{"id":"269e7170-cb4b-4882-98db-d88c476f87ef","arxiv_id":"2505.11819","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":2.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"A stochastic LGD model with Gaussian disorder and depolarization fields yields correlation functions that fit some, but not all, TGS domain-growth data; the paper largely restates the authors' earlier results.","lead":"This paper reviews a stochastic Landau-Ginzburg-Devonshire model for how polarization domains form in uniaxial ferroelectrics after rapid cooling. It argues the model can explain three decades of triglycine sulfate experiments and predict how quench conditions set the coercive field.","discovery_kind":"review","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The Gaussian-random-field closure and per-dataset choice of initial correlation shape are the load-bearing assumptions; the paper's own fits show the shape is selected per experiment, undermining the predictive claim.","rationale":"I agree with the reader's assessment: the weakest assumption is the Gaussian-random-field closure plus hand-chosen initial correlation shapes, fitted to the same experiments used for validation. This is the most load-bearing concern because the abstract's headline claims (analytical solvability, explanations of kinetics, coercive-field prediction) all depend on this closure. The paper is internally consistent—the mathematics of the evolution equations appears to follow from the Gaussian closure—but the external validity is not established. The paper's own text concedes that transverse correlation oscillations are not reproduced, that near-transition oscillations (ΔT = 0.1 K, Golitsina) are not described, and that coercive fields are too large, all pointing to the closure limits. The reader's verdict CONDITIONAL is appropriate: the review/synthesis value is real, but the predictive claims need an out-of-sample test and clearer fitting/uncertainty reporting. My concrete test (out-of-sample coercive field prediction from measured initial correlations) is exactly the check that would settle whether the model has predictive power or is just a fitting scheme. I see no reason to change the verdict to ACCEPT or REJECT; the paper is a useful synthesis with a genuine but unproven predictive claim.","tokens_in":25419,"tokens_out":1603,"duration_ms":15030,"concrete_test":"Perform an out-of-sample validation: use the analytical expressions in Supplemental S1 to predict the coercive field E_c for a TGS sample whose initial disorder amplitude and correlation radius are determined directly from AFM/SFM images taken immediately after quench (or from the fitted initial correlation function on one dataset), then compare with the measured switching field on a different dataset not used to select the initial correlation shape (Gaussian vs exponential vs erfc). If the predicted E_c is within the experimental range and the monotonic trend in (D(0), rc) holds without re-fitting, the central claim survives; if the shape or parameters must be re-fit per dataset to match, the model is descriptive rather than predictive.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim—that the model reproduces TGS domain kinetics and predicts coercive fields from quench parameters—rests entirely on the Gaussian-random-field closure introduced around Eq. (2)-(4). This closure has three unvalidated parts: (i) the post-quench polarization field is taken as a Gaussian random field, which is asserted in Section 2 ('considered in this model as Gauss random fields'); (ii) the initial correlation function is chosen as Gaussian, exponential, or erfc (Sections 2 and 3.2), with the shape selected post hoc per experiment; Section 3.2 states that the exponential shape is preferred for Tomita et al. yet 'a validation against additional experiments remains imperative'; (iii) Eq. (4) closes by approximating higher-order correlations in terms of K and the mean polarization. The paper concedes observable failures of this closure: transverse correlations do not fit the oscillations in Tomita et al. (Section 3.2), and the predicted coercive fields are acknowledged as 'still too large' (Section 2). The claimed predictive link to quenching parameters is not demonstrated: the initial parameters (rc, D(0)) are fitted to the same experimental data used for validation, and no out-of-sample test is provided. Thus the strongest claim—that the model predicts coercive-field dependence on initial disorder—is supported only by a fit and a monotonic trend, not by an independent measurement of the initial disorder.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops and reviews a stochastic Landau-Ginzburg-Devonshire (LGD) model for quenched uniaxial ferroelectrics, treating the polarization field as a Gaussian random field with an initial correlation function that can be Gaussian, exponential, or complementary-error-function in form. From a closed system of integrodifferential equations for the mean polarization and the polarization correlation function, the authors derive analytical expressions for correlation lengths, polarization correlation coefficients, electric-field correlations, and cross-correlations. They fit these expressions to TGS experimental data from Tomita et al., Likodimos et al., and Golitsyna et al. in Section 3.2, discuss the resulting phase portrait and coercive-field behavior in Section 2, and present new predictions for the spatial and temporal evolution of correlations