{"id":"d4252109-ee3d-46bd-b1c1-72bd815aff87","arxiv_id":"2608.02400","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"GRADAR computes which grains of an EBSD-mapped polycrystal can reach a nominated two-beam ECCI condition and ranks them by predicted clean contrast.","lead":"A new computational method, GRADAR, reads a polycrystalline sample's orientation map and computes which grains can be brought into the precise two-beam diffraction condition that dislocation imaging needs, and at which stage settings. It replaces a per-grain manual search with a ranked session plan; the geometry was checked on a published silicon measurement.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Ranking premise is unvalidated: darkest-clean selection rests on internal r=+0.40 and a single-crystal check; per-grain polycrystal validation is deferred, so the session-plan output lacks experimental support.","rationale":"The reader identified the same load-bearing concern: the ranking premise is structurally distinct from the geometric reachability claim and is unvalidated on a polycrystal. My review of the paper confirms that the reachability criterion (eq. 2) follows directly from the cone geometry and is checked against the published Si precession series, so it is robust. The ranking step, however, depends on a decomposition of intensity into 'own band' and 'rival contamination' that is supported only by an internal correlation and a single-crystal check. The paper openly defers per-grain validation. This makes the instrument's headline output — ranked per-grain two-beam conditions for a polycrystal — conditional on future measurement. The reader's CONDITIONAL verdict is appropriate; I see no additional concern that would change it, and no reason to downgrade the geometric contribution.","tokens_in":13610,"tokens_out":7092,"duration_ms":81079,"concrete_test":"Run the deferred per-grain polycrystal validation on a measured EBSD map of a polycrystal (e.g., the 163-grain austenitic steel map): select at least 20 grains that each have at least three {111} two-beam candidates spanning the clearance range from <1° to >5°. For each candidate, acquire a BSE image or SA-ECP intensity trace at the predicted stage rotation at fixed tilt (7°). Score each condition by (a) measured channeling intensity (normalized against the grain's rotation series) and (b) dislocation contrast-to-noise ratio for a known dislocation population. Then test whether the GRADAR-selected darkest-clean candidate is darker and yields better contrast than the raw darkest candidate and than a high-clearance random candidate across the grain set. If the ordering is not reproduced, the ranking premise fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The geometric core of the paper — the reachability criterion of eq. (2) — is simple, parameter-free, and convincing. The load-bearing weak point is the ranking step that turns reachability into a session plan. The paper claims (Section 3, Rank) that at exact Bragg incidence all same-family candidates are equivalent with respect to their own band, so that the remaining intensity differences are dominated by rival-band contamination, and that the darkest candidate inside the clearance gate is the best two-beam condition. This premise is supported only by (i) an internal correlation r=+0.40 between predicted intensity and rival clearance computed on the same map's candidates, and (ii) the single-crystal Si check where measured intensity is dark at 16 predicted edges. Neither tests the ranking across a polycrystal. The paper itself explicitly defers per-grain ranking validation to a companion study (Section 6.2). If this premise is wrong — e.g., if the own-band contribution is not actually identical for all same-family candidates because of absorption, boundary conditions, or surface orientation effects, or if the clearance floor max(0.5°, θ_B) does not correspond to the angular scale on which rival contamination degrades dislocation contrast — then the ranked candidate list and the resulting stage moves could select suboptimal grains. This is a disclosed limitation, not a hidden flaw, but it is the least secure link between the proven geometry and the applied instrument claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents GRADAR, a computational procedure that takes an EBSD orientation map of a polycrystal, a nominated reflection family, and a stage tilt/rotation envelope, and determines which grains can be brought to a two-beam electron channeling condition for that family, then ranks the survivors. The central geometric result is eq. (2): at fixed stage tilt t, stage rotation sweeps the incident beam along a cone of half-angle t in each grain's crystal frame, so a plane normal with polar angle α from the specimen normal reaches a Bragg angle θ_B iff α lies within [90°−(t+θ_B), 90°+(t+θ_B)]. This criterion is applied to a measured 163-grain austenitic steel map at 20 kV, producing reachability fractions that vary strongly with tilt and family ({111} rises from 38.7% at 7° to 77.3% at 15°; {220} and {311} reach 79.8% and 91.4% at 7°). A 'darkest-clean' ranking with a rival-clearance gate (eq. 4) is proposed to select among reachable candidates. The predicted precession sweep is checked against the published AstroECP