{"id":"9f5fa0c9-cd3b-455d-ae32-35f3c4b62965","arxiv_id":"2411.14660","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"A tensorially generalised version of the Terracotta clay model, specialised to direct shear, yields a friction law for fault gouges whose rate-and-state, dilational, and thickness-dependent behaviour is validated against clay-gouge experiments.","lead":"Fault gouge friction is derived from a hydrodynamic clay model, giving a friction law that depends on gouge layer thickness, solid fraction, and normal stress as well as slip rate. The law reproduces laboratory velocity-step tests and, coupled to a spring-dashpot host rock, produces stick-slip earthquake cycles with elastic buildup before rupture.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Uniform-strain assumption γ̇=v/h with full gouge thickness is the most load-bearing simplification: thickness-dependent earthquake predictions scale with it, while real clay gouges localize to much thinner bands.","rationale":"The paper's central claim combines a new friction law with earthquake-cycle predictions. The homogeneous-shear assumption is the weakest step in that chain because it directly controls the strain rate and hence the viscous stress, the thickness scaling, and the dynamic response. The authors themselves flag it, and the reader correctly identified it as the most fragile link. I agree with the CONDITIONAL verdict: the derivation and laboratory validation are reasonably convincing (especially the predictive Ashman & Faulkner test), but the earthquake predictions require the uniform-strain assumption. A single numerical test—rerunning the benchmark with a thinner, microstructurally plausible h0—would quantify the sensitivity. If the results are robust, the verdict should stand; if not, the earthquake sections need revision or a localization model. No ad hominem; the concern is on the argument, and the test is concrete.","tokens_in":33409,"tokens_out":12477,"duration_ms":128696,"concrete_test":"Re-run the Terracotta-spring benchmark and the thickness parametric study (Figs. 12-13) with h0 set to the localized shear-band thickness inferred from clay microstructure, e.g., h0 = 50 µm instead of 0.1-10 mm, while adjusting φ0 to preserve the same overconsolidation ratio. If the stick-slip regime, stress drops, or recurrence intervals change by more than a factor of two, the uniform-strain assumption is load-bearing and the earthquake predictions are conditional on it.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 2.3.2 and Eq. (26a) set γ̇ = v/h with h the full gouge thickness, an assumption the authors explicitly acknowledge 'neglects the coseismic evolution of shear localisation from a full gouge thickness to a localised band.' This is not a minor detail: the viscous term in Eq. (24) is B T_m v/h, and the steady-state friction Eqs. (37)-(38) scale with (v/h)^2. Consequently, all thickness-dependent results in Sections 4-5—the critical thickness separating stick-slip from stable sliding, the amplitude of friction drops, and recurrence intervals (Figs. 12-13)—are direct consequences of this 1/h scaling. In clay-rich gouges, shear localizes to a band of 10-100 times the platelet diameter (≈10-100 µm), which is 1-2 orders of magnitude thinner than the h0 = 0.1-10 mm used in the spring-slider simulations. During an earthquake the active band thickness can evolve, so h in Eq. (24) should be a dynamic state variable, not an imposed full layer thickness. The model has no internal mechanism to predict h (the second-gradient meso-temperature term that could provide a length scale is explicitly beyond scope). Thus the bridge from the laboratory-validated friction law to field-scale earthquake predictions rests on an unquantified and potentially incorrect kinematic assumption.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper derives a friction law for clay-rich fault gouges by starting from the Terracotta constitutive model for clay, generalizing it to tensorial form, and then specializing it to direct shear boundary conditions. The resulting Terracotta friction law, Eq. (24), expresses the shear stress as an elastic term proportional to ε_v^e γ^e / h^6 and a viscous term proportional to T_m v / h, with state evolution equations for elastic strains, meso-temperature, and thickness. The authors calibrate the model to published direct-shear experiments on illite, smectite, and kaolinite gouges, compare the predictions against additional normal-stress experiments, and then couple the friction law to a spring-dashpot host-rock model to simulate earthquake nucleation, stick-slip cycles, and recurrence statistics. The paper claims that the resulting framework is a physics-based alternative to empirical rate-and-state friction laws, with explicit dependence on gouge thickness, normal stress, and solid fraction.","tokens_in":33680,"tokens_out":4902,"duration_ms":54991,"significance":"If the central claims are accepted, the paper would offer a useful route from a thermodynamically grounded clay constitutive model to a friction law that naturally includes layer thickness, density state, and dilatancy, and it would connect laboratory friction