in Sections 3.3, 4, and 5. The abstract claims that the system is analytically solvable, that the model explains TGS polarization-formation kinetics, and that it predicts the coercive-field dependence on the initial disordered state.","tokens_in":25624,"tokens_out":4757,"duration_ms":47916,"significance":"If the central claims are fully justified, the model would offer a rare analytical route to domain kinetics on macroscopic scales and a concrete link between quench protocols and switching properties. The paper has genuine strengths: it provides closed-form correlation functions for a model with self-consistent depolarization fields; it engages with a substantial body of TGS experiments; and it transparently acknowledges some limitations, such as the failure to describe transverse correlation oscillations and the overestimated coercive-field magnitude. However, the predictive content is weakened by the circular use of experimental data: the initial correlation shape and its length scale and amplitude are fitted to the same datasets that are used for validation, and no out-of-sample test is provided. The abstract overstates the analytical solvability and predictive status relative to the body of the paper. These issues are load-bearing for the main claims and require careful revision.","major_comments":[{"comment":"The model's predictive claim is undermined by a fitting circularity. The initial correlation function shape is selected per dataset in Section 3.2 (Gaussian, exponential, or erfc), and the parameters rc or ξ, D(0), π̄(0), and t0 are fitted to the same experimental measurements used to claim validation. The paper even states that 'a validation against additional experiments remains imperative' for the exponential-shape preference. Because the predicted coercive-field dependence inherits these fitted initial parameters, the abstract's claim that the model 'predicts the dependence of the ferroelectric coercive field on the initial disordered state characteristics' is not supported by an independent test. The authors should provide an out-of-sample prediction, or derive the initial correlation shape and amplitude from the quench protocol (e.g., from the cooling dynamics of the LGD equation) rather than choosing them post hoc.","section":"§2, Eqs. (4)–(5), and §3.2"},{"comment":"The agreement with experiment is presented mainly by visual inspection and is qualified by the paper's own statements: transverse correlations 'cannot be fitted good within the outcomes of our model', the oscillations in Fig. 6(a) 'defy quantitative description within the confines of the isotropic model', and for the Golitsyna et al. data all three correlation forms are 'deemed suitable only with a notable margin of error'. Given these admitted failures, the abstract's statement that the model 'provides explanations to a range of experimental results ... including the time-dependent correlation lengths and correlation functions' is too strong. The authors should restrict the claim to the quantities and datasets that are actually described well, and add quantitative goodness-of-fit measures or at least clearly state the fitted parameter values and their uncertainties.","section":"§3.2, Figs. 4–6"},{"comment":"The claimed coercive-field prediction is not demonstrated against experiment. The paper states that the predicted coercive fields are 'still too large in comparison with experimental values' and that the 'most significant result' is a monotonic increase of the coercive field with the initial amplitude and spatial size of fluctuations, obtained from numerical solutions of Eq. (5) with fitted inputs. No direct measurement of the initial fluctuation amplitude D(0) or correlation radius rc is presented, so the link between quench parameters (initial temperature, cooling rate) and coercive field remains a model-based trend rather than a tested prediction. The authors should either compare the predicted coercive-field trend with experimental data under varied quench conditions or explicitly label this as a prediction to be tested in future work, and adjust the abstract accordingly.","section":"§2, coercive-field discussion"},{"comment":"The abstract claims that the system of integrodifferential equations for correlation functions is 'analytically solvable', but the text in Section 2 (after Eq. (4)) and the concluding section state that 'the full analysis requires numerical solution of the system of nonlinear differential equations' and that 'the derived equations necessitate numerical solutions'. The analytical results apply to the correlation functions once the mean-field trajectory and the initial correlation form are supplied, not to the full evolution system. This discrepancy should be resolved by stating precisely what is analytically solved and what requires numerical treatment, both in the abstract and in Section 2.","section":"Abstract and §2, Eq. (4)"}],"minor_comments":[{"comment":"The heading 'CONCUSION' should be corrected to 'CONCLUSION'.","section":"§6 heading"},{"comment":"In the sentence introducing the Fourier transforms, the correlation function Rαβ(s, τ) appears without its time argument: 'Rαβ(s,)' should read 'Rαβ(s, τ)'.","section":"§2, after Eq. (4)"},{"comment":"The term '(q2 + η qz2/q2)' contains an unreadable glyph for η in the typeset equation; please ensure the equation renders correctly.","section":"§2, Eq. (4)"},{"comment":"The caption 'with exponential correlation coefficient [60]' appears