{"id":"6ca9a537-6573-47ff-8205-fa65155dab84","arxiv_id":"1908.08102","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":7,"one_line_summary":"The apparent first diffraction peak of liquid water is a doublet, a Lorentzian from tetrahedral structures and a Gaussian from disordered structures, supporting a two-state picture of water.","lead":"This paper argues that the first peak in the scattering pattern of liquid water is actually two overlapping peaks, one from orderly tetrahedral clusters and one from disordered regions. This supports the idea that water is a dynamic mixture of two local structures, and offers a way to measure the amount of ordering from experiment.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The experimental proof of the doublet is not yet decisive: the fitted model imposes both the two-peak decomposition and fT1 ∝ s, and the reported residual comparison does not rule out a single asymmetric peak.","rationale":"The reader's weakest assumption — that the Lorentzian+Gaussian decomposition and fT1 ∝ s are assumed rather than derived — is also the most load-bearing weakness I find. I agree that the central claim is plausible and well supported on the simulation side, where the ζ-resolved Debye scattering function shows kT1 and kD1 peaks in different structural domains and where ζ and Nfs distributions are bimodal in three water models. That is real, non-circular evidence. The experimental proof, however, is weaker: the fit constrains fT1 to follow the two-state fraction, and for real water the fraction itself is obtained from the same scattering data via gOO(r), so the agreement between fT1 and s is not an independent confirmation. Moreover, the Fig. S12 comparison varies both the number of peaks and the global parameterization, so its residual difference cannot be attributed solely to the presence of a doublet. My proposed test would settle the matter by comparing equal-flexibility models and by checking whether an unconstrained fT1 actually tracks s. Until that is done, the reader's CONDITIONAL verdict is appropriate; I see no reason to move it to ACCEPT or REJECT.","tokens_in":17780,"tokens_out":6435,"duration_ms":70990,"concrete_test":"Re-fit the experimental ambient-pressure S_OO(k) datasets (Refs. 29–31) in two stages. First, at each temperature, fit the first-peak region with candidate models of equal parameter count: one Gaussian, one Lorentzian, one asymmetric pseudo-Voigt, and a Lorentzian+Gaussian doublet with kT1 and kD1 free but fT1 and fD1 unconstrained; compare by AIC/BIC and by bootstrap over the reported experimental noise. Second, using the doublet model, leave fT1 and fD1 free at each temperature and test whether the fitted fT1 tracks the independently determined s(T) from ζ and Nfs for the model waters, and s(T) = 1 − gOO(r_HB) for real water, within propagated errors. Include the three points excluded in Fig. S13 in a robustness variant. If a single asymmetric peak is preferred, or the free fT1 does not track s, the two-motif conclusion from scattering alone is not supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing concern is that the experimental evidence for two motifs is generated by a fitting model that already contains the two-motif answer. Equation S11 decomposes the first peak into a Lorentzian at kT1 and a Gaussian at kD1, and Eqs. S16 and S18 then impose fT1 = a·s and fD1 = b·(1−s), with s from the two-state equation S17. For real water, the parameters entering s are calibrated from gOO(r) via s = 1 − gOO(r_HB) (Methods), i.e., from the same scattering data that are then fitted; a, b, ΔE, and Δσ are free parameters. The comparison in Fig. S12 does not isolate the doublet hypothesis: Scheme I uses 8–9 free parameters per temperature, while Scheme II uses 25 parameters with globally smoothed temperature dependence (Eqs. S12–S15), and three deviant experimental points are set aside post hoc (Fig. S13). No confidence intervals are reported for the fitted kT1, kD1, ΓT1, or fT1 values, which is especially important because overlapping peaks are strongly covariant. The simulation-side ζ-resolved Debye structure factor (Fig. 2d,e) is genuine independent support for two motifs in the models, but it does not establish that the experimental line shape is a doublet or that the Lorentzian weight equals a·s. Therefore the claim that the scattering data 'unambiguously prove' coexistence is conditional on a model-comparison and a proportionality test that are not reported.