silicon series: the simulated trace correlates at r=0.84 with the SA-ECP centre trace, and the measured intensity is dark at all 16 predicted two-beam edges. Per-grain polycrystal validation of the ranking is explicitly deferred to a companion study.","tokens_in":13929,"tokens_out":9936,"duration_ms":109066,"significance":"If the reachability result is correct, it closes the forward question posed by L'hôte et al. in 2019 and turns map-wide ECCI session planning into a computed procedure: the operator can nominate a g vector and immediately see which grains of a mapped polycrystal can serve it, and at what stage tilt and rotation. The eq. (2) criterion is parameter-free and hand-checkable; the reachability tables are deterministic computations on a public dataset with a disclosed random-texture baseline; the implementation is reproducible via shipped notebooks, with two independent implementations agreeing to <0.006° in crossing angles. The experimental check on published silicon data is genuine and anchors the sweep geometry at two levels (orientation fitting and intensity correlation). The main weakness is that the ranking step—the 'R' in GRADAR—is validated only indirectly: an internal correlation of r=+0.40 on the same map's candidates and darkness at 16 predicted edges in a single crystal. Neither test demonstrates that the darkest-clean candidate is the best two-beam condition in a polycrystal. This is disclosed in §6.2, but it is the load-bearing link between the proven geometry and the applie","major_comments":[{"comment":"The ranking premise—that at exact Bragg incidence all same-family candidates are equivalent with respect to their own band, so intensity differences are dominated by rival contamination—is not theoretically established. In the two-beam dynamical theory, the tie point and branch excitation at sg=0 depend on the boundary conditions at the specimen surface, which vary along the Kossel cone as the orientation of the diffracting planes relative to the surface changes; the 'own-band' contribution may therefore not be identical for all candidates. The paper's empirical support is the internal correlation r=+0.40 (§4) and the single-crystal dark-at-16-edges check (§5). Neither tests the ranking across a polycrystal. Because the ranked candidate table is the main output of the tool, this is load-bearing. The manuscript should either derive the equivalence from a suitable dynamical model, present","section":"§3 Rank; eq. (4)"},{"comment":"The experimental validation is confined to a single [001] silicon crystal at ~7° tilt and does not exercise the ranking across grains. The r=+0.40 correlation in §4 is computed on 244 candidates from 163 grains, with multiple candidates per grain, so the effective number of independent observations is smaller and within-grain clustering is not accounted for; moreover, a correlation with rival clearance only supports the contamination gradient, not the correctness of the darkest-clean choice. A dedicated experiment comparing predicted dark-clean candidates with measured dislocation contrast in several grains of a polycrystal is necessary to support the session-plan output. As written, the paper's strongest validated contribution is the reachability geometry; the ranking contribution remains unsupported.","section":"§5; Table 2; §6.2"}],"minor_comments":[{"comment":"The random-texture baseline paragraph says 'The exact isotropic value sits between the two' (single-plane probability and union bound). Please give the actual Monte-Carlo value in the text for {111} at 7°, and add the union-bound row to Table 1 or state it explicitly; the current wording is slightly unclear.","section":"§4, Table 1"},{"comment":"The BSE trace comparison uses a linear detrend and a −3° lag, but neither is visible in Figure 8. Please mark the detrend and the lag in the figure or its caption so that the reader can see what was adjusted.","section":"§5, Figure 8"},{"comment":"The notation γ_r = |90° − angle(k, g_r)| is understandable but the subscript braces and the absolute value could be annotated. Also, 'kinematically forbidden rivals are dropped by structure factor' should specify whether the structure-factor cut-off is user-configurable.","section":"§3, eq. (4)"},{"comment":"The column header 'gRotation (deg)' is ambiguous; consider splitting it into 'g' and 'Stage rotation (°)' for clarity.","section":"Table 2"},{"comment":"The statement that the code is 'available from the author while the public release archive is being prepared' is at odds with the reproducibility promise. A permanent DOI or repository link would make the reproducibility claim fully checkable.","section":"§6.3"}],"recommendation":"major_revision","confidential_remarks":"The reachability core is solid, well-presented, and genuinely useful; eq. (2) is a clean result that should be published. The ranking gap is the only serious issue. I would encourage the editor to request a revision that either adds a small but direct polycrystal validation of the ranking (even a few grains) or substantially softens the ranking claim in the title, abstract, and conclusions; the manuscript's own deferred-validation statement already acknowledges the gap, but the framing still presents the ranked session plan as established. A major revision, not rejection, is appropriate."