to contain a citation error; the exponential-form correlation fits are from the authors' prior work [70], not from reference [60].","section":"Figure 6 caption"},{"comment":"Reference [59] spells the author name 'Kolomgorov' instead of 'Kolmogorov', and several references inconsistently omit article titles; please standardize the reference format.","section":"Reference list"},{"comment":"The names 'Golitsina' and 'Golitsyna' are used inconsistently in the text and figure captions; the spelling should be unified to match the cited references.","section":"Throughout, esp. §3.2"},{"comment":"The symbol ξ is used for the exponential correlation length in Section 2 and also for the long-range correlation length in Eq. (6); consider using a distinct notation (e.g., λ_lr) to avoid confusion.","section":"§3.1, Eq. (6)"}],"recommendation":"major_revision","confidential_remarks":"The manuscript overlaps heavily with the authors' own previous PRB articles [68–70], and the new material consists mainly of additional correlation-profile visualizations and a review-style synthesis. The editor may wish to verify compliance with the journal's policy on prior publication and self-overlap. The central modeling framework is promising, but the circular fitting and overstatements in the abstract need to be addressed before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my take. This is a review of the authors' own stochastic LGD model, not a new result. The core equations, the Gaussian-closure correlation functions, and the coercive-field dependence are all in their earlier PRB papers [68-70]; what this preprint adds is a systematic presentation, parameter scans, and movies of correlation functions, plus a side-by-side comparison with the TGS data of Tomita, Likodimos, and Golitsyna. That is genuinely useful for someone who wants the toolkit in one place.\n\nThe model itself deserves credit. The analytical solvability of the closure is not trivial, and the correlation-length expressions do capture the linear growth seen in several TGS experiments, with an honestly reported dependence on the assumed initial correlation shape. I also appreciate that the authors are upfront about the limitations: transverse correlations miss the oscillations, near-transition oscillations 'defy quantitative description' within the isotropic model, and the predicted coercive fields are too large. That candor is real.\n\nThe soft spots are in the framing. The abstract says no theoretical tools existed until now, which is simply wrong — Tomita et al. used Ising-model fits, and the paper itself cites them. More important, the central claim that the model 'predicts' the coercive field from the initial disordered state overstates what is done. The initial correlation function shape and its length scale are chosen per experiment and fitted to the same data used for validation; the coercive-field result then inherits those fitted inputs, and no out-of-sample test is offered. There are also no error bars on the fits and no deposited code or data. The authors themselves say that validation against additional experiments is imperative, but the abstract has already claimed the prediction.\n\nWho is this for? People working on TGS domain kinetics and stochastic LGD modeling. As a review article it is worth a serious referee, but the referee should require a rewritten abstract, a table of fitted parameters with uncertainties, and at least one out-of-sample check of the coercive-field trend before publication. As a new-results paper it would not pass.","headline":"A useful but overclaiming review of the authors' own stochastic LGD model; the new material is mostly parameter scans and movies, and the flagship coercive-field 'prediction' is fitted to the same data it is validated against.","tokens_in":26224,"tokens_out":2616,"would_cite":false,"duration_ms":27689,"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":"This paper argues that the pattern of domains left by quenching a uniaxial ferroelectric is governed by an analytically solvable stochastic Landau-Ginzburg-Devonshire model in which quenched Gaussian polarization disorder sets correlation…","keywords":["stochastic ferroelectric domain structure formation","polarization correlations","Gauss random variables","triglycine sulfate","Landau-Ginzburg-Devonshire theory","quenched disorder","coercive field","correlation length"],"falsifier":"Measure the full two-point polarization correlation function in a freshly quenched TGS crystal at short times and compare it, without refitting, with the three predicted closed forms; a correlation curve that cannot be represented by any of them would falsify the Gaussian-random-field ansatz. Independently, vary the initial temperature and cooling rate, measure the initial correlation radius immediately after the quench, and check whether the coercive field rises monotonically with that radius as the model predicts.","tokens_in":25128,"feed_emoji":"⚡","tokens_out":14036,"duration_ms":125562,"temperature":0.7,"pith_summary":"The paper argues that the disordered polarization pattern that appears when a uniaxial ferroelectric is quenched from the paraelectric into the ferroelectric phase can be described by a Landau-Ginzburg-Devonshire model in which the polarization is a Gaussian random field and quenched disorder dominates thermal fluctuations. In this formulation the coupled system for the mean polarization and the two-point correlation functions of polarization