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper argues that the apparent first diffraction peak in the O-O partial structure factor of water is actually a doublet: a Lorentzian peak at kT1 ≈ 3/4 (in units of k rOO / 2π), attributed to locally favored tetrahedral structures (LFTS), and a Gaussian peak at kD1 ≈ 1, attributed to disordered normal-liquid structures (DNLS). The authors support this with simulations of TIP4P/2005, TIP5P, and ST2 water, using a ζ-resolved Debye scattering function that shows distinct peaks in different ζ domains, and with fits to experimental and simulated S(k) using a two-Lorentzian/two-Gaussian model. From the fits they extract the LFTS fraction s and a coherence length, claiming that this 'unambiguously proves the coexistence of two local structural motifs' in liquid water.","tokens_in":18200,"tokens_out":3361,"duration_ms":31469,"significance":"If the central claim is correct, the paper provides an experimentally accessible order parameter for the two-state model of water and a way to measure the range of tetrahedral ordering, which would be a substantial contribution to the long-standing mixture-versus-continuum debate. The simulation-side evidence—especially the ζ-resolved Debye scattering function in Fig. 2d–e showing two distinct peaks at kT1 and kD1 in different ζ domains—is a genuine and valuable independent demonstration that the two structural motifs produce distinct scattering signatures in the simulated models. The paper also connects water to the prior silica work (Ref. 26) in a physically appealing way. However, the experimental proof rests on a fitting model whose line-shape decomposition and proportionality relations are assumed rather than derived, and the model comparison does not rule out simpler alternatives.","major_comments":[{"comment":"The decomposition of the apparent first diffraction peak into a Lorentzian (L1) at kT1 and a Gaussian (G1) at kD1 is assumed in the fitting model, not derived from the data. The comparison in Fig. S12 between Scheme I (one Gaussian for the first peak) and Scheme II (Lorentzian plus Gaussian) does not rule out other line shapes, because the two schemes differ in parameter count (8–9 per temperature versus 25 globally) and no information criterion, cross-validation, or confidence intervals are reported. As written, the experimental evidence for the doublet is conditional on the very model it is meant to prove.","section":"Methods, Eq. S11"},{"comment":"The proportionality fT1 = a·s and fD1 = b·(1−s) is imported from liquid silica (Ref. 26) and the constants a and b are fitted to the water data; s itself is determined from the same experimental scattering data via s = 1 − gOO(r_HB) (Methods, 'Fitting formula for the structure factor'). This makes the agreement between the fitted Lorentzian intensity and the two-state fraction (Fig. 4a) partly circular: the model constrains the fit with the quantity it then claims to extract. The paper should provide an independent determination of s (e.g., from simulations of the same models) and test the proportionality with a and b reported with uncertainties.","section":"Methods, Eqs. S16–S18"},{"comment":"No statistical uncertainties are reported for the fitted peak positions kT1, kD1, widths, or intensities. For heavily overlapped peaks these parameters are strongly covariant, so the claimed separation at kT1 ≈ 3/4 and kD1 ≈ 1 cannot be assessed without confidence intervals or a bootstrap analysis. At minimum, the authors should report error bars from the simultaneous fit, particularly for the experimental water data.","section":"Fig. 3 and Fig. S12"},{"comment":"Three experimental temperatures (234.8, 264.0, and 268.1 K) are excluded post hoc as outliers based on their peak/trough heights, and the master-curve collapse in Fig. S12 is shown after this exclusion. Because Scheme II is a global fit, excluding points can change the fitted parameters at all temperatures; the authors should show the fit with and without these points and justify the exclusion a priori rather than after seeing the residuals.","section":"Fig. S13"}],"minor_comments":[{"comment":"The word 'discontinuosly' should be 'discontinuously', and in the Fig. 4 caption 'propotional' should be 'proportional'.","section":"Abstract"},{"comment":"The phrase 'only board isotropic amorphous halos' should read 'only broad isotropic amorphous halos'.","section":"Main text, near Eq. (1)"},{"comment":"There are several typographical errors: 'Lorentizan' and 'Guassian' appear in the Fig. 3 caption and elsewhere, and 'cooresponds' appears in the Fig. 2 caption. These should be corrected.","section":"Throughout"},{"comment":"The threshold value ζc (≃0.5 Å) is mentioned only in the caption; the procedure for choosing this threshold and its effect on s should be described in the Methods, since s depends on it.","section":"Fig. S5 caption"},{"comment":"The statement that 'only 25 free fitting parameters' are necessary is not fully enumerated; the constraints (e.g., setting k̃x2 = 0, fixing kT3, fixing the ratios of k̃T11/k̃D11) should be tabulated so the reader can reproduce the parameter count.","section":"Methods, Eqs. S12–S15"},{"comment":"The filled and open symbols in panels a and b are said to correspond to Refs. [29] and [30], respectively, but the figure legend does not state this explicitly; please add a legend or note in the caption.","section":"Fig. S13"}],"recommendation":"major_revision","confidential_remarks":"The paper builds heavily on the authors' prior work on silica (Refs. 12 and 26), and the novelty here is the extension to water and the experimental claim. The central assertion—that the scattering data 'unambiguously prove' coexistence—is not yet supported because the experimental analysis assumes the two-peak structure and the proportionality fT1 ∝ s. The simulation-side evidence is strong and would support a more measured claim. I recommend major revision with a request for a non-circular test of the proportionality, uncertainty estimates on the fitted parameters, and a model-comparison statistic that accounts for the differing number of parameters."