{"id":"6e5e77b3-21b4-4d16-aa44-ff2f69145cf2","arxiv_id":"2608.00138","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"The interlayer bond in hexagonal diamond is about 24 mÅ longer than the intralayer bonds, and Raman spectra confirm this asymmetry against two contradictory refinements.","lead":"First-principles calculations and Raman analysis show that hexagonal diamond's interlayer bond is about 24 thousandths of an angstrom longer than its intralayer bonds. The result settles a disputed structural question and challenges two recent refinements that claimed the opposite.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Raman-derived 24±3 mÅ interval depends on an unverifiable two-component fit of Lai's envelope; a misassignment or >3 cm−1 splitting error would shift the headline, though not the sign.","rationale":"The paper's central claim — intrinsic hexagonal diamond has OB−OA ≈ +24 mÅ — is supported by two DFT functionals, by the eclipsed-bond mechanism, by the older Yoshiasa refinement, and by the pattern-level Raman comparisons. The weakest point is the specific quantitative inversion of the phase-pure Raman spectrum into 24±3 mÅ. That inversion uses a measured splitting of 28 cm−1 between two fitted components of one asymmetric envelope; the sensitivity is 0.41 mÅ per cm−1 of splitting, so the result is directly controlled by the fit that produced 1,310 and 1,338 cm−1. The manuscript states in S9 that the underlying traces are not among the deposited source data, so the fit cannot be independently reproduced. The uncertainty budget also relies on an assumed σ≈3 cm−1 with no formal probability model for the '95%' label. These are not fatal to the sign or approximate magnitude, because the DFT and diffraction evidence stand on their own, but they mean the Raman-derived interval should be understood as model-dependent and not independently verifiable from the archived materials. The reader's weakest assumption identifies the same decomposition and assignment dependence, so I agree with the conditional verdict and recommend no change.","tokens_in":26484,"tokens_out":10414,"duration_ms":129164,"concrete_test":"Obtain the raw Raman spectrum behind Lai et al.'s Supplementary Fig. 4 (or digitize the published trace) and independently re-fit the 1,250–1,450 cm−1 envelope with: (i) two Lorentzians (A1g/E1g), (ii) three Lorentzians adding E2g, (iii) A1g plus a cubic-diamond T2g component, and (iv) inverted A1g/E1g ordering, with a common background model. Bootstrap the component positions. If the 1,310–1,338 cm−1 splitting changes by more than ~5 cm−1 under these models, or a non-preferred model is preferred by AIC/BIC, the 24±3 mÅ interval is not robust and should be re-labeled or widened. A complementary check: polarization-resolved UV Raman on a phase-pure sample would directly establish E1g visibility and settle the assignment.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The quantitative Raman claim rests on the splitting between two fitted components (1,310 and 1,338 cm−1) in a single asymmetric envelope from Lai et al. Passing this splitting through the DFT sensitivity map (Eq. S1) gives OB−OA = 24±3 mÅ, with ∂(OB−OA)/∂(splitting) = 0.41 mÅ/cm−1. The interval therefore inherits the component-fit uncertainty of the splitting, assumed to be σ ≈ 3 cm−1 (SI S1, S9). The raw traces are not in the deposited source data, so this assumption cannot be checked independently. If the 1,338 cm−1 component is not E1g — for example if it is a cubic-diamond T2g line (1,332 cm−1), a third unresolved component, or the E2g mode — the inversion changes qualitatively; the paper's assignment matrix argues strongly against these, but the argument is conditional on the same E1g visibility it uses to admit the component. The sign and rough magnitude of the asymmetry are independently supported by DFT (24 mÅ in two functionals) and the Yoshiasa refinement (+60±45 mÅ), so the concern is with the precision and 'measurement' status of 24±3, not with the central structural conclusion.