behavior to a spring-slider seismicity model. The strengths of the manuscript are the analytic steady-state reduction in Appendix B, the availability of the numerical code, and the out-of-sample normal-stress comparison with Ashman and Faulkner (2023), which is genuinely predictive because no additional parameter adjustment is made for that dataset. However, the significance is limited by two load-bearing issues: most of the validation is in-sample calibration rather than independent prediction, and the field-scale earthquake predictions rest on the explicit assumption of uniform shear across the full gouge thickness, an assumption the authors themselves acknowledge neglects shear localization. These issues do not invalidate the derivation, but they need to be addressed or clearly bounded before the broader claims can be accepted.","major_comments":[{"comment":"The validation protocol is largely in-sample calibration, not independent prediction. The opening paragraph of Section 3 states that the parameters are 'iteratively updated to fit the experimental curves,' and Section 3.1 additionally recalibrates c, a, η, and M for the 100 MPa illite and smectite tests (c = 45·10^6 Ks, η = 53·10^3 K^-1s^-1, M = 0.86 for illite, with different values for smectite). Consequently, the agreement in Figures 4 and 5 is expected, and the claim that the law 'validates' rate-and-state behavior is overstated. The only truly out-of-sample test is the Ashman and Faulkner (2023) normal-stress comparison in Figure 7. Please report the full parameter set for each protocol, report misfit or uncertainty measures, and clearly distinguish fitted curves from predictive curves throughout Sections 3 and 6.","section":"§3 and Table 1"},{"comment":"The uniform-strain assumption γ̇ = v/h with h equal to the full gouge thickness is load-bearing for the earthquake predictions. The paper explicitly acknowledges in Section 2.3.2 that this 'neglects the coseismic evolution of shear localisation from a full gouge thickness to a localised band,' but the friction law Eq. (24) and the steady-state expression Eq. (38) scale with 1/h and 1/h_s^2, respectively. All thickness-dependent results in Section 5.2.1 and Figures 12-13, including the critical thickness separating stick-slip from stable sliding, the stress-drop amplitudes, and the recurrence intervals, therefore depend directly on the choice of full-layer h. In clay-rich gouges the localized band is often one to two orders of magnitude thinner than the h0 = 0.1-10 mm used in the simulations, and the model has no internal equation for the active band thickness because the second-gradient meso-temperature term is explicitly set aside in Section 2.2.1. Please quantify the sensitivity of the Section 5 predictions to h, or reformulate the model so that the active thickness is a dynamic field, before presenting the field-scale stick-slip and recurrence results as quantitative predictions.","section":"§2.3.2 and Eq. (26a)"},{"comment":"The steady-state velocity dependence is not independently predicted: the steady-state friction coefficient in Eq. (38) contains B, c, η, M, and h_s, and h_s in Eq. (35) depends on φ0, h0, ω, p_I, and λ. All of these quantities are calibrated, either directly or indirectly, from the same velocity-step experiments used for the validation in Section 3. In other words, the rate-strengthening or rate-weakening behavior reproduced in Figures 4-6 is largely a consequence of the fitted transport coefficients rather than a parameter-free consequence of the hydrodynamic framework. To make the predictive claim credible, please show which parameters can be fixed from independent triaxial, isotropic compression, or elastic-wave tests, and which ones necessarily require direct-shear velocity steps, and discuss how parameter uncertainty propagates into the steady-state friction law.","section":"§B, Eq. (38)"}],"minor_comments":[{"comment":"The parameter values used for the 100 MPa illite and smectite predictions are reported only in the text and are not entered in Table 1, which makes the Figure 5 results difficult to reproduce; please add them to the table or to a supplementary table.","section":"§3.1 and Figure 5"},{"comment":"The numerical implementation uses the initial thickness h0 rather than the current thickness h in the constraint (26b), with the justification that the normal strain remains below about 5%. Please state this approximation directly in the main text where Eq. (26b) is introduced, rather than only in the appendix, and justify that the 5% bound holds for all simulations including the parametric sweeps in Section 5.","section":"Appendix A"},{"comment":"The notation in Eq. (24), with the underbrace labels 'state' and 'rate', is visually unclear; the variables v and v_n are rates, but T_m and the elastic strains are also evolving state variables. A cleaner grouping of arguments would improve readability.","section":"§2.3, Eq. (24)"},{"comment":"The conclusion states that the friction law 'explains for the first time analytically the buildups of an elastic stress towards first rupture.' The elastic buildup is already evident from the elastic component in Eq. (24) and from the