to zero yields exactly Eq. (6), l_c = 2(n/m - 1) E_S-UN / ΔE_S. Thus the per-atom convention is not an arithmetic trick under the model's assumptions; the fixed-N ensemble gives the same critical thickness. The load-bearing residual concern is that the DFT validation does not implement either the fixed-area per-atom model or the physical fixed-N comparison: each supercell is fully optimized, so the unreconstructed and reconstructed cells have different atom counts and different equilibrium in-plane areas. The deviations visible for Pt and Au in Figs. 2(e) and 2(f) show that surface stress is not negligible, so the linearity assumption is imperfect. Consequently, the agreement in Table I validates the per-atom model but does not by itself prove that a real fixed-N sheet reverses at l_c. In addition, the grand-canonical ensemble (a sheet exchanging atoms with a reservoir) would show no thickness dependence at all, so the 'universal' claim is ensemble-dependent; the paper never states the assumed boundary conditions. A fixed-N DFT calculation is therefore the missing test.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a thickness criterion for surface reconstruction in ultrathin metal nanosheets. Starting from a slab model in which each surface is either unreconstructed or reconstructed, the authors derive an expression for the difference in energy per atom, ΔE, between the two structures, and from it a critical thickness l_c = 2(n/m − 1) E_S−UN / ΔE_S at which the reconstruction tendency reverses. They argue that thinning always favors structural changes that increase the surface atomic density, unifying the suppression of the (1×2) missing-row reconstruction on 5d noble-metal (110) sheets and the induction of quasi-hexagonal reconstruction on 4d noble-metal (001) sheets. The prediction is compared with PBEsol DFT calculations for Ir, Pt, Au, Rh, Pd, and Ag slabs, and the critical thicknesses from the model, l_c, are compared with estimates from direct total-energy calculations, l_c*. The paper concludes that surface stress and quantum oscillations introduce only minor corrections and that the rule is not limited to noble metals.","tokens_in":7347,"tokens_out":7228,"duration_ms":70556,"significance":"If correct, Eq. (6) would be a useful parameter-free design rule: it predicts a nanosheet's reconstruction reversal using only bulk surface energies and surface atomic density ratios, without computing the sheet itself. The derivation is transparent, the algebra of Eqs. (1)–(6) checks out, and the DFT trends are internally consistent with the model's sign structure. The paper also gives credit-worthy attention to the Pd(001) magnetic state, showing that ferromagnetism disappears upon full optimization and therefore does not affect the reconstruction tendency. The main significance is tempered, however, by three issues: the thermodynamic ensemble underlying the per-atom comparison is not stated; the DFT validation is performed in a different ensemble from the physical fixed-N nanosheet; and the 'universal' claim extends beyond the tested materials and reconstruction patterns.","major_comments":[{"comment":"The stability criterion in Eq. (3) compares average energies per atom for structures with different total atom counts, which is only one possible thermodynamic convention. For an isolated nanosheet, fixed N is the physical ensemble, while for a sheet exchanging atoms with a reservoir the grand potential should be used; the paper cites refs. [15,16] but does not justify the choice or state the assumed boundary condition. I checked that a fixed-N comparison under the same linear model (same total N, lateral area adjusted) reproduces exactly the same critical thickness as Eq. (6), so the sign structure is not a pure arithmetic artifact. However, the paper does not make this check, and the DFT validation does not implement the fixed-N ensemble: each supercell is fully optimized, so the unreconstructed and reconstructed cells have different atom counts and different equilibrium in-plane areas. Thus the black dots in Figs. 2 and 3 validate the per-atom model of Eq. (4), but they do not by themselves prove that a real fixed-N sheet reverses at l_c. The manuscript should either state the ensemble explicitly and discuss the fixed-N equivalence, or test the fixed-N constraint directly in the DFT calculations.","section":"Eqs. (1)–(6), surrounding text"},{"comment":"The agreement between l_c and l_c* is used as the main evidence that surface stress and quantum oscillations are minor corrections. For Pt and Au, however, the directly calculated ΔE lies systematically below the model curve, an effect the authors attribute to surface stress without quantifying it. For the (001) systems, l_c* differs from l_c by one layer for Rh, Pd, and Ag, i.e., 20–33% of l_c for these 3–6-layer sheets. Given that l_c is only a few layers, a one-layer uncertainty can change the qualitative prediction for a specific thickness. The paper should quantify the surface-stress contribution (for