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Core idea: the apparent first diffraction peak of water is a doublet—a tetrahedral component near k r_OO/2π ≈ 3/4 and a disordered component near ≈ 1—and the low-k peak's integrated intensity tracks the two-state fraction s. If true, that would give experimental access to tetrahedral order. The simulation half, especially the ζ-resolved Debye scattering function (Fig. 2), is the genuinely new and convincing part: it shows two distinct local environments producing peaks at different wave numbers in three water models. That is a real contribution beyond the earlier silica work.\n\nThe soft spot is the bridge to real water. The experimental fit already contains the doublet: Eq. S11 fixes a Lorentzian at k_T1 and a Gaussian at k_D1, and Eqs. S16/S18 force the intensities to be a·s and b·(1−s), with s calibrated from g_OO(r) via s=1−g_OO(r_HB)—the very same scattering data being fitted. The residual comparison (Fig. S12) pits a 9-parameter-per-temperature scheme against a 25-parameter global scheme, so the better χ² for Scheme II is not independent evidence for two peaks; it partly reflects extra flexibility. Three real-water points are discarded post hoc as noisy, and no error bars are reported for fitted positions, widths, or intensities. Given that, the word 'unambiguously' in the abstract overstates the case.\n\nThis doesn't mean the doublet is wrong. The silica analogy and the simulation decomposition make it plausible, and the collapse of Δχ² onto a master curve across all systems is appealing. But as a proof of coexistence in real water, the paper is interpretative. A serious referee would ask for a parameter-count-adjusted model comparison (AIC/BIC), error propagation on the fitted quantities, and a justification for importing Lorentzian/Gaussian line shapes from silica to water. The central argument holds only if those blanks are filled.\n\nI'd send this to peer review rather than desk-reject—the claim is important and the simulation evidence is solid—but I'd expect major revisions. For my own work, I'd cite the ζ-resolved simulation result with caution and avoid citing the doublet as established experiment.","headline":"Simulation evidence for two motifs is solid; the experimental doublet proof is model-dependent and overclaimed.","tokens_in":18678,"tokens_out":2919,"would_cite":true,"duration_ms":27647,"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":"The apparent first diffraction peak in liquid water is really two overlapping peaks, one from tetrahedral clusters and one from disordered surroundings — evidence that two local structural motifs coexist.","keywords":["liquid water","two-state model","locally favored tetrahedral structure","first sharp diffraction peak","structure factor","Debye scattering function","supercooled water","tetrahedral liquids"],"falsifier":"One could re-fit the experimental and simulated O–O structure factors below 240 K with a single asymmetric peak or with two Gaussians of free position, and compare models by an information criterion; the two-motif claim would be falsified if the Lorentzian-plus-Gaussian doublet is not clearly preferred and its $k_{T1}$ integrated intensity does not scale with the independently measured fraction of tetrahedral structures across temperature. A second check: resolve $S(k,\\zeta)$ by molecular dynamics and verify that the $k_{T1}$ peak appears only in the high-$\\zeta$ tetrahedral subpopulation; if the disordered subpopulation contributes comparably at $k_{T1}$, the assignment collapses.","tokens_in":17546,"feed_emoji":"💧","tokens_out":9219,"duration_ms":81987,"temperature":0.7,"pith_summary":"This paper argues that liquid water is not a single continuously distorted hydrogen-bond network but a dynamic mixture of two local structural motifs: tetrahedrally ordered locally favored structures and disordered normal-liquid structures. The evidence is a new reading of the oxygen–oxygen structure factor: its apparent first diffraction peak is actually two peaks, one at $k r_{OO}/2\\pi \\approx 3/4$ and one at $\\approx 1$, which grow and shrink with temperature in a way that tracks the fraction of the two motifs. The same two-peak structure appears in three widely used water models and in x-ray scattering data from real supercooled water. If right, the finding turns ordinary scattering measurements into a direct experimental probe of the degree and range of tetrahedral ordering in water, a quantity that has been accessible only through simulation. This would help settle the century-old continuum-versus-mixture debate about water's structure.","feed_headline":"Two hidden peaks reveal water's dual local structure","feed_subtitle":"Scattering data reveal tetrahedral and disordered motifs coexisting, settling a century-old debate.","key_machinery":"The