spring-slider integration, but the claim of analytic explanation should be supported by a direct analytic expression for the pre-rupture stress evolution, or softened.","section":"§6"},{"comment":"The energy budget quantities in Figure 16(a) are defined visually rather than by explicit formulas in the text; please provide the equations used to compute E_R, E_G, and E_F so that the reported values E_G = 0.21 J/m^2, E_F = 170 J/m^2, and E_R = 0.04 J/m^2 can be verified.","section":"§5.1 and Figure 16"}],"recommendation":"major_revision","confidential_remarks":"This is a substantively interesting paper with a sound thermodynamic derivation and a genuine out-of-sample check in the Ashman and Faulkner comparison. The main risk is rhetorical overclaiming: the validation is largely in-sample calibration, and the earthquake-scale results depend on a uniform-strain assumption that the authors themselves flag as neglecting localization. I do not see a fatal internal inconsistency, so the appropriate route is major revision rather than rejection. I would encourage the editor to ask for a revised version that separates calibrated fits from true predictions, quantifies the thickness sensitivity, and either adds an evolution equation for the active shear-band thickness or explicitly restricts the field-scale claims to the uniform-band idealization."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, the paper actually delivers a new friction law: Eq. (24) with explicit dependence on gouge thickness, solid fraction, normal stress, and meso-temperature, derived by tensorially generalizing the Terracotta clay model and specializing to direct shear. The steady-state closed form in Appendix B is a genuine addition. Second, the Ashman–Faulkner comparison is the honest highlight: same kaolinite parameters as the Bedford calibration, different normal stresses, predictions hold. That is real predictive evidence, not just curve fitting.\n\nThe soft spots are in proportion. The velocity-dependent behavior is not independently predicted: the parameters B, c, eta, M, and h_s that enter the steady-state friction coefficient in Eq. (38) are calibrated against the same velocity-step experiments used for validation. Table 1 is explicit that parameters are 'iteratively updated to fit the experimental curves.' So the rate-and-state-like behavior is demonstrated as a modeling capability, not as a forecast. The authors note the quadratic velocity dependence is a limitation, and they are honest about it.\n\nThe bigger issue is the uniform-strain assumption gamma_dot = v/h with h the full gouge thickness. Section 2.3.2 concedes this neglects coseismic shear localization. Everything in Sections 4–5 that depends on thickness—critical thickness for stick-slip, stress drop amplitudes, recurrence intervals—scales with that 1/h. Real clay gouges localize to bands one to two orders of magnitude thinner than the h0 values used in the spring-slider simulations, and the model has no internal mechanism to track the active band thickness. The second-gradient meso-temperature term that could provide a length scale is explicitly left out. So the earthquake-cycle predictions are a plausible demonstration of the model's qualitative repertoire, not a robust quantitative bridge to the field.\n\nMinor: the claim to explain elastic buildup 'for the first time analytically' is overstated; elastic precursor behavior exists in other spring-slider treatments. Not a big deal.\n\nNet: this deserves a serious referee. The derivation is coherent, the normal-stress prediction is one independent check, and the thickness dependence is structurally new. Referee time should push on the validation protocol and the localization assumption. I would not cite it as a validated rate-and-state alternative yet, but I would cite it as a new physics-based friction law with a genuinely predictive component.\n\nRecommendation: send to peer review with a request for major revision—make the fitting explicit, show which parameters are fixed across datasets, and add a discussion (if not a solution) of how the active shear-band thickness should evolve.","headline":"A genuinely new physics-based friction law with one solid holdout prediction, but the earthquake-cycle extension rests on an acknowledged and unquantified uniform-thickness assumption.","tokens_in":34265,"tokens_out":2653,"would_cite":true,"duration_ms":26587,"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":"The paper claims that a tensorially generalised hydrodynamic clay model yields a friction law whose thickness, normal-stress, and solid-fraction dependence reproduces laboratory velocity steps and stick-slip earthquake cycles.","keywords":["fault gouge friction","rate-and-state friction","hydrodynamic constitutive model","meso-temperature","clay","stick-slip","earthquake nucleation","dilatancy"],"falsifier":"Resolve the shear strain profile inside a clay gouge layer during direct shear while measuring stress drops and recurrence intervals for initial thicknesses from about 0.01 mm to 10 mm; if a persistent shear band much thinner than the nominal layer carries the slip, or if the peak-and-decrease pattern of stress drop versus