example, by computing surface stress tensors for the reconstructed surfaces) and provide a less ambiguous definition of l_c* than the thickness at which ΔE is closest to zero, ideally with an interpolation or a fitted crossing point.","section":"Table I, Figs. 2(e)–2(f), 3(c)–3(f)"},{"comment":"The statement that 'thinning always favors structural changes that increase surface atomic density' is presented as a universal rule, but the derivation assumes the reconstruction is confined to the two surface layers and that E_S−UN and ΔE_S are thickness-independent. The paper itself adds a caveat for the n=m case (W(001), Mo(001)), where l_c=0 and stress or quantum oscillations could cause a crossing, which undercuts the word 'always'. The DFT evidence covers only fcc noble metals and two reconstruction patterns, so the concluding extrapolation that the criterion is 'not specific to noble metals' is not supported by the data presented. I recommend either restricting the claims to the tested class or providing explicit arguments (or additional test cases) showing that the sign rule survives beyond the present materials.","section":"Section 'Finally, we comment...', Fig. 4"}],"minor_comments":[{"comment":"There are several typographical artifacts, including '5d noble – metal' with a stray dash in the abstract and '10-6 eV' which should be set as '10^{−6} eV'.","section":"Abstract and text formatting"},{"comment":"The surface energies used in the dashed curves and in Table I are not listed; please state whether they were computed here with PBEsol and the same cutoff/k-point settings, and ideally tabulate E_S−UN and ΔE_S for each surface.","section":"Eqs. (4)–(6), dashed curves in Figs. 2 and 3"},{"comment":"The sentence stating that all studied (110) nanosheets transform to (001) surfaces when sufficiently thin is an additional structural-prediction claim that is not supported by any energy comparison between (110) and (001) orientations in the paper; either provide the supporting evidence or delete the sentence.","section":"Band/gap around Fig. 2"},{"comment":"For Ir(110), l_c = 55.2 layers while the DFT calculations only extend to 40 layers, so the entry '>40' is a weak consistency check; this should be stated as a lower bound rather than a validation.","section":"Table I"},{"comment":"The dashed curves are continuous functions of l, but ΔE is only meaningful for integer numbers of layers; a note explaining that the curves are guides for the discrete data points would improve clarity.","section":"Fig. 2 caption"}],"recommendation":"major_revision","confidential_remarks":"The central algebraic result is correct under its stated assumptions, and the DFT trends are broadly consistent. The main concerns are the unstated thermodynamic ensemble and the mismatch between the ensemble used in the DFT validation and the ensemble that governs a physical nanosheet; these are fixable with additional analysis or a recalculation under a fixed-N constraint. The title and abstract advertise a 'universal' rule, which is considerably stronger than what the evidence supports; I would encourage the editor to require that the generality claims be tempered or explicitly delimited."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper gives a simple, testable rule: thinning always favors reconstructions that raise surface atomic density, with critical thickness Eq. (6) from bulk surface energies only. That is genuinely new and useful. The derivation is elementary, but the formula organizes two opposite thickness trends, and the DFT numbers support it. I think the reader's main worry about per-atom normalization is real but not fatal: the stress-test note is right that in the same linear surface-energy model, a fixed-N comparison yields exactly the same l_c. So Eq. (6) is not an arithmetic trick. The load-bearing soft spot is different: the DFT validation never actually tests the physical fixed-N ensemble that governs an isolated nanosheet. Each supercell is fully optimized, so the unreconstructed and reconstructed cells have different atom counts and different equilibrium in-plane areas. That means Table I validates the per-atom model, not the fixed-N prediction. The deviations for Pt/Au and for (001) sheets also show surface stress and quantum oscillations are not negligible, so the linearity assumption is imperfect. And the 'universal' claim is ensemble-dependent: in a grand-canonical picture (sheet in contact with an atom reservoir) there is no thickness dependence at all. The paper never states its assumed boundary conditions. A fixed-N DFT slab calculation is the missing test.\n\nWhat is good: the algebra is clean, the DFT trends are internally consistent, the Pd magnetism check is a nice extra, and the predicted l_c values in Table I agree with direct DFT for all six cases within about one layer (Ir is a lower bound). The paper is honest about not considering the complex Pt/Au(001) reconstructions and about open competition with other superstructures. Citations look fair; the per-atom convention is cited to refs [15,16].