{"id":"90e07486-ffb2-4e26-bb42-e00fc6ad68a1","arxiv_id":"2607.11176","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"The generalized Frank-Bilby equation, b(p)=T p with T=([[F^-1]], n_R^T), and its quantized form BN^T=T∥ predict interfacial disconnection networks with both Burgers-vector and step-height content.","lead":"The paper derives an extended Frank-Bilby equation that adds step height to Burgers-vector content, so interface defects ('disconnections') can be predicted as arrays with both dislocation and step character. It applies the framework to twin boundaries, Au-Pd twist interfaces, and titanium martensite faces, matching several known results and experiments.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Reference-state selection is not unique: the paper's stated heuristic fails to pick the experimentally consistent Au-Pd reference, and martensite cases require ad hoc criteria, undermining the claim that GFBE predicts interface structure without extra input.","rationale":"The reader's weakest assumption — that the method's predictive power rests on a non-unique reference-state choice and post-hoc selection rules — is exactly the most load-bearing concern I find. The paper itself demonstrates the ambiguity: for Au-Pd, the heuristic does not select the reference that matches experiment, and for martensite the solution is a manifold requiring an extra criterion. The final paragraph's limitation to DSC disconnections and exclusion of reactions further restricts the scope, but the reference-state ambiguity is more immediate because it affects even the simplest examples. The derivation of the GFBE equations themselves appears algebraically consistent; the weak link is the interpretive step from 'a geometrically possible network' to 'the predicted interface structure.' A concrete numerical re-computation using the heuristic-selected RS would settle whether the predicted network is unique and experimentally valid. Since this is the same concern the reader raised, my verdict is unchanged.","tokens_in":30403,"tokens_out":8214,"duration_ms":74772,"concrete_test":"Recompute the Au-Pd (001) twist-interface predictions using only the reference state selected by the paper's own heuristic (minimal ||JF^-1K|| and minimal inclination). For the 9:10 coherent reference, evaluate the qGFBE d(θ) and ω(θ) curves from Eqs. (71)-(72) and overlay them on the experimental data in Fig. 2c,d. If the heuristic-selected curves do not match the black crosses within experimental scatter while the 1:1 curves do, the reference-state ambiguity is unresolved and the central predictive claim fails as stated. A secondary check: repeat the broad-face martensite calculation with each of the three selection rules and quantify the spread in θ, φ, and d^(1); if the spread exceeds the experimental uncertainty, the selection-rule dependence is significant.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central predictive claim — that the GFBE (Eq. 3) and its quantized form (Eq. 6) determine interface structure from lattice mismatch, reference normal, and DSC modes alone — depends critically on a unique, physically meaningful choice of the reference state (RS). The paper's heuristic (close-packed CSL plane, minimal ||JF^-1K|| and minimal inclination) is not proved to be unique or correct. In Sec. III, the 1:1 and 9:10 references for the Au-Pd (001) twist interface both have zero inclination; the 9:10 reference has markedly smaller in-plane strain and hence smaller ||JF^-1K||, so the stated heuristic selects 9:10. Yet the experimental data (Fig. 2e) are reported consistent with the 1:1 reference. Thus the selection rule as written points to the reference that experiment excludes. In the martensite broad-face case (Sec. IV B), the qGFBE solution is a one-dimensional manifold; the paper imposes an un-derived rule (min Q, min θ, max d^(2)) to choose a point, and different rules yield different geometries (Figs. 3d,e). The final paragraph acknowledges an additional domain restriction: the method excludes disconnection reactions. Together, these gaps mean the GFBE can rationalize a known network but does not, as presented, uniquely predict it from geometry alone.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper derives a generalized Frank-Bilby equation (GFBE) that augments the classical Burgers-vector balance b(p)=JF^{-1}K p with a step-height balance h(p)=n_R·p, giving a 4-vector incompatibility tensor T. Quantization via DSC disconnection modes yields BN^T=T_∥ (Eq. 6), which determines line directions and spacings of disconnection arrays. Applications are presented to misoriented Σ3 twin boundaries (recovering Frank's formula), Au-Pd (001) twist interfaces (recovering the Jesser-Kuhlmann-Wilsdorf result and predicting four-set disconnection networks), and α/β martensite broad and side faces (matching PTMC and experiments). The stated aim is to predict interface structure from lattice geometry, reference state, and DSC modes without invoking energy or kinetics.","tokens_in":30814,"tokens_out":8388,"duration_ms":80590,"significance":"The