load-bearing object is a decomposition of the apparent first diffraction peak into a Lorentzian at $k_{T1} \\approx 3/4$ and a Gaussian at $k_{D1} \\approx 1$, tied to the thermodynamic two-state model through the proportionalities $f_{T1} = a\\,s$ and $f_{D1} = b(1-s)$. The Lorentzian represents scattering from the density wave of characteristic wavelength set by the height of a tetrahedral locally favored structure, and its Fourier transform in real space is an exponentially decaying correlation whose decay length is read as the coherence length of tetrahedral ordering; the Gaussian represents the ordinary nearest-neighbor correlation of the disordered component. Supporting this assignment, a Debye-scattering analysis resolves the structure factor by the local structural descriptor $\\zeta$ and shows the $k_{T1}$ and $k_{D1}$ peaks arising from different $\\zeta$ subpopulations. The whole fitting scheme uses four peak functions across the first three apparent peaks and only about 25 free parameters to describe the temperature series.","core_discovery":"The paper's central claim is that the feature usually labeled the first diffraction peak in the O–O partial structure factor of water is a doublet. A low-wave-number Lorentzian peak at $k_{T1} = k r_{OO}/2\\pi \\approx 3/4$ is the first sharp diffraction peak of tetrahedral order, produced by the density wave along the height of a tetrahedral locally favored structure; a higher-wave-number Gaussian peak at $k_{D1} \\approx 1$ is the ordinary nearest-neighbor peak of disordered normal-liquid structures. Fitting real-water x-ray data and simulations of TIP4P/2005, TIP5P, and ST2 water with this doublet plus two higher peaks, the authors find that the integrated intensity of the $k_{T1}$ component is proportional to the fraction $s$ of tetrahedral structures independently obtained from a structural descriptor and from the coordination-number distribution, and that it follows the thermodynamic two-state equation. Below about 1.1 times the Schottky temperature a single-Gaussian description of the first peak fails while the doublet succeeds, and the same trend collapses across all systems. The paper therefore concludes that the two motifs coexist in liquid water and that the scattering function gives experimental access to both the degree ($s$) and the range (coherence length, from the Lorentzian width) of tetrahedral ordering.","pith_inferences":["If the decomposition holds, the same Lorentzian-plus-Gaussian analysis could be applied to other tetrahedral liquids and to neutron-scattering data with isotope substitution to test whether the two-motif picture is universal.","The proportionality between Lorentzian intensity and the two-state fraction implies that the FSDP intensity should track thermodynamic response functions such as the compressibility maximum; checking that correlation in existing data would be a cheap independent test.","The assumed Lorentzian line shape itself is a testable physical statement: it says tetrahedral order decays exponentially in space. Comparing the fitted real-space correlation with direct pair-correlation analysis from simulations would distinguish it from other decay laws.","If the two-peak decomposition remains stable under a Bayesian model comparison with asymmetric or alternative two-peak shapes, the coexistence conclusion would be substantially strengthened; the current evidence rests on the chosen line shapes."],"forward_implications":["The integrated intensity of the first sharp diffraction peak becomes a direct experimental order parameter for tetrahedral ordering in liquid water, replacing simulation-only descriptors.","The width of that peak gives the coherence length of tetrahedral order, which grows on cooling and is bounded between about 2 Å (single tetrahedron) and 6.5 Å (LDA ice), quantifying how short-ranged the ordering is.","Below roughly $1.1\\,T_{s=1/2}$, a one-peak description of the first diffraction peak fails for both real and model water, explaining why the two-state signature is invisible at ambient conditions where $s$ is small.","The same doublet appears in real water and in three popular models, with model differences showing up as different rates of growth of $s$, meaning the method can rank how over-structured a model is.","The result supports the two-state picture of water as a mixture of locally favored tetrahedral structures and disordered normal-liquid structures, distinct from macroscopic low-density and high-density liquid phases."],"supporting_citations":[{"why":"Establishes that the FSDP of tetrahedral liquids comes from the tetrahedral-unit density wave at $k_{T1}\\approx 3/4$ and that its intensity is proportional to the LFTS fraction; the water analysis imports this directly.","marker":"[26]"},{"why":"Introduces the locally favored structure view of water anomalies and the structural descriptor used to obtain $s$ independently.","marker":"[11]"},{"why":"Shows the two-state decomposition of the scattering function in silica, the template for the water fit.","marker":"[12]"},{"why":"Defines the microscopic structural descriptor $\\zeta$ for liquid water used