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The genuinely new thing here is the forward inversion: nominate a reflection family and a stage envelope, then ask which grains of an EBSD map can reach that condition. Every prior tool I know solves the inverse problem per grain. The geometric core, eq. (2), is correct and checkable by hand, and the gearing relation (dβ = sin t dφ) is a nice observation that gives real precision from a cheap rotation stage. The computed reachability tables on the public openECCI map, with a random-texture baseline, are reproducible in principle, and the paper explicitly flags the RANSAC-seed caveat that moves the grain count by a few points. That is the kind of transparency I want to see.\n\nThe single-crystal silicon check is genuine partial evidence: the simulated sweep tracks the measured SA-ECP trace at r = 0.84, and the intensity is dark at all 16 predicted two-beam edges. That validates the sweep geometry and the darkness property at edges. But the paper's applied output — a ranked session plan for a polycrystal — rests on a much weaker footing. The darkest-clean ranking premise is supported only by an internal correlation of r = +0.40 on the same map's candidates and by that single-crystal check. No per-grain polycrystal measurement tests the ranking. The paper admits this plainly in Section 6.2 and defers the closure to a companion study, so it is not a hidden flaw. Still, it is the load-bearing link between the proven geometry and the instrument claim. If the own-band equivalence fails in real polycrystalline conditions, or the clearance floor is the wrong angular scale, the ranked list could select suboptimal grains. I also note the code is not yet public with a commit hash, though the reproduction notebooks are promised.\n\nMy read agrees with the stress-test note: the geometric core holds up, the ranking premise is the soft spot, and it is disclosed rather than fatal. The paper deserves a serious referee because the forward problem is important to the ECCI community and the central derivation is solid. Who is this for? SEM-based defect imagers and tools developers, plus anyone thinking about automating ECCI. I would cite the geometry now and wait for the companion study before relying on the ranking. Send it to peer review, with the expectation that the per-grain validation and public artifacts are required before the applied claims are accepted.","headline":"The forward reachability geometry is solid, simple, and worth referee time; the headline ranking output is honest but unvalidated on polycrystals, so treat the session-plan claim as provisional.","tokens_in":14462,"tokens_out":1452,"would_cite":true,"duration_ms":17470,"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":"A geometric cone criterion decides which grains of a polycrystal can reach a nominated two-beam condition, and ranking by darkness alone selects the wrong candidates.","keywords":["electron channeling contrast imaging","two-beam condition","EBSD orientation map","reachability criterion","stage rotation","clearance ranking","grain targeting","channeling simulation"],"falsifier":"A decisive check: on a polycrystalline sample, measure actual dislocation-channeling contrast at the method-selected condition for each grain and compare darkest-clean versus globally darkest candidates; if per-grain contrast does not favor the darkest-clean selection, or if a grain predicted unreachable at a given tilt shows a usable two-beam band edge after all, the ranking or the reachability criterion fails.","tokens_in":13452,"feed_emoji":"🔬","tokens_out":6077,"duration_ms":56848,"temperature":0.7,"pith_summary":"This paper claims that the tedious per-grain hunt for a two-beam diffraction condition in electron channeling contrast imaging (ECCI) can be inverted: the operator names one reflection family and a stage envelope, and a computed sweep over an EBSD orientation map returns every grain that can reach that family's condition, with the stage rotation and tilt to dial. The reachability test reduces to a two-line inequality: at fixed stage tilt t, the plane normal of a reachable reflection must lie within (t + θ_B) of the specimen surface. On a measured 163-grain austenitic steel map at 20 kV, this makes {111} reachability rise from 38.7% at 7° tilt to 77.3% at 15°, while {220} and {311} already reach 79.8% and 91.4% at 7°. The paper further claims that among reachable grains, picking the darkest candidate is wrong—same-family candidates excite their own band identically at Bragg incidence, so darkness tracks rival-band contamination; the method instead gates candidates on a minimum angular clearance from all rival reflections and ranks the darkest survivor. If true, high-throughput defect studies would switch from per-grain manual alignment to a computed session plan under one declared diffraction vector.","feed_headline":"Predict which grains can hit a chosen two-beam condition","feed_subtitle":"On a 163-grain steel map, {111} reachability doubles from 38.7% at 7° to 77.3% at 15° tilt.","key_machinery":"The central object is the rotation cone: at fixed stage tilt t, stage rotation φ sweeps the incident beam direction around a cone of half-angle t about the specimen normal, in each grain's