thickness is absent, the uniform-strain foundation of Eq. (24) fails.","tokens_in":33148,"feed_emoji":"🌋","tokens_out":10108,"duration_ms":89926,"temperature":0.7,"pith_summary":"This paper claims that the frictional behaviour of clay-rich fault gouges, including its rate-and-state character, dilation, and dependence on layer thickness, normal stress, and solid fraction, can be derived rather than fitted. The route is to take the Terracotta hydrodynamic constitutive model for clay, which was built for triaxial loading, generalise it to tensorial form, and then specialise it to direct shear. The resulting Terracotta friction law has two explicit terms, one elastic and one viscous, with the meso-temperature of the clay as the state variable that controls transients. Validated against velocity-stepping experiments on illite, smectite, and kaolinite, the law retains the triaxial model's predictive power and explains dilatancy through evolution of gouge thickness. Coupled to a spring-dashpot host rock, it produces elastic stress buildup, stick-slip nucleation, and periodic seismicity whose stress drops and recurrence intervals match laboratory and field trends, so a sympathetic reader would take away that rate-and-state friction is not a standalone empirical law but a special case of a broader material physics.","feed_headline":"Hydrodynamics of clay yields a fault-gouge friction law","feed_subtitle":"The derived law makes layer thickness, normal stress, and solid fraction explicit, and reproduces stick-slip cycles.","key_machinery":"The carrying object is the Terracotta friction law, Eq. (24), produced by the hydrodynamic machinery of two-scale temperatures: a thermal temperature for atomic fluctuations and a meso-temperature $T_m$ for the kinetic energy of clay platelets and aggregates, with energy flowing from meso to micro scale through a sink term. Around that object, the argument uses four pieces: thermodynamically reciprocal transport coefficients that couple volumetric and deviatoric plastic strain rates; a critical-state compression line $p_c(\\phi) = p_I \\phi^{\\lambda}$ that ties solid fraction to pressure; the tensorial generalisation of the Terracotta elastic energy with elastic instability criterion $q^e/p^e \\le M^e$; and direct-shear boundary conditions, $v = \\dot{\\gamma} h$ and constant $\\sigma_n$, that reduce the tensorial model to the two-term friction law. The evolution of $T_m$, Eq. (18), controls the transient and steady-state viscous response, while the evolution of $h$, Eq. (25d), makes dilatancy and solid-fraction changes part of the state. This machinery converts what are usually empirical rate-and-state parameters into measurable hydrodynamic, elastic, and critical-state quantities.","core_discovery":"On its own terms, the paper's central discovery is a closed friction law, Eq. (24): $\\tau = (A/h^6)\\,\\varepsilon_v^e\\,\\gamma^e + (B/h)\\,T_m\\,v$, where $h$ is gouge thickness, $\\varepsilon_v^e$ and $\\gamma^e$ are elastic volumetric and shear strains, $T_m$ is the meso-temperature, and $A$, $B$ lump elastic and rheological constants. The first term is an elastic stress that builds before rupture and carries the model's density- and pressure-dependent elasticity; the second is a viscous stress controlled by the fluctuating kinetic energy of clay aggregates and platelets. The law is obtained by writing the Terracotta clay model in full tensorial form and imposing constant-normal-stress direct-shear kinematics, so the thickness $h$ evolves with compaction and dilation and the solid fraction follows $\\phi = \\phi_0 h_0 / h$. The paper shows that this single law reproduces velocity-step experiments on illite shale, smectite, and kaolinite, including transient peaks, isotach behaviour, dilation and compaction, and normal-stress dependence without recalibrating material constants. When the law is attached to a spring-dashpot host rock, the combined Terracotta-spring model predicts an elastic loading phase, a first earthquake, subsequent periodic stick-slip events, and eventual convergence to aseismic sliding, with stress drops and recurrence statistics consistent with laboratory and natural seismicity data.","pith_inferences":["If the derivation is as general as claimed, the same hydrodynamic route could produce friction laws for saturated, chemically active, or higher-temperature gouges by adding the corresponding degrees of freedom, since the underlying balance and entropy-production structure already accommodates multiphysics couplings.","The steady-state law's quadratic dependence on slip velocity, Eq. (38), is a concrete, testable signature that distinguishes it from logarithmic rate-and-state laws; velocity-step experiments spanning several decades of $v$ would separate the two.","Because solid fraction was used as a fitting parameter in the validations, direct measurements of $\\phi_0$ and of the strain profile across the layer, rather than inferred values, would provide the sharpest independent test of the model's thickness and volumetric predictions."],"forward_implications":["Rate-and-state friction parameters acquire physical meaning: they follow from elastic moduli, critical-state