\n\nWho this is for: people working on 2D metal stability, nanosheet synthesis, and adsorbate-induced reconstruction. It deserves a serious referee. The main revision should add a fixed-N DFT test, or at least an explicit statement of the ensemble assumption, and ideally deposit raw energies.","headline":"A clean, useful criterion for surface reconstruction in 2D metals, but the DFT validation does not test the fixed-N ensemble that governs a real nanosheet, so the 'universal' claim is weaker than stated.","tokens_in":7769,"tokens_out":1850,"would_cite":true,"duration_ms":17169,"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 single thickness criterion, built from bulk surface energies and density ratios, predicts when two-dimensional metal sheets reconstruct: thinning always favors the structure with higher surface atomic density.","keywords":["surface reconstruction","two-dimensional metals","noble metal nanosheets","critical thickness","surface atomic density","density functional theory","(1×2) missing-row reconstruction","quasi-hexagonal (5×1) reconstruction"],"falsifier":"Grow a freestanding Au(110) sheet about 10–12 atomic layers thick and inspect its surface by STM or LEED at low temperature. The criterion predicts the (1×2) missing-row reconstruction should be absent, so observing that pattern at this thickness would falsify the claimed thickness reversal.","tokens_in":1907,"feed_emoji":"⚛️","tokens_out":2023,"duration_ms":63677,"temperature":0.7,"pith_summary":"This paper tries to establish a single rule for when the surface of a two-dimensional metal sheet reconstructs: thinning always favors the structural change that raises the density of atoms in the surface layer. Starting from bulk surface energies and the atom-count ratio of the reconstruction, the authors derive a critical thickness at which the reconstruction preference flips, and validate it with density-functional calculations on noble-metal (110) and (001) sheets. If the rule holds, the surface structure of an ultrathin metal can be predicted from bulk data alone, without computing the sheet. It also unifies two seemingly opposite observations: thinning suppresses the (1×2) reconstruction on 5d metal (110) sheets and promotes the quasi-hexagonal reconstruction on 4d metal (001) sheets.","feed_headline":"Thinning rewrites the rules for metal surface reconstruction","feed_subtitle":"A single formula from bulk energies predicts whether ultra-thin metal sheets reconstruct.","key_machinery":"The central object is the energy-per-atom difference $\\Delta E$ between reconstructed and unreconstructed sheets, Eq. (3), which the paper reduces to Eq. (4) and solves for the critical thickness $l_c = 2(n/m - 1)E_{S-UN}/\\Delta E_S$. Here $m$ is the number of atoms in a bulk (1×1) layer, $n$ is the number of atoms in a reconstructed surface layer over the same lateral area, $E_{S-UN}$ is the unreconstructed surface energy per area, and $\\Delta E_S = E_{S-RE} - E_{S-UN}$ is the surface-energy change upon reconstruction. The sign of $(n/m - 1)\\Delta E_S$ determines whether thinning can flip the preference: when $l_c>0$ the crossover occurs at $l=l_c$, and when $l_c\\le 0$ the bulk tendency persists at all thicknesses. This turns reconstruction prediction into arithmetic on bulk quantities.","core_discovery":"The paper claims that in an ultrathin fcc metal sheet the thermodynamic preference for surface reconstruction is controlled by whether the reconstruction changes the number of atoms in the surface layer. Writing the total energy of an $l$-layer slab as bulk energy plus surface terms, and comparing average energy per atom, gives a critical thickness $l_c = 2(n/m - 1)E_{S-UN}/\\Delta E_S$. If $l_c>0$, the reconstruction tendency reverses at that thickness: for 5d noble-metal (110) sheets ($n/m=1/2$, $\\Delta E_S<0$) thinning lifts the (1×2) missing-row reconstruction, while for 4d noble-metal (001) sheets ($n/m>1$, $\\Delta E_S>0$) thinning induces the quasi-hexagonal (5×1) reconstruction. In both cases thinning favors the structure with higher surface atomic density. Direct density-functional total-energy calculations reproduce the predicted critical thicknesses within a few layers, with surface stress and quantum-size oscillations giving only minor corrections.","pith_inferences":["The same density criterion should apply to non-noble fcc metals and to alloys that share the fcc layer geometry; the paper argues for universality but tests only noble metals, so this extension is an inference rather than a demonstrated result.","Because surface stress already lowers the actual crossover for soft metals like Au and Pt, applying tensile or compressive strain could systematically tune the critical thickness and potentially switch reconstruction on or off at a chosen sheet thickness.","Adsorbate-induced reconstructions, such as the CO-induced Pd nanosheet reconstruction, could be reinterpreted as the adsorbate flipping the effective $\\Delta E_S$, turning a bulk-unreconstructed