in the Debye-scattering analysis.","marker":"[13]"},{"why":"Provides the two-state thermodynamic parameters for TIP5P and ST2 water used to constrain the temperature dependence of $s$.","marker":"[15]"},{"why":"Supplies the Debye scattering equation that allows the $\\zeta$-resolved local structure factor $S(k,\\zeta)$.","marker":"[28]"},{"why":"Provides the x-ray O–O partial structure factor data of real supercooled water down to 254.1 K used in the experimental fits.","marker":"[29]"},{"why":"Provides additional supercooled-water x-ray data down to about 235 K that extend the experimental test below the Schottky temperature.","marker":"[30]"},{"why":"Provides the LDA ice structure factor used as the high-order limit for coherence length and FSDP shape.","marker":"[32]"},{"why":"Gives the independent small-angle x-ray estimate of correlation length (about 4.1 Å) used to validate the coherence length scale.","marker":"[35]"}],"fun_headline_variants":["Water's first diffraction peak actually hides two local structural motifs","Scattering data split water's first peak into tetrahedral and ordinary components","The century-old water structure debate is settled by a hidden peak doublet","Two overlapping peaks in water's structure factor reveal coexisting structural motifs"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is that the apparent first diffraction peak is exactly a Lorentzian component at $k_{T1}\\approx 3/4$ plus a Gaussian component at $k_{D1}\\approx 1$, with the Lorentzian's integrated intensity strictly proportional to the tetrahedral fraction $s$; these line shapes and the proportionality constants are assumed and fitted, not derived, so if a single asymmetric peak or a different two-component model describes the same data equally well, the coexistence conclusion does not follow.","fun_headline_variants_meta":{"raw":{"variants":["Water's first diffraction peak actually hides two local structural motifs","Scattering data split water's first peak into tetrahedral and ordinary components","The century-old water structure debate is settled by a hidden peak doublet","Two overlapping peaks in water's structure factor reveal coexisting structural motifs"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000296,"raw_usage":{"total_tokens":1723,"prompt_tokens":955,"completion_tokens":768,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":571,"completion_tokens_details":{"reasoning_tokens":693}},"tokens_in":571,"tokens_out":768,"duration_ms":7539,"temperature":1.0,"reasoning_tokens":693,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:49:19.585801+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"One could re-fit the experimental and simulated O–O structure factors below 240 K with a single asymmetric peak or with two Gaussians of free position, and compare models by an information criterion; the two-motif claim would be falsified if the Lorentzian-plus-Gaussian doublet is not clearly preferred and its $k_{T1}$ integrated intensity does not scale with the independently measured fraction of tetrahedral structures across temperature. A second check: resolve $S(k,\\zeta)$ by molecular dynamics and verify that the $k_{T1}$ peak appears only in the high-$\\zeta$ tetrahedral subpopulation; if the disordered subpopulation contributes comparably at $k_{T1}$, the assignment collapses.","supporting_citations":[{"cited_title":"& Tanaka, H","cited_arxiv_id":null,"evidence_quote":"Establishes that the FSDP of tetrahedral liquids comes from the tetrahedral-unit density wave at $k_{T1}\\approx 3/4$ and that its intensity is proportional to the LFTS fraction; the water analysis imports this directly."},{"cited_title":"& Tanaka, H","cited_arxiv_id":null,"evidence_quote":"Shows the two-state decomposition of the scattering function in silica, the template for the water fit."},{"cited_title":"& Tanaka, H","cited_arxiv_id":null,"evidence_quote":"Defines the microscopic structural descriptor $\\zeta$ for liquid water used in the Debye-scattering analysis."},{"cited_title":"& Tanaka, H","cited_arxiv_id":null,"evidence_quote":"Provides the two-state thermodynamic parameters for TIP5P and ST2 water used to constrain the temperature dependence of $s$."},{"cited_title":"Zerstreuung von röntgenstrahlen","cited_arxiv_id":null,"evidence_quote":"Supplies the Debye scattering equation that allows the $\\zeta$-resolved local structure factor $S(k,\\zeta)$."},{"cited_title":"B., Benmore, C., Neuefeind, J","cited_arxiv_id":null,"evidence_quote":"Provides the x-ray O–O partial structure factor data of real supercooled water down to 254.1 K used in the experimental fits."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides additional supercooled-water x-ray data down to about 235 K that extend the experimental test below the Schottky temperature."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the LDA ice structure factor used as the high-order limit for coherence length and FSDP shape."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the independent small-angle x-ray estimate of correlation length (about 4.1 Å) used to validate the coherence length scale."}],"review_version":1}