constants, and rheological coefficients that can be calibrated in triaxial or isotropic tests, so the same material description works under non-fault loadings.","Gouge thickness becomes a dynamic state variable rather than a fixed surface property, so the model predicts how stress drops and return periods vary with layer thickness, including a critical thickness that maximises stress drop.","The model provides an analytical account of elastic shear-stress buildup before the first rupture, a precursor phase that phenomenological rate-and-state laws cannot represent.","Coupled to a spring-dashpot host rock, the model generates complete seismic cycles, including an isolated first event, periodic stick-slip, and eventual stable sliding, with dilation-compaction cycles synchronised to the meso-temperature and matching observed trends."],"supporting_citations":[{"why":"Supplies the Terracotta clay constitutive model in triaxial form that the paper generalises and then specialises to direct shear.","marker":"(Wiebicke and Einav, 2024)"},{"why":"Provides the hydrodynamic procedure for conservation laws, entropy production, and thermodynamic fluxes underlying the model.","marker":"(Landau and Lifshitz, 2013)"},{"why":"Introduces the two-stage irreversibility and two-temperature framework from which the meso-temperature originates.","marker":"(Jiang and Liu, 2009)"},{"why":"Gives the reciprocal relations that set the plastic and viscous transport coefficients coupling volumetric and deviatoric fluxes.","marker":"(Onsager, 1931)"},{"why":"Provides the illite and smectite velocity-step experiments used to calibrate and validate the friction law.","marker":"(Saffer and Marone, 2003)"},{"why":"Supplies kaolinite direct-shear data with measured gouge thickness and normal velocity used to validate the volumetric behaviour.","marker":"(Bedford et al., 2022)"},{"why":"Provides kaolinite velocity-step experiments under different normal stresses used to test the predicted normal-stress dependence without recalibration.","marker":"(Ashman and Faulkner, 2023)"},{"why":"Gives laboratory stick-slip data on granular fault gouge showing thickness-dependent stress drops and return intervals that the Terracotta-spring model reproduces.","marker":"(Lyu et al., 2019)"},{"why":"Establishes the localised shear-zone thickness scale and coseismic weakening context used to set initial gouge thickness values.","marker":"(Rice, 2006)"}],"fun_headline_variants":["New friction law for fault gouges from clay hydrodynamics","Hydrodynamics of clay captures fault gouge stick-slip","Fault gouge law links thickness, stress, and sliding","From clay physics to earthquake cycles in one law"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is that the gouge shears uniformly through its full thickness, so the strain rate is exactly $v/h$; if slip localises into a thinner band whose width changes during an earthquake, the predicted stress drops, thickness scaling, and recurrence intervals would change.","fun_headline_variants_meta":{"raw":{"variants":["New friction law for fault gouges from clay hydrodynamics","Hydrodynamics of clay captures fault gouge stick-slip","Fault gouge law links thickness, stress, and sliding","From clay physics to earthquake cycles in one law"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000444,"raw_usage":{"total_tokens":2323,"prompt_tokens":1101,"completion_tokens":1222,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":717,"completion_tokens_details":{"reasoning_tokens":1156}},"tokens_in":717,"tokens_out":1222,"duration_ms":10084,"temperature":1.0,"reasoning_tokens":1156,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T15:03:32.396402+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Resolve the shear strain profile inside a clay gouge layer during direct shear while measuring stress drops and recurrence intervals for initial thicknesses from about 0.01 mm to 10 mm; if a persistent shear band much thinner than the nominal layer carries the slip, or if the peak-and-decrease pattern of stress drop versus thickness is absent, the uniform-strain foundation of Eq. (24) fails.","supporting_citations":[{"cited_title":"\\ Einav, I","cited_arxiv_id":null,"evidence_quote":"Supplies the Terracotta clay constitutive model in triaxial form that the paper generalises and then specialises to direct shear."},{"cited_title":"\\ Liu, M","cited_arxiv_id":null,"evidence_quote":"Introduces the two-stage irreversibility and two-temperature framework from which the meso-temperature originates."},{"cited_title":"\\ Marone, C","cited_arxiv_id":null,"evidence_quote":"Provides the illite and smectite velocity-step experiments used to calibrate and validate the friction law."},{"cited_title":", Faulkner, D R","cited_arxiv_id":null,"evidence_quote":"Supplies kaolinite direct-shear data with measured gouge thickness and normal velocity used to validate the volumetric behaviour."},{"cited_title":", Rivi \\`e re, J","cited_arxiv_id":null,"evidence_quote":"Gives laboratory stick-slip data on granular fault gouge showing thickness-dependent stress drops and return intervals that the Terracotta-spring model reproduces."}],"review_version":1}