and electric field becomes analytically tractable, giving closed-form correlation lengths and correlation coefficients. Applied to triglycine sulfate, the model reproduces measured time-dependent correlation lengths and correlation functions on macroscopic scales, and it predicts that the coercive field grows monotonically with the initial amplitude and spatial size of the polarization fluctuations produced by the quench. If correct, this turns the apparently stochastic outcome of a quench into a design parameter: initial temperature and cooling rate can be chosen to tailor domain sizes, hysteresis loops, and switching fields.","feed_headline":"One solvable model sets the rules for ferroelectric quench domains","feed_subtitle":"Closed-form correlations link cooling rate to domain size and coercive field in TGS.","key_machinery":"The load-bearing mechanism is the replacement of the random polarization field by its statistical description: a Gaussian random field with mean $\\bar{\\pi}(\\tau)$ and two-point correlation function $K(s,\\tau)=\\langle \\xi(\\mathbf{r}_1,\\tau)\\,\\xi(\\mathbf{r}_2,\\tau)\\rangle$. Fourier-transforming all correlation functions reduces the stochastic LGD dynamics to a closed pair of equations for $\\bar{\\pi}(\\tau)$ and $\\tilde{K}(\\mathbf{q},\\tau)$, whose zero-mode part becomes a system for the mean and the variance $D(\\tau)=K(0,\\tau)$. The choice of initial correlation shape (Gaussian, exponential, or complementary error function) enters through the correlation length $L(\\tau)$ and therefore controls both the kinetics of ordering and the predicted coercive field. The assumption that quenched disorder dominates thermal fluctuations is what removes the thermal-noise source term and leaves the analytically solvable deterministic system for the correlations.","core_discovery":"In the slab geometry of the TGS experiments, the authors take the post-quench polarization state to be $\\pi(\\mathbf{r},\\tau)=\\bar{\\pi}(\\tau)+\\xi(\\mathbf{r},\\tau)$, with $\\xi$ a zero-mean Gaussian random field, and derive from the LGD functional a closed system of integrodifferential equations for the mean polarization and the Fourier transform of the polarization correlation function $\\tilde{K}(\\mathbf{q},\\tau)$. The central discovery is that for three choices of the initial correlation shape — Gaussian $\\exp(-s^2/r_c^2)$, exponential $\\exp(-s/\\xi)$, and complementary error function — the correlation functions admit closed-form expressions, including the time-dependent correlation length, for example $L(\\tau)=\\sqrt{(r_c^2+4\\tau)/3}$ in the Gaussian case, together with the full set of polarization, electric-field, and cross-correlation coefficients; the accompanying nonlinear evolution of the mean polarization and variance is integrated numerically. These expressions fit the TGS experimental correlation-length data and reproduce the qualitative behavior of the correlation curves along the lamellar domains. The model further predicts that the coercive field — the applied field separating final multidomain from single-domain states — increases monotonically with the initial disorder amplitude and correlation radius, so that quench parameters such as initial temperature and cooling rate control the switching properties.","pith_inferences":["A sharper test than correlation-length fitting would be to record the full two-point correlation function at the earliest accessible time after the quench and see whether any of the three closed forms — Gaussian, exponential, or complementary error function — survives without refitting; the shape is not preserved by the dynamics, so early curves carry the most information about the true initial di","The monotonic coercive-field prediction suggests a controlled quench-protocol experiment: quench TGS from different initial temperatures and at different cooling rates, measure the initial correlation radius immediately after the quench by AFM or PFM, and check that the switching field follows that radius.","The same Gaussian-field machinery could be carried over to other uniaxial ferroelectrics, such as lead germanate, if the missing anisotropy terms are added; whether the analytical solvability survives that extension is an open question with direct bearing on materials where domain kinetics are too fast to observe directly.","The model's suppressed bound-charge correlations imply that apparently uncharged head-to-head and tail-to-tail walls need not contain free carriers; a direct comparison of charge-density signals measured at such walls with the model's predicted order-of-magnitude suppression would test that mechanism."],"forward_implications":["If the model is right, the final domain state after a quench is set by a small number of measurable inputs: initial mean polarization, initial variance, correlation radius, depolarization coefficient, susceptibility, and applied field.","The analytic correlation functions turn domain images from AFM, SFM, and SHG into a direct quantitative readout of quench conditions, without time-consuming phase-field or atomistic simulations.","The coercive field increases monotonically with the initial amplitude and spatial radius of quenched polarization fluctuations, so the same material can be made harder or easier to switch by changing cooling rate and starting temperature.","Including polarization-electric-field cross-correlations makes the multidomain state more stable and raises the coercive field by roughly