step-height generalization is a natural and potentially important extension of the Frank-Bilby formalism. The compact 4D formulation unifies dislocation and step content and, as shown, reduces to several known geometric theories (Frank's formula, Jesser-Kuhlmann-Wilsdorf, PTMC) in appropriate limits. The agreement with atomistic simulations (CTB case) and with experiments (Au-Pd, martensite) is encouraging evidence that the core derivation is correct. The paper also provides closed-form analytical predictions and a clear linear-algebra procedure for the quantized problem, which should be useful to the interface community if the caveats below are properly addressed.","major_comments":[{"comment":"The qGFBE solution for the broad face is a one-dimensional manifold, and the paper selects a point using min Q, min θ, or max d^(2) — criteria not derived from the GFBE. Different criteria give different geometries (Figs. 3d, e, g). This directly conflicts with the abstract/introduction claim that interface structure is predicted 'without invoking energetic or kinetic information,' since Q is an energy measure. The authors should either derive a selection rule from the geometric framework or explicitly restate the contribution as providing necessary geometric conditions, with a separate physical criterion needed to close the problem.","section":"Sec. IV B and C (Eqs. 92–93, Fig. 3)"},{"comment":"The reference state is non-unique, and the stated heuristic (close-packed CSL plane, minimal norm of JF^{-1}K, minimal inclination) is not formulated as a well-defined algorithm over the infinite set of possible coincidence references. Predictions depend quantitatively on the chosen reference (color-coded curves in Fig. 2), and the manuscript does not specify how the minimization is carried out or why the chosen reference is the one realized experimentally. This is load-bearing for the claimed predictive power. Please clarify the selection procedure and its uniqueness, or present the GFBE as conditional on a reference state chosen by external physical arguments.","section":"Sec. III, reference-state construction (p. 2, Figs. 2c–e)"},{"comment":"The paper acknowledges that the GFBE excludes disconnection reactions and assumes a coherent CSL/DSC terrace reference. This is a substantial domain restriction for real interfaces. Since the title and abstract promise a general framework, the limitations should be stated more prominently and the practical conditions under which reactions are negligible should be identified. As written, the scope is narrower than the abstract suggests.","section":"Final paragraph, Sec. IV C"}],"minor_comments":[{"comment":"The notation ΣAu:ΣPd is used for heterophase coincidence ratios; a brief definition in the main text would help readers not familiar with this convention.","section":"Eq. (4) and Table I"},{"comment":"The relative position of the two disconnection sets is stated to be undetermined by qGFBE and 'minimizing the elastic energy suggests d(1)/2.' This is an example of the energy-dependence issue; it would be useful to state explicitly that such a determination is outside the geometric framework.","section":"Sec. II (CTB case)"},{"comment":"The expression for ω(1) contains an arctangent with a denominator that can approach zero; specifying the branch or giving an equivalent principal-value expression would improve clarity.","section":"Eq. (65)"},{"comment":"The subscripts 'Au' and 'Pd' in Eq. (69) are defined in the text but the notation ¯a_Au is introduced abruptly; a sentence defining all barred lattice parameters would help.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The core derivation appears sound and the paper is likely a useful contribution, but the predictive claim is currently overstated relative to the actual role of the selection criteria and reference-state choice. The revision should either supply a rigorous selection principle or carefully delimit the claim to conditional geometric constraints. I see no fatal technical error in the main equations."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First thing to know: this paper derives a generalized Frank-Bilby equation that puts Burgers vector and step height on the same footing, and the quantized form BN^T=T∥ is a genuinely useful extension. It cleanly reproduces Frank's formula, the Jesser-Kuhlmann-Wilsdorf result, and PTMC as special cases, and the martensite applications are new. The derivation is careful and self-contained; the math checks out as far as I can tell.\n\nNow the soft spots, in proportion. The central claim is that the GFBE predicts interface structure from geometry alone, but that requires a unique reference state. The paper offers a heuristic—close-packed CSL plane, minimal ||JF^-1K|| and minimal inclination—but it doesn't always pick the physically right one. In the Au-Pd twist case, both 1:1 and 9:10 references have zero inclination; 9:10 has smaller in-plane strain and hence smaller ||JF^-1K||, so the heuristic selects 9:10. Yet the experimental data the authors cite are consistent with the 1:1 reference. The paper doesn't hide this—it shows both predictions and notes the experiment matches 1:1—but that means the stated selection rule is not reliable. If the reference state is part of the input, the GFBE can rationalize a network, but it isn't uniquely predicting it from geometry alone.