{"id":"0330e040-fecd-495f-a9c4-5127765820d5","arxiv_id":"2601.03581","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"In a gradient-annealed Fe-based glassy ribbon, the anomalous Ettingshausen (transverse heat) response peaks before crystallization, at positions containing 1.7-nm Cu nanoclusters — a variation invisible to XRD and longitudinal transport.","lead":"Heated unevenly, a glassy metal ribbon showed its strongest heat-to-electricity conversion just before it began to crystallize — a 'sweet spot' that ordinary X-ray and conductivity measurements completely miss. The result, mapped with an infrared lock-in camera on a single graded sample, points to nanometer-scale copper clusters as the cause.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central claim — maximum AEE 'well before crystallization' — rests on an unmeasured annealing-temperature gradient; §2.1 and Methods admit the profile is assumed and 'may not be strictly linear,' so the x→Ta mapping and the peak's temperature-derived structural meaning are unverified.","rationale":"The reader's weakest assumption is also, in my reading, the single most load-bearing concern. The non-monotonicity of Aodd is not in serious doubt: even with the electrode-gap at x ≈ 75 mm and no error bars, the flanking data (4.90 mK at x = 0, 6.65 mK near the gap, 2.29 mK at x = 130) establish a broad pre-crystallization maximum; this is also consistent with the group's prior discrete-sample work (Refs 40,42,43). What is not established is the abstract's operative phrase 'well before crystallization': that is a temperature claim about the annealing state at the peak, and the only temperature data in the paper are two non-isothermal DSC exotherms. The XRD onset at x ≈ 110 mm gives a partial internal anchor (it is consistent with a near-linear gradient putting T(110) ≈ 711 K), but the 711 K value is a 10-K/min heating value; a 15-min isothermal hold crystallizes at a lower temperature, so the peak's margin below crystallization is genuinely unknown — and if the gradient is steeper than linear, the peak could sit near the crystallization boundary, making the 'atomic-heterogeneity regime' interpretation an incipient-crystallization effect. The paper's own admissions (§2.1 'may not be strictly linear'; Methods 'assumed to be one-dimensional'; §2.4 STEM resolution limit) are the strongest evidence that this calibration is missing, and I weigh them as the reader did — as care rather than concealment, but still as the load-bearing gap. I considered two alternative concerns and rank them lower: (i) the Aodd peak abuts the electrode gap — this affects the peak's position/value precision, not the qualitative result; (ii) an internal-consistency check suggesting the reported αANE(TH)/αANE(TM) ≈ 1.31 exceeds the value (~0.9–1.0) recoverable from the reported Aodd, σ and D via Eq. (3) unless κ(TH) exceeds what the measured D implies — a real quantitative worry, but one confined to the derived Fig. 5c quantities rather than the central AEE profile. The Ta(x) concern is directly testable with a thermocouple-instrumented dummy run, and the verdict CONDITIONAL remains correct: acceptance should require the temperature-profile calibration (or an explicit structural calibration of x against effective annealing state).","tokens_in":17365,"tokens_out":36477,"duration_ms":341581,"concrete_test":"Replicate the annealing with a dummy ribbon (same geometry, same furnace position) instrumented with fine thermocouples at 5–10 mm intervals along its length; record Ta(x) over the full cycle (10 K/min ramp, 773 K hot-zone hold for 15 min, natural cooling, fan + wet cloth active). Verify Ta(x) is monotonic; read Ta at the AEE peak (x = 70–80 mm) and at the XRD crystallization onset (x ≈ 110 mm). Determine the effective 15-min isothermal crystallization temperature via isothermal DSC (or a discrete annealing series at 550–750 K for 15 min) to correct the 711 K/10-K/min DSC value. If Ta(70–80) is ≥50 K below the effective crystallization temperature and falls inside the 523–673 K optimum window, the 'well before crystallization' claim is upheld. If Ta(70–80) is at or above the threshold, or Ta(x) is non-monotonic, the central claim is falsified and the FGM-to-discrete-annealing corresponde","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim — 'maximum AEE … appearing in the atomic-heterogeneity regime well before crystallization' (Abstract) — requires each x-position to map to a distinct, monotonically ordered structural state set by the annealing temperature Ta(x). That mