{"id":"02594f9c-97f0-4505-9650-050720d923ca","arxiv_id":"1908.04739","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":7,"one_line_summary":"A modified Coulomb friction law with a state variable that decays with sliding distance reproduces the measured torque-angle curve of mode III implant debonding, but only after its parameters are fitted to that same curve.","lead":"This paper replaces the constant friction coefficient in Coulomb's law with a sliding-distance-dependent coefficient to model how osseointegrated implants debond from bone under twisting. It calibrates the new law against two published torque experiments and reports errors below 2.25 percent for the fitted curves.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Eq. (49) in Appendix B is internally inconsistent with Eqs. (17)-(18): the analytical sliding distance is negative or zero for points that Eq. (18) places in the sliding region.","rationale":"The reader's verdict of CONDITIONAL is appropriate, and the reader's weakest-assumption analysis (unmeasured normal pressure, parameter identifiability) is a valid and clearly acknowledged limitation. However, the single most load-bearing concern is more specific and internal: the analytical derivation in Appendix B contains an inconsistency between Eq. (49) and the stick/slip kinematics defined by Eq. (17). This is not a matter of interpretation but a checkable mathematical fault. It directly affects the paper's central claim because the abstract and the results present the analytical model as reproducing the experimental torque curve with small error. If Eq. (49) is wrong, the analytical curves in Fig. 4 are not derived from the proposed friction law, and the parameter estimates claimed to be consistent between the analytical and numerical models are not trustworthy. The FE implementation uses a different, path-integrated sliding distance (Eq. (32)), so the numerical results may remain valid; therefore rejection is too strong. The paper should be revised to correct the analytical kinematics and recompute the analytical torque curves and any parameters obtained from them, which is consistent with a conditional acceptance. My agreement with the reader is partial because I identify a different load-bearing concern than the one emphasized in the reader's weakest_assumption, though both point to weaknesses in the physical interpretation of the model.","tokens_in":21531,"tokens_out":14274,"duration_ms":143013,"concrete_test":"Re-derive the analytical torque curve using the physically correct sliding distance for the sliding region: g_s = 0 for r < c(θ), and g_s = max(0, rθ − θ_lin R) for r ≥ c(θ). With the parameters in Table 4 for data set 1, compute T(θ) from Eq. (37) and compare it against the experimental data and the published 'ana. Eq. (18)' curves. If the corrected curve differs by more than 2.25% relative error or shows a visibly different softening region, Eq. (49) is confirmed as an error. A quick arithmetic check: at θ = 1°, r = R, θ_lin = 0.13°, θ_max = 1.13°, R = 2.5 mm, Eq. (49) gives g_s = 0 μm, whereas the physical value is R(1° − 0.13°) ≈ 38 μm; the analytical torque curve for 0.13° < θ < 1.0° therefore does not represent the proposed friction law.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The analytical model underpinning the paper's central claim contains a kinematic inconsistency. In Appendix B, Eq. (49) states that in the sliding region g_s = r(θ − θ_max + θ_lin). For a point at radius r, sliding begins when the critical radius c(θ) = θ_lin R/θ reaches r, i.e. at θ_start = θ_lin R/r. The accumulated sliding distance should then be g_s = r(θ − θ_start) = rθ − θ_lin R, not the expression in Eq. (49). The two formulas coincide only if θ_lin R = r(θ_max − θ_lin), which is not generally true. For data set 1 (θ_lin = 0.13°, θ_max = 1.13°, R = 2.5 mm), Eq. (49) gives g_s = R(1° − 1.13° + 0.13°) = 0 μm at θ = 1° and r = R, while the physical sliding distance is R(1° − 0.13°) ≈ 38 μm. More seriously, for 0.13° < θ < 1.0°, Eq. (49) yields negative sliding distances, which are unphysical. Since Eq. (18) defines a sliding region for r ≥ c(θ) at every θ > θ_lin, the analytical torque curves attributed to 'ana. Eq. (18)' in Fig. 4 are not valid consequences of the proposed friction law. The finite element model, which computes g_s by direct accumulation (Eq. (32)), may be unaffected, but the paper's central claim explicitly includes the analytical model of Sec. 3.3, and the reported analytical errors of 2.18% and 2.83% are therefore suspect. The authors need to correct Eq. (49) or clarify the definition of θ_max; until then the analytical section cannot be regarded as a correct derivation.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a Modified Coulomb's law with a state-variable friction coefficient that decays as a function of accumulated sliding distance (Eqs. (11)-(13)) to describe the tangential debonding of osseointegrated implants. The law is implemented in a 3D nonlinear finite element framework and in a new analytical model (Sec. 3.3, Appendix B) for mode III cleavage of a coin-shaped titanium implant, and is calibrated against two experimental torque-angle datasets from Mathieu et al. [32]. The numerical results reproduce the experimental curves with a mean relative error below 2.25% (Table 4), and the paper reports improved estimates of the bone shear modulus and adhesion energy compared with the reference analytical model.","tokens_in":21965,"tokens_out":11283,"duration_ms":106389,"significance":"If the proposed law is accepted, it offers a simple and computationally efficient way to model the transition from intact osseointegration to frictional sliding, and the FE implementation naturally handles partial osseointegration patterns. The paper is thorough in its parameter studies and mesh convergence analysis, and it explicitly lists the limitations of the experimental data. However, the central claim is weakened by the fact that the reported errors are in-sample calibration errors and by an error in the analytical derivation that calls the analytical torque curves into question.","major_comments":[{"comment":"The accumulated sliding distance in the sliding region is incorrect. For a point at radius r, sliding begins at θ_start = θlin R/r, so the physical sliding distance is g_s = r(θ - θ_start) = rθ - θlin R. The paper's expression g_s = r(θ - θmax + θlin) is negative for a significant range of θ (for data set 1, for θ between 0.13° and 1.0° at r = R) and yields zero at θ = 1.0°, whereas the correct value is approximately 38 μm at θ = 1.0°. Since Eq. (18) places all points with r ≥ c(θ) in the sliding region, the analytical torque curves in Fig. 4 labeled 'ana. Eq. (18)' are not valid consequences of the proposed friction law. The analytical errors of 2.18% and 2.83% are therefore expressed relative to a model that is inconsistently derived.","section":"Appendix B, Eq. (49)"},{"comment":"The parameters μub, as, bs, d, and Gb are determined by minimizing the mean relative error defined in Eq. (19) on the very same experimental curves that are later used to report the error values. The statement that the model reproduces the experimental curves 'with a relative error of less than 2.25%' is thus an in-sample calibration error, not a measure of predictive accuracy. With approximately five free parameters fitted to a single torque curve per data set, the close agreement is to be expected. The paper should state this explicitly and, if possible, provide a validation on an independent data set (e.g., one of the two datasets held out during calibration) to support the claim of predictive power.","section":"Sec. 4.1, Eq. (19), Table 4"},{"comment":"Because the contact pressure p is not measured and is instead generated by an imposed displacement d that is itself calibrated from the same torque data, the friction coefficients μub and μb are not individually identifiable; only the products μub p and μb p are determined. The paper acknowledges that no statement can be made about the accuracy of p, but it still reports μub values with three significant digits and interprets them as physiological parameters. The authors should discuss the non-identifiability more explicitly and perhaps report the results in terms of μ p or provide bounds on p from independent measurements.","section":"Sec. 4.5, Table 4"},{"comment":"The average initial osseointegration fraction for the first data set is given as φ̄0 = 0.73 in Sec. 4.3 (and used in Eq. (21) and Table 5), while Sec. 4.4 states that 'the sample in Fig. 7a is considered to have average osseointegration φ̄0 = 0.55' for the reconstructed pattern. This discrepancy is not explained and changes the computed values of the area-specific work of adhesion substantially. The authors need to clarify which value is used and why the same photograph yields two different estimates.","section":"Sec. 4.3 vs. Sec. 4.4"}],"minor_comments":[{"comment":"'Circular friction law' is a typo; it should likely read 'Coulomb friction law' or 'state-variable friction law'.","section":"Sec. 4.5"},{"comment":"The statement 'c≥r≥R denotes the sliding region' contains a reversed inequality; it should read 'c ≤ r ≤ R' since the critical radius c separates the sticking region r < c from the sliding region r ≥ c.","section":"Sec. 3.3"},{"comment":"The phrase 'relative error of less than 2.25%' should specify that this