\n\nThe martensite broad-face case has a similar issue: the qGFBE solution is a one-parameter family, and the paper needs an extra criterion (min Q, min θ, or max d^(2)) to select a point. They are transparent about this, but it further narrows the \"no energy/kinetics needed\" claim.\n\nTwo more things. The relation to prior generalized FBE treatments—especially Hirth, Pond, Hoagland et al. 2013—is not clearly demarcated; the step-height concept and CSL/DSC disconnection modes are largely present there. And the limitation to DSC disconnections, with disconnection reactions excluded, is acknowledged in the final paragraph, but it's a real restriction on the generalization.\n\nThat said, the paper deserves a serious referee, but it should be sent back for substantial revision: either give a principled, validated way to choose the reference state, or tone down the uniqueness claim and present the GFBE as a flexible framework that requires physical input to close. I'd also ask for code or at least the numerical details for the Au-Pd and martensite plots. Bottom line: worth reading, worth citing, but treat the predictive claim with caution until the reference-state problem is addressed.","headline":"The generalized Frank-Bilby equation is a genuine extension that puts Burgers and step content on the same footing, but the paper's predictive promise is undercut by non-unique reference-state selection, and its own Au-Pd example contradicts the stated heuristic.","tokens_in":31261,"tokens_out":3234,"would_cite":true,"duration_ms":29324,"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":"This paper derives a generalized Frank-Bilby equation that tracks both Burgers-vector and step-height content of interfacial line defects, and argues that for most crystalline interfaces the resulting quantized equation BN^T=T∥ predicts the","keywords":["Frank-Bilby equation","disconnections","step height","interface structure","coincidence-site lattice","DSC lattice","martensite crystallography","heterophase interfaces"],"falsifier":"For a given interface, compare the predicted disconnection spacings and line directions for two different reference states (e.g., Au-Pd twist interface with Σ=1:1 vs Σ=9:10). If neither matches electron microscopy observations, or if the observed network requires a disconnection reaction outside the DSC set, the 'geometry alone' claim fails.","tokens_in":1621,"feed_emoji":"📐","tokens_out":2692,"duration_ms":101910,"temperature":0.7,"pith_summary":"This paper argues that the structure of a crystalline interface can be predicted from geometry alone, as long as the interfacial line defects are treated as disconnections—defects carrying both a Burgers vector and a step height—rather than as pure dislocations. The authors derive a generalized Frank-Bilby equation (GFBE) that combines the classical Burgers-vector balance b(p) = JF^{-1}K·p with an analogous step-height balance h(p) = n_R·p, producing a single extended equation b(p) = T p. Quantizing against the coincidence-site/DSC lattice turns this into a linear system BN^T = T∥ whose solution gives the spacing, line direction, and step content of every disconnection array on the interface. In the cases worked out—a misoriented coherent twin boundary, Au-Pd twist heterophase interfaces, and the broad and side faces of martensite nuclei—the predictions match atomistic simulations, experiments, and, as limiting cases, Frank's formula, the classical twist-interface result, and the phenomenological theory of martensite crystallography. If the formulation is right, then for many interfaces the disconnection network is fixed by lattice mismatch and reference-state choice alone, with no energy or kinetics required.","feed_headline":"Step height joins Burgers vector to fix interface structure","feed_subtitle":"A generalized Frank-Bilby equation predicts dislocation-step arrays from crystallography alone, matching experiments and simulations.","key_machinery":"The load-bearing object is the extended incompatibility tensor T = (JF^{-1}K ; n_R^T), where JF^{-1}K is the jump in inverse deformation gradients between the two crystals and n_R is the reference-interface normal. Stacking the step-height equation below the Burgers-vector equation turns the interface into a four-component 'Burgers-step' space, and quantizing with the CSL/DSC lattice yields the matrix equation BN^T = T∥. N carries the physical network: each column η^(m) points normal to a disconnection array, with magnitude equal to the reciprocal spacing. The rank structure of B and T∥ decides whether the network is uniquely determined, underdetermined, or impossible with the chosen modes.","core_discovery":"The central claim is that a complete geometric description of an interface requires both a Burgers vector and a step height, and