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper uses first-principles lattice dynamics and Raman-spectrum inversion to argue that the intrinsic structure of hexagonal diamond has a long interlayer bond, OB − OA ≈ +24 mÅ, in agreement with an overlooked 2003 Yoshiasa refinement but in conflict with two recent Rietveld refinements (Yang et al., Lai et al.) that give −55 and +183 mÅ, respectively. The authors compute exact phonons for the relaxed, refined, and several constrained structures; benchmark against cubic-diamond T2g and graphite G; show that the A1g mode acts as an interlayer-bond gauge with sensitivity ≈ −2150 cm−1 Å−1; invert the two-component Raman envelope of Lai's phase-pure sample to obtain OB − OA = 24 ± 3 mÅ (95%); and reanalyze Lai's diffraction pattern to argue that it cannot discriminate the DFT-relaxed coordinate. They also attribute the 1529 cm−1 band of Yang's sample to strained/disordered sp2 carbon. The central claim is that vibrational spectroscopy, together with diffraction, selects a small positive bond asymmetry and invalidates both recent refinements.","tokens_in":26743,"tokens_out":7090,"duration_ms":81336,"significance":"If the central conclusion stands, it resolves a high-profile structural controversy in a material of current experimental interest, and it provides a concrete, falsifiable prediction: the A1g–E1g splitting is a linear gauge of the interlayer bond. The paper's computational strengths are substantial: exact phonons at every candidate structure, two independent functionals, benchmarks against known carbon modes, an assignment matrix that tests all six component orderings, a measured-cell consistency check, a diffraction sensitivity analysis on source data, and a fully archived 2,611-file data package with a manifest. The sign and rough magnitude of the asymmetry are supported by multiple lines of evidence (DFT, Yoshiasa diffraction, Goryainov impact-diamond Raman), so the main structural conclusion is defensible. However, the quantitative '24 ± 3 mÅ' headline is conditional on an externally fitted two-component decomposition of a Raman envelope whose raw traces are not deposited, and the inversion is calibrated by the same DFT physics it confirms; these caveats need to be foregrounded.","major_comments":[{"comment":"The headline 24 ± 3 mÅ (Eq. S4) is obtained by passing the two fitted components (1,310 and 1,338 cm−1) of Lai et al.'s envelope through the linearized DFT map (Eq. S1). As S9 admits, the underlying Raman traces of Lai et al. are not in any deposited source data, so the two-component decomposition and its covariance cannot be independently checked. The ±3 mÅ budget uses an assumed σ ≈ 3 cm−1 for the splitting; if the envelope contains a third resolved component or if the line-shape model differs, the splitting and hence the inferred asymmetry would shift. The assignment matrix (Table S1) addresses which modes are assigned, but not how many components exist or whether a two-component fit is statistically justified. Since the abstract presents 24 ± 3 mÅ as a primary result, the paper should either obtain and archive the raw traces (or a re-fit of them) or clearly label this value as condit","section":"SI S1 / SI S9"},{"comment":"The Raman inversion is not an independent measurement of the bond asymmetry: the sensitivity matrix J and the reference frequencies ω0 are computed by DFT at the DFT-relaxed asymmetric structure, and the measured splitting (28 cm−1) coincides exactly with the calculated E1g − A1g splitting (28 cm−1). The conclusion that the inversion 'gives' 24 ± 3 mÅ therefore largely returns the DFT input through a one-parameter map. The paper's insensitivity to the offset δ is correct, but the J matrix and ω0 carry the DFT physics. The claim should be phrased as a DFT-calibrated consistency check, not as an independent spectroscopic determination. This distinction matters for how a reader weighs the 24 ± 3 mÅ interval against the two recent refinements.","section":"SI S1 Eqs. (1)–(3)"},{"comment":"The statement that the two recent refinements are 'inconsistent with ... the earlier refinement, at 2.3σ and 3.0σ' compares z-coordinates using the Yoshiasa σ = 0.008. That σ comes from a two-phase powder refinement with a stated u uncertainty of ±0.008, and the '0.7σ' agreement between Yoshiasa