crystal frame. The reachability criterion is an annulus requiring a plane normal within (t + θ_B) of the specimen surface; it reduces the targeting problem to a two-line geometric check. The ranking mechanism is the clearance gate: for each candidate at a two-beam edge of the nominated family, compute the minimum angular distance from any rival reflection to its own Bragg condition, discard candidates below max(0.5°, θ_B), and rank the rest by predicted channeling darkness. The gearing relation dβ = sin t","core_discovery":"At fixed stage tilt t, rotating the stage sweeps the incident beam around a cone of half-angle t in each grain's crystal frame. A reflection family with Bragg angle θ_B reaches its two-beam condition exactly when one of its plane normals lies within (t + θ_B) of the specimen surface; this is an annulus on the orientation sphere. The paper shows on a measured 163-grain map that this geometric criterion, not texture intuition, governs usability: {111} reachability jumps from 38.7% at 7° tilt to 77.3% at 15°, and the higher-multiplicity {220}/{311} families outperform {111} at low tilt. For ranking, all same-family candidates are equivalent with respect to their own band at exact Bragg incidenc","pith_inferences":["Editorial inference: the reachability inequality is purely geometric and energy-dependent only through θ_B, so the same annulus criterion should transfer to other channeling-based modalities, such as TEM two-beam settings or other crystal systems, with the stage model modified.","Editorial inference: because the sweep depends on one reference orientation plus stage kinematics, the method could be extended to a live feedback loop that recomputes candidates from a locally updated orientation after each stage move, correcting map calibration drift during the session.","Editorial inference: the reported positive correlation between predicted intensity and rival clearance suggests a testable meta-prediction: measured per-grain contrast should be higher for darkest-clean candidates than for the globally darkest ones, which the companion polycrystal validation can settle directly.","Editorial inference: the random-texture baseline implies that for weakly textured specimens, reachability fractions are near-universal functions of family, energy, and tilt, so session planning could be precomputed from material class alone before any map is taken."],"forward_implications":["Map-wide ECCI session planning becomes a computed procedure: nominate a diffraction vector and a stage envelope, and get a ranked candidate table with dial-ready stage moves instead of per-grain manual searches.","Reachability fractions are strongly stage-tilt dependent: on the measured map at 7° tilt, {111} serves only 38.7% of grains while {220} and {311} serve 79.8% and 91.4%, guiding family choice before the session.","The clearance gate implies that scientifically usable candidates are not the darkest; the optimal floor trades contamination margin against yield and darkness, an experimental parameter left to validation.","The gearing result means sub-0.1° beam alignment is attainable on an ordinary rotation stage at low tilt, meeting the sub-half-degree precision demand without specialized channeling hardware.","The inverse formulation allows statistically representative defect populations to be collected under one declared condition across many grains, enabling comparisons across processing or testing histories."],"fun_headline_variants":["Tilt sweeps beam cone; geometry picks two-beam grains","Reachability jumps from 38.7% to 77.3% for {111} at 15° tilt","GRADAR ranks darkest-clean grains, avoiding rival Bragg edges","Inverse ECCI: choose a reflection, get all usable grains"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The ranking rests on the assumption that every member of the chosen reflection family produces identical channeling darkness at its own Bragg condition, so differences between candidates come only from nearby competing reflections; this premise has not yet been verified grain-by-grain on a polycrystal.","fun_headline_variants_meta":{"raw":{"variants":["Tilt sweeps beam cone; geometry picks two-beam grains","Reachability jumps from 38.7% to 77.3% for {111} at 15° tilt","GRADAR ranks darkest-clean grains, avoiding rival Bragg edges","Inverse ECCI: choose a reflection, get all usable grains"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000428,"raw_usage":{"total_tokens":2137,"prompt_tokens":969,"completion_tokens":1168,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":713,"completion_tokens_details":{"reasoning_tokens":1082}},"tokens_in":713,"tokens_out":1168,"duration_ms":13590,"temperature":1.0,"reasoning_tokens":1082,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-04T07:54:50.560293+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A decisive check: on a polycrystalline sample, measure actual dislocation-channeling contrast at the method-selected condition for each grain and compare darkest-clean versus globally darkest candidates; if per-grain contrast does not favor the darkest-clean selection, or if a grain predicted unreachable at a given tilt shows a usable two-beam band edge after all, the ranking or the reachability criterion fails.","supporting_citations":[],"review_version":1}