applies to the numerical solution; the analytical solution errors are 2.18% and 2.83%.","section":"Abstract"},{"comment":"The rows labeled 'ana. [32]' report errors for the reference analytical model; clarifying this in the caption would avoid confusion with the new analytical model.","section":"Table 4"},{"comment":"The sentence 'The model only depends on four physiological parameters (µub, p, as, and bs)' treats the contact pressure p as a parameter; p is a mechanical field rather than a physiological parameter, and µb is also fixed from literature.","section":"Sec. 4.4"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the scope of the journal and the numerical framework is well-executed, but the analytical inconsistency and the in-sample calibration caveat need to be resolved before publication. I would suggest the editor send the paper back for a major revision with explicit requests to correct Eq. (49), re-evaluate the analytical results, and clarify the calibration and identifiability issues."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The core idea here is worth taking seriously: a sliding-distance-dependent friction coefficient that smoothly degrades from an unbroken to a broken state, implemented in a NURBS-based contact FE framework. That gives computational implant mechanics something it lacks — a pointwise, partial-osseointegration-capable debonding law. The FE results genuinely match the two experimental torque curves after calibration, the mesh and load-step sensitivity study is competent, and the partial-osseointegration examples are a nice demonstration of the model's reach. Credit is also due for the Discussion section, which candidly lists the unmeasured contact pressure, the lack of a priori parameter determination, and the limited experimental basis.\n\nThe soft spots are real, though. The analytical derivation in Appendix B has a load-bearing error. Eq. (49) gives the sliding distance in the sliding region as g_s = r(θ − θ_max + θ_lin). For a point at radius r, sliding actually begins when c(θ) = θ_lin R/θ reaches r, i.e. at θ_start = θ_lin R/r, so the accumulated sliding distance is rθ − θ_lin R, not the paper's expression. The two coincide only in special cases. For data set 1, the paper's formula gives g_s = 0 at θ = 1° and r = R, and negative values for θ < 1°, which is unphysical because sliding at that radius starts at θ_lin = 0.13°. So the analytical torque curves labeled \"ana. Eq. (18)\" in Fig. 4 are not valid consequences of the proposed friction law, and the reported analytical errors of 2.18% and 2.83% should not be trusted. The FE model accumulates sliding directly and may be unaffected, but the paper's central claims explicitly include the analytical model.\n\nBeyond that, the calibration is entirely on the same two curves used to report error: all four friction parameters plus the bone shear modulus and the contact pressure are fitted from those curves. There is no independent validation, so the 2.25% error is a measure of curve fit, not prediction. The unmeasured contact pressure also weakens physical identifiability of the extracted parameters. The authors acknowledge most of this, which is fair, but it limits what the paper establishes.\n\nWho should read this? People working on bone–implant interface modeling, especially those needing a smooth debonding law for computational studies. It deserves a serious referee, but only with a major revision that fixes Eq. (49), re-runs the analytical model, and either validates on unseen data or clearly separates calibration claims from prediction claims. As it stands, the numerical model is a promising tool, but the analytical part is not correct.","headline":"The state-variable friction law for implant debonding is a useful idea and the FE implementation fits the data well, but the analytical model in Appendix B contains a clear kinematic error that invalidates the reported analytical torque curves.","tokens_in":22526,"tokens_out":3765,"would_cite":false,"duration_ms":38216,"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 proposes a modified Coulomb friction law whose coefficient decays with sliding distance, and shows it reproduces measured torque-versus-angle curves for osseointegrated implant debonding within 2.25% error.","keywords":["osseointegration","bone-implant interface","friction law","state variable friction","debonding","finite element method","mode III cleavage","adhesion energy"],"falsifier":"Measure the normal contact pressure independently during a torsion debonding test, or run the test under controlled zero normal load, and check whether the fully debonded torque equals $T_\\infty = \\frac{2}{3}\\pi R^3 \\mu_b p$ with the