and DFT may be a coincidence of the large uncertainty. This is not an error, but the sigma language suggests a quantitative verdict that is only as strong as the 2003 refinement's error model. The Raman-based rejection of the two recent structures rests on pattern-level discrepancies of 78–350 cm−1, which are far more robust; the sigma phrasing should be softened or its provenance clarified.","section":"Main text, Discussion (ii)"}],"minor_comments":[{"comment":"The abstract uses '238 mÅ' without defining the unit; define mÅ (10−3 Å) at first use, since many readers will misread it as an SI prefix on Å.","section":"Abstract"},{"comment":"The shaded measured bands in Fig. 4 are described as visual guides with fixed width, which is fine, but the figure would benefit from explicitly marking the Lai measurement window at 1,100 cm−1 so that 'below window' entries (e.g., Lai-Rietveld A1g at 959 cm−1) are visually clear.","section":"Fig. 4 / Table 1"},{"comment":"The intensity ordering argument uses static, non-resonant PBE activities, while both experiments used UV excitation; the text correctly caveats this, but the caveat should be repeated near the sentence excluding the E1g/A1g swap structure, since that exclusion partly rests on equalized activities (98:100:103).","section":"S5"},{"comment":"The diffraction sensitivity analysis uses a one-parameter March–Dollase texture proxy, not the original spherical-harmonic texture model; this is acknowledged, but the conclusion that the Lai pattern 'does not discriminate' should be stated as a sensitivity result for this proxy model, not as a full reproduction of Lai's refinement.","section":"S6"}],"recommendation":"major_revision","confidential_remarks":"The paper is scientifically rich and likely correct in its main qualitative conclusion, and the computational deposit is exemplary. The main risk to the journal's readers is that the abstract and introduction overstate the independence and precision of the Raman-derived 24 ± 3 mÅ: the interval is conditional on Lai et al.'s two-component fit, whose raw data are not available, and on a DFT-calibrated sensitivity map that already contains the +24 mÅ physics. I would encourage revision that (1) explicitly disclaims the unverifiable external fit in the abstract, (2) rephrases the Raman result as a DFT-calibrated consistency check rather than an independent measurement, and (3) demotes the sigma-statements about the earlier refinement to a cautious footnote. If these caveats are made prominent, the paper's central claim is defensible."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First thing to know: this is the strongest entry yet in the HD bond-asymmetry fight. Zhu computes the phonons of all candidate structures, builds a benchmarked A1g sensitivity map (about -2,100 cm^-1/A) and uses it to invert published Raman spectra to OB-OA = +24±3 mÅ for the phase-pure sample, with +33±8 and +60±45 from two other determinations. The paper also shows that both recent Rietveld refinements (Yang 2025, Lai 2026) fail to reproduce the band patterns measured on their own samples, and that the only homogeneous 2H structure matching Yang's three bands needs ~77 GPa axial stress and lattice constants excluded by their own XRD. If right, the sign controversy is closed: the interlayer bond is longer, as Yoshiasa found in 2003.\n\nWhat's new and good: the sensitivity-map idea is genuinely new, and Zhu is careful with it. The mode map is benchmarked against diamond T2g and graphite G, the assignment matrix covers all six orderings, the measured-cell consistency check works, and the diffraction sensitivity analysis of Lai's pattern shows the z-coordinate is basically unconstrained by their data. The paper also credits prior equilibrium phonon calculations, cites Yoshiasa (which the recent papers missed), and is transparent about what it cannot do: it cannot identify the 1529 cm^-1 carrier, and it says so.