that both are governed by one linear equation. The generalized Frank-Bilby equation (GFBE) reads b(p) = T p for every in-plane probe vector p, where b = (b, h) is the extended Burgers-step vector and T = (JF^{-1}K ; n_R^T) is the incompatibility tensor built from the jump in inverse deformation gradients and the reference-interface normal. Quantizing on the CSL/DSC lattice gives the qGFBE: BN^T = T∥, with B listing admissible disconnection modes and N containing the reciprocal spacing vectors of the unknown arrays. When the rank condition rank([B T∥]) = rank(B) = M","pith_inferences":["If the GFBE is right that geometry fixes the network, then the reference state is not a mathematical convenience but a physical property of the interface; the Au-Pd case, where different reference choices predict different arrays, offers a way to infer which coincidence-site construction a real interface adopts by comparing measured spacings with predictions.","The step-height balance could be embedded as a conservation law in phase-field or continuum models, so that the predicted disconnection network serves as an initial condition or constraint for relaxations that include energy and kinetics—potentially extending the method to the honeycomb networks the paper excludes.","For underdetermined martensite cases, the one-dimensional solution manifolds imply that observable habit-plane data can be read as a selection experiment: measuring which facet forms effectively identifies whether the system minimizes the disconnection-energy measure Q, the deviation from the Burgers orientation relationship, or the mode-2 spacing."],"forward_implications":["For any interface where the rank condition holds, the disconnection network—line spacings, line directions, Burgers vectors, and step heights—is uniquely determined by lattice mismatch, reference normal, and the admissible CSL/DSC modes, with no energy calculation.","The misoriented coherent twin boundary prediction reduces to Frank's formula, d = a/(√3 sin(Δθ/2)), and agrees with atomistic simulations for Ni.","Twist heterophase Au-Pd interfaces require four sets of disconnections with finite step heights at higher-order coincidence orientations (e.g., ΣAu:ΣPd = 9:10), content that the classical FBE cannot represent; the predicted network energy measure Q has no cusp at coherent misorientations.","For the martensite habit plane, the qGFBE predictions coincide with the phenomenological theory of martensite crystallography but additionally give the disconnection network and remain valid when no invariant plane exists.","The broad and side faces of martensite nuclei are underdetermined by geometry alone, yielding one-dimensional solution manifolds; the paper's selection criteria (min Q, min deviation from the Burgers orientation relationship, max mode-2 spacing) identify different experimentally observed facets.","The method explicitly does not describe interfaces with disconnection reactions, such as the honeycomb dislocation networks observed in twist grain boundaries."],"fun_headline_variants":["Generalized Frank-Bilby equation adds step height to dislocations","From Burgers vectors to step heights: interface structure now fully described","One equation now captures both dislocation and step defects","New Frank-Bilby theory predicts dislocation-step arrays","Including steps in Frank-Bilby: a complete interface theory"],"cache_read_input_tokens":32512,"weakest_assumption_plain":"The method assumes a real interface can be represented as a coherent reference terrace decorated only by quantized DSC disconnections, with a reference state chosen by a heuristic (close-packed CSL plane, minimal mismatch and inclination) and, in underdetermined cases, by an additional selection rule; the paper itself concedes that disconnection reactions (e.g., honeycomb twist networks) fall outside this description.","fun_headline_variants_meta":{"raw":{"variants":["Generalized Frank-Bilby equation adds step height to dislocations","From Burgers vectors to step heights: interface structure now fully described","One equation now captures both dislocation and step defects","New Frank-Bilby theory predicts dislocation-step arrays","Including steps in Frank-Bilby: a complete interface theory"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000209,"raw_usage":{"total_tokens":1191,"prompt_tokens":641,"completion_tokens":550,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":385,"completion_tokens_details":{"reasoning_tokens":476}},"tokens_in":385,"tokens_out":550,"duration_ms":5078,"temperature":1.0,"reasoning_tokens":476,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T06:58:59.845061+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For a given interface, compare the predicted disconnection spacings and line directions for two different reference states (e.g., Au-Pd twist interface with Σ=1:1 vs Σ=9:10). If neither matches electron microscopy observations, or if the observed network requires a disconnection reaction outside the DSC set, the 'geometry alone' claim fails.","supporting_citations":[],"review_version":2}