calibrated friction coefficient; a mismatch, or a nonzero peak under zero pressure, would indicate that the interface resistance is not purely Coulomb-type friction with the proposed sliding-distance decay.","tokens_in":21308,"feed_emoji":"🦴","tokens_out":4694,"duration_ms":47246,"temperature":0.7,"pith_summary":"The paper aims to predict debonding of osseointegrated implants by replacing the constant friction coefficient of Coulomb's law with one that depends on the accumulated sliding distance. The friction coefficient starts at a high, unbroken value and smoothly decays to a lower, broken value, capturing the transition from a bonded to a sliding interface. This modified law is implemented in a 3D finite element framework and tested against a mode III cleavage experiment of a coin-shaped titanium implant in bone. The authors show that the torque-versus-angle curve is reproduced within 2.25% relative error, and that the model yields better estimates of bone shear modulus and adhesion energy than the reference analytical model. If correct, the model offers a simple macro-scale description of interfacial debonding that can handle partial osseointegration, which is relevant to implant stability prediction.","feed_headline":"Friction law that decays with sliding matches implant torque to 2.25%","feed_subtitle":"A sliding-distance-dependent friction coefficient also extracts bone shear modulus and adhesion energy from torsion tests.","key_machinery":"The central object is the modified Coulomb law with a state-dependent friction coefficient, defined by $\\mu(\\varphi) = \\varphi \\mu_{\\mathrm{ub}} + (1-\\varphi)\\mu_{\\mathrm{b}}$ and the smooth state function $\\varphi(g_s) = \\varphi_0$ times a sinusoidal plateau–transition–zero profile in the normalized sliding distance $g_s/a_s$. This law converts Coulomb's friction into a local, history-dependent model that can represent a moving crack front and nonuniform bonding. It is implemented in a 3D nonlinear finite element framework with NURBS-enriched contact surfaces and also solved analytically for a radially symmetric mode III debonding, giving the traction distribution $\\sigma_{\\theta z} = t_{\\max}^t r/c(\\theta)$ in the sticking region and $\\sigma_{\\theta z} = \\mu p$ in the sliding region.","core_discovery":"The central claim is that a state-variable friction law of the form $\\mu(\\varphi) = \\varphi \\mu_{\\mathrm{ub}} + (1-\\varphi)\\mu_{\\mathrm{b}}$, with $\\varphi$ a smooth function of the accumulated sliding distance $g_s$, reproduces the full torque-angle response of an osseointegrated implant undergoing torsional debonding. The state function $\\varphi(g_s)$ stays at 1 up to a sliding threshold $a_s$, then decays sinusoidally to 0 over a transition zone set by $b_s$. This turns the friction coefficient into a local, history-dependent quantity that describes the progressive loss of bonding. Calibrated against two experimental data sets, the model matches the initial stiffness, peak torque, softening branch, and residual torque, with mean relative errors below 2.25%, and yields bone shear moduli of 7–8 GPa and Young's moduli of 18–21 GPa, which are consistent with literature values, unlike the reference analytical model. The model also shows that partial osseointegration changes the extracted parameters but that the torque curve alone cannot uniquely determine the degree or distribution of osseointegration without knowing the friction coefficients.","pith_inferences":["The sliding threshold $a_s$ (about 22–26 µm here) may correspond to the clinically reported micro-motion tolerance of the bone–implant interface, suggesting a direct link between the friction-state parameter and biological tolerance limits.","The same state-variable philosophy could be extended to other loading modes such as pull-out or push-in, where the accumulated sliding distance would again drive the transition from bonded to frictional response.","Since the pressure $p$ is inferred but not measured, a natural testable extension is to design an experiment with independently measured normal force; such a test would either validate the identified pressure or reveal that the unbroken resistance is dominated by adhesion rather than friction.","The success of a single scalar state variable suggests that complex micro-mechanical adhesion and interlocking processes can be effectively coarse-grained into a history-dependent friction coefficient, which could simplify patient-specific implant stability simulations."],"forward_implications":["The same set of four calibrated parameters ($\\mu_{\\mathrm{ub}}$, $p$, $a_s$, $b_s$) together with a literature value of $\\mu_{\\mathrm{b}}$ reproduces the entire torque curve, so