a factor of three relative to calculations that ignore them.","The predicted correlation functions evolve toward isotropy in the polar plane as domains grow, so long-time monitoring of correlations can distinguish single-domain from multidomain final states."],"supporting_citations":[{"why":"Supplies the first quantitative characterization of ferroelectric domain correlations in a quenched TGS crystal; its correlation length and power-law growth are the benchmark used for fitting.","marker":"[44]"},{"why":"Provides scanning force microscopy measurements of domain growth kinetics in TGS whose time-dependent correlation lengths are compared with the model's predictions.","marker":"[46]"},{"why":"Provides AFM correlation functions and correlation lengths in the polar plane of TGS, used to test the exponential and complementary-error-function initial correlations.","marker":"[55]"},{"why":"Establishes the narrow temperature window near the transition where thermal fluctuations dominate, supporting the assumption that quenched disorder dominates elsewhere.","marker":"[57]"},{"why":"Derives the coupled LGD evolution equations and the Gaussian-random-field closure for the correlation functions on which this review is based.","marker":"[68]"},{"why":"Supplies the analytical expressions for the complete set of polarization, electric-field, and cross-correlation coefficients.","marker":"[69]"},{"why":"Introduces the exponential and complementary-error-function initial correlation shapes and derives the coercive-field dependence on the initial disordered state.","marker":"[70]"}],"fun_headline_variants":["Solvable model decodes ferroelectric quench domains","Cooling rate shapes ferroelectric domains: a solvable model","Exact correlations in quenched ferroelectrics","TGS quench domains predicted by closed-form theory","Ferroelectric quench: solvable theory matches experiments"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire closed-form solution rests on treating the state immediately after the quench as a Gaussian random field with one of three hand-chosen initial correlation shapes, whose amplitude, correlation radius, and time scale are fitted to the same experiments used for validation; if the real quench disorder is non-Gaussian, anisotropic, or described by a different initial correlation function, the analytic correlation functions and the coercive-field trend would not apply.","fun_headline_variants_meta":{"raw":{"variants":["Solvable model decodes ferroelectric quench domains","Cooling rate shapes ferroelectric domains: a solvable model","Exact correlations in quenched ferroelectrics","TGS quench domains predicted by closed-form theory","Ferroelectric quench: solvable theory matches experiments"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000239,"raw_usage":{"total_tokens":1561,"prompt_tokens":1037,"completion_tokens":524,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":653,"completion_tokens_details":{"reasoning_tokens":444}},"tokens_in":653,"tokens_out":524,"duration_ms":5152,"temperature":1.0,"reasoning_tokens":444,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T20:48:24.591525+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the full two-point polarization correlation function in a freshly quenched TGS crystal at short times and compare it, without refitting, with the three predicted closed forms; a correlation curve that cannot be represented by any of them would falsify the Gaussian-random-field ansatz. Independently, vary the initial temperature and cooling rate, measure the initial correlation radius immediately after the quench, and check whether the coercive field rises monotonically with that radius as the model predicts.","supporting_citations":[{"cited_title":"Tomita, H","cited_arxiv_id":null,"evidence_quote":"Supplies the first quantitative characterization of ferroelectric domain correlations in a quenched TGS crystal; its correlation length and power-law growth are the benchmark used for fitting."},{"cited_title":"Likodimos, M","cited_arxiv_id":null,"evidence_quote":"Provides scanning force microscopy measurements of domain growth kinetics in TGS whose time-dependent correlation lengths are compared with the model's predictions."},{"cited_title":"Golitsyna and S.N","cited_arxiv_id":null,"evidence_quote":"Provides AFM correlation functions and correlation lengths in the polar plane of TGS, used to test the exponential and complementary-error-function initial correlations."},{"cited_title":"Nattermann, Static and dynamic critical behaviour of uniaxial ferroelectrics and the phase transition in TGS, Phys","cited_arxiv_id":null,"evidence_quote":"Establishes the narrow temperature window near the transition where thermal fluctuations dominate, supporting the assumption that quenched disorder dominates elsewhere."},{"cited_title":"Mazur, L.I","cited_arxiv_id":null,"evidence_quote":"Derives the coupled LGD evolution equations and the Gaussian-random-field closure for the correlation functions on which this review is based."},{"cited_title":"Genenko, O.Y","cited_arxiv_id":null,"evidence_quote":"Supplies the analytical expressions for the complete set of polarization, electric-field, and cross-correlation coefficients."},{"cited_title":"Mazur, L.I","cited_arxiv_id":null,"evidence_quote":"Introduces the exponential and complementary-error-function initial correlation shapes and derives the coercive-field dependence on the initial disordered state."}],"review_version":1}