\n\nThe soft spots are not fatal. The '95% interval' label is not earned: the budget is a quadrature of effective uncertainties with no formal probability model, which the paper itself admits in SI S1. Calling that a 95% interval is misleading, and a referee should ask for a re-label. The 24±3 also rests on the two-component fit of Lai's envelope, and the raw traces aren't in the deposited data; the paper flags this too (SI S9) and the assignment matrix makes a strong case, but independent verification of the splitting is still missing. The E1g-visibility point has a whiff of circularity, though the third-sample (Goryainov) spectrum and the intensity ordering give independent support.\n\nBottom line: the sign and tens-of-mÅ magnitude are solid; the precise number 24±3 should be treated as a model-based estimate, not a measurement with a genuine 95% CI. This paper deserves a serious referee. It is reproducible (data and scripts on Zenodo), the claims are falsifiable (polarization-resolved Raman, raw traces), and the conclusion matters for anyone working on carbon polytypes or high-pressure synthesis. I'd bring it to a reading group and would cite it.","headline":"Zhu's Raman-inversion paper is the strongest entry yet in the HD bond-asymmetry fight: sign is positive, magnitude ~24 mÅ, both recent refinements fail the spectral test, and the only real flaws are the '95%' label and missing raw traces.","tokens_in":27277,"tokens_out":3982,"would_cite":true,"duration_ms":45231,"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":"Hexagonal diamond's interlayer bond is longer than its intralayer bonds by 24 mÅ, and Raman spectra support it.","keywords":["hexagonal diamond","lonsdaleite","bond asymmetry","interlayer bond","Raman spectroscopy","A1g mode","density functional theory","eclipsed bond"],"falsifier":"Polarization-resolved Raman on an oriented phase-pure hexagonal-diamond crystal: if the 1,338 cm−1 component shows E1g symmetry and a third band appears near 1,221 cm−1 (E2g), the A1g/E1g assignment and the 24±3 mÅ value are supported; if the 1,338 cm−1 component is A1g, or if a third resolved component emerges in the 1,310/1,338 envelope, the inverted asymmetry shifts and the central claim fails.","tokens_in":26296,"feed_emoji":"💎","tokens_out":4654,"duration_ms":52702,"temperature":0.7,"pith_summary":"Hexagonal diamond (lonsdaleite) has two inequivalent carbon–carbon bonds: three intralayer bonds within buckled honeycomb layers and one interlayer bond along the hex axis. The paper argues that the interlayer bond is intrinsically longer than the intralayer bonds by about 24 mÅ, and that this asymmetry—not a near-zero value and not the opposite sign—is what the measured Raman spectra show. Two recent diffraction refinements disagree with each other and with a 2003 refinement; the paper uses first-principles lattice dynamics to show that the bright A1g Raman mode acts as a direct gauge of the interlayer bond, turning a measured Raman splitting into a bond-length difference. If correct, this settles a contested structural question and makes Raman spectroscopy a quantitative probe of bond asymmetry in diamond polytypes.","feed_headline":"Raman pins hexagonal diamond's interlayer bond at +24 mÅ","feed_subtitle":"A contested structure: the A1g mode acts as a bond ruler, resolving a 238 mÅ disagreement between refinements.","key_machinery":"The central object is the zone-center A1g mode, which is essentially the interlayer-bond stretch. Its frequency shifts by about −2,100 cm−1 per ångström of interlayer-bond length, while the two E modes track the intralayer bonds; this sensitivity map converts a measured Raman splitting directly into a bond-length difference. The underlying physical mechanism is the eclipsed conformation of the interlayer bond, which weakens and lengthens it, and the effect is local and transferable across the diamond polytypes examined.","core_discovery":"Relaxed hexagonal diamond has a longer interlayer bond: OB − OA = +24 mÅ in both density functionals tested, driven by the eclipsed conformation of the interlayer bond. The same local asymmetry appears in hexagonally stacked layers of other polytypes and disappears at cubic stacking faults. Inverting the phase-pure measured Raman envelope with a DFT-calibrated sensitivity map gives OB − OA = 24±3 mÅ (95% interval), with two independent determinations on other samples giving +33±8 and +60±45 mÅ. The