the model can serve as a predictive tool for torsional debonding of planar implants.","The model extracts bone shear modulus and adhesion energy more accurately than the previous analytical model, giving values consistent with independent measurements of cortical bone.","Because the state variable is local, the model naturally handles partial and inhomogeneous osseointegration, which is common after short healing times.","The torque-angle curve alone does not uniquely determine the distribution of osseointegration when friction coefficients are unknown, implying that additional measurements are needed for full characterization.","Under perfect twisting, the sliding distance at the center of the implant remains zero, so that point never debonds; this is a prediction about the crack-front arrest that could be checked experimentally."],"supporting_citations":[{"why":"Provides the experimental mode III cleavage data and the reference analytical model that the new model is calibrated against and compared with.","marker":"[32]"},{"why":"Introduces state variable friction laws, the theoretical basis for making the friction coefficient depend on a history variable.","marker":"[40]"},{"why":"Develops slip instability and state variable friction laws, supporting the interpretation of the state variable as a bonding state.","marker":"[44]"},{"why":"Supplies the frictional contact formulation based on surface potentials and isogeometric discretization that the finite element implementation builds on.","marker":"[18]"},{"why":"Provides the NURBS-enriched contact finite element framework used for the numerical solution.","marker":"[12]"}],"fun_headline_variants":["Friction fades with sliding, matching implant torque to 2.25%","Sliding-distance friction law fits implant torque within 2.25%","A sliding-dependent friction coefficient predicts implant debonding","New friction model captures osseointegrated implant debonding","Sliding-based friction law reveals partial osseointegration effects"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The model assumes that the resistance of the osseointegrated interface is governed by Coulomb friction with a decaying friction coefficient, and that the normal pressure $p$ entering that law is real and identifiable from the torque data—but the experiments could not measure or eliminate $p$, and the intact state is treated as high-friction rather than as an adhesive or cohesive interface.","fun_headline_variants_meta":{"raw":{"variants":["Friction fades with sliding, matching implant torque to 2.25%","Sliding-distance friction law fits implant torque within 2.25%","A sliding-dependent friction coefficient predicts implant debonding","New friction model captures osseointegrated implant debonding","Sliding-based friction law reveals partial osseointegration effects"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001398,"raw_usage":{"total_tokens":5701,"prompt_tokens":1041,"completion_tokens":4660,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":657,"completion_tokens_details":{"reasoning_tokens":4568}},"tokens_in":657,"tokens_out":4660,"duration_ms":35942,"temperature":1.0,"reasoning_tokens":4568,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:51:55.983798+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the normal contact pressure independently during a torsion debonding test, or run the test under controlled zero normal load, and check whether the fully debonded torque equals $T_\\infty = \\frac{2}{3}\\pi R^3 \\mu_b p$ with the calibrated friction coefficient; a mismatch, or a nonzero peak under zero pressure, would indicate that the interface resistance is not purely Coulomb-type friction with the proposed sliding-distance decay.","supporting_citations":[{"cited_title":"Mathieu, R","cited_arxiv_id":null,"evidence_quote":"Provides the experimental mode III cleavage data and the reference analytical model that the new model is calibrated against and compared with."},{"cited_title":"Rice and A","cited_arxiv_id":null,"evidence_quote":"Introduces state variable friction laws, the theoretical basis for making the friction coefficient depend on a history variable."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Develops slip instability and state variable friction laws, supporting the interpretation of the state variable as a bonding state."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the frictional contact formulation based on surface potentials and isogeometric discretization that the finite element implementation builds on."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the NURBS-enriched contact finite element framework used for the numerical solution."}],"review_version":1}