surface into one that reconstructs once the sheet is thin enough.","For 4d (001) sheets, the paper leaves open which superstructure wins among (5×1), c(28×48), and c(26.6×118); calculations at the predicted 3–5 layer crossover would settle which reconstructed pattern actually forms."],"forward_implications":["For 5d noble-metal (110) sheets, the (1×2) missing-row reconstruction should be suppressed below the critical thickness: roughly 11 layers for Pt, 15 for Au, and more than 40 for Ir.","For 4d noble-metal (001) sheets, the quasi-hexagonal (5×1) reconstruction should appear below roughly 6 layers for Rh, 5 for Pd, and 4 for Ag, even though the bulk surfaces do not reconstruct.","The critical thickness can be computed from bulk surface energies and the density ratio $n/m$ alone, so no nanosheet calculation is required to predict whether an ultrathin metal reconstructs.","When the surface atomic density does not change upon reconstruction ($n=m$, as reported for W(001) and Mo(001)), no thickness-driven change in reconstruction tendency is expected.","Surface stress and quantum-size oscillations shift the actual crossover by at most a few atomic layers in the metals studied, so the bulk-derived formula remains a good predictor."],"supporting_citations":[{"why":"Document the bulk reconstruction trends of 5d versus 4d noble metal surfaces, the empirical baseline that Eq. (6) must reproduce.","marker":"[1,3]"},{"why":"Supply the average-energy-per-atom convention for comparing structures with different atom counts over the same lateral area, which underlies Eq. (3).","marker":"[15,16]"},{"why":"Provide the surface atomic density changes of the (110)-(1×2) and (001) reconstructions, fixing the $n/m$ ratios used in the criterion.","marker":"[17-19]"},{"why":"Supplies the plane-wave density-functional implementation used for all total-energy calculations.","marker":"[23]"},{"why":"Supplies the exchange-correlation functional used in the calculations, chosen for reproducing transition-metal surface properties.","marker":"[24]"},{"why":"Provides surface stress values for reconstructed surfaces, invoked to explain the small deviations between derived and directly calculated energy differences.","marker":"[26]"},{"why":"Reports ferromagnetism in unreconstructed Pd(001) sheets at fixed bulk lattice constants; the paper checks and rules out magnetism as a factor in the reconstruction crossover.","marker":"[29]"},{"why":"Reports CO-induced reconstruction on Pd nanosheets, the adsorbate case that the clean-surface criterion is meant to inform.","marker":"[14]"}],"fun_headline_variants":["Density rules surface reconstruction in ultra-thin metals","Thinning selects denser atom packing on metal surfaces","A single thickness formula predicts metal surface structure","For thin metals, surface density dominates stress","Thickness flips metal surface reconstruction by density"],"cache_read_input_tokens":9856,"weakest_assumption_plain":"The argument assumes that stability is judged by average energy per atom over a fixed lateral area, so structures with different total atom counts are compared on an equal-area footing; if the correct comparison instead fixes the number of atoms, the sign structure of the critical-thickness formula changes.","fun_headline_variants_meta":{"raw":{"variants":["Density rules surface reconstruction in ultra-thin metals","Thinning selects denser atom packing on metal surfaces","A single thickness formula predicts metal surface structure","For thin metals, surface density dominates stress","Thickness flips metal surface reconstruction by density"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000366,"raw_usage":{"total_tokens":1904,"prompt_tokens":820,"completion_tokens":1084,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":436,"completion_tokens_details":{"reasoning_tokens":1013}},"tokens_in":436,"tokens_out":1084,"duration_ms":10239,"temperature":1.0,"reasoning_tokens":1013,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T14:21:46.419386+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Grow a freestanding Au(110) sheet about 10–12 atomic layers thick and inspect its surface by STM or LEED at low temperature. The criterion predicts the (1×2) missing-row reconstruction should be absent, so observing that pattern at this thickness would falsify the claimed thickness reversal.","supporting_citations":[{"cited_title":"Olivier, G","cited_arxiv_id":null,"evidence_quote":"Provides surface stress values for reconstructed surfaces, invoked to explain the small deviations between derived and directly calculated energy differences."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reports ferromagnetism in unreconstructed Pd(001) sheets at fixed bulk lattice constants; the paper checks and rules out magnetism as a factor in the reconstruction crossover."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reports CO-induced reconstruction on Pd nanosheets, the adsorbate case that the clean-surface criterion is meant to inform."}],"review_version":1}