two recent Rietveld structures fail the Raman test on their own samples, and no mechanically viable, diffraction-consistent homogeneous 2H structure reproduces the twinned sample's three Raman ban","pith_inferences":["If the calibrated sensitivity map transfers to other materials, the same strategy—using a bright, bond-selective zone-center mode as a 'bond ruler'—could be applied to contested bond-length asymmetries in other tetrahedral polytypes, such as wurtzite boron nitride or silicon carbide polytypes.","A direct testable extension: measuring the A1g frequency of hexagonal diamond under controlled uniaxial stress should reproduce the predicted ∼−2,100 cm−1/Å shift and the sign inversion near ±1.8% strain.","The diffraction sensitivity re-analysis implies that preferred orientation may make the internal coordinate of the phase-pure sample underdetermined; re-refining with a more robust texture model could resolve the contradiction from the diffraction side without invoking Raman.","If the one-phonon density of states indeed terminates at 1,329 cm−1, any future observation of a strong 1,529 cm−1 band in a clean, phase-pure sample would force a reassignment, so the sp2 explanation is falsifiable by cleaner spectroscopy."],"forward_implications":["The two recent diffraction refinements of bulk hexagonal diamond are inconsistent with each other, with the Raman spectra measured on their own samples, and with the 2003 refinement; their internal coordinates should be treated with caution.","Raman spectroscopy, through the A1g mode, can measure the interlayer bond asymmetry to about ±3 mÅ, making it a sharper structural probe on these samples than the available diffraction.","The 1,529 cm−1 feature cannot be a first-order mode of ideal 2H diamond; the leading explanation is minority sp2 carbon under roughly 1% biaxial strain or with significant disorder.","The bond asymmetry scales with hexagonality and is erased at cubic stacking faults, so Raman frequencies report the average stacking order within the optical probe volume.","The 2003 diffraction refinement, though imprecise in its internal coordinate, agrees with theory and with Raman on the sign, axial ratio, bond budget, and bond-strength ordering."],"fun_headline_variants":["Raman resolves hexagonal diamond's bond asymmetry: +24 mÅ","Hexagonal diamond's long interlayer bond measured: +24 mÅ","Vibrational proof: hexagonal diamond has asymmetric bonds","Bond ruler: Raman pins hexagonal diamond at +24 mÅ","Raman settles hexagonal diamond bond debate: +24 mÅ"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that the measured Raman envelope of the phase-pure sample is correctly decomposed into exactly two components at 1,310 and 1,338 cm−1 assigned to A1g and E1g, and that the DFT-computed sensitivity of the A1g frequency to interlayer bond length is accurate enough to convert the 28 cm−1 splitting into a bond-length difference of 24 mÅ.","fun_headline_variants_meta":{"raw":{"variants":["Raman resolves hexagonal diamond's bond asymmetry: +24 mÅ","Hexagonal diamond's long interlayer bond measured: +24 mÅ","Vibrational proof: hexagonal diamond has asymmetric bonds","Bond ruler: Raman pins hexagonal diamond at +24 mÅ","Raman settles hexagonal diamond bond debate: +24 mÅ"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000203,"raw_usage":{"total_tokens":1259,"prompt_tokens":817,"completion_tokens":442,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":561,"completion_tokens_details":{"reasoning_tokens":354}},"tokens_in":561,"tokens_out":442,"duration_ms":5410,"temperature":1.0,"reasoning_tokens":354,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-04T01:11:56.366201+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Polarization-resolved Raman on an oriented phase-pure hexagonal-diamond crystal: if the 1,338 cm−1 component shows E1g symmetry and a third band appears near 1,221 cm−1 (E2g), the A1g/E1g assignment and the 24±3 mÅ value are supported; if the 1,338 cm−1 component is A1g, or if a third resolved component emerges in the 1,310/1,338 envelope, the inverted asymmetry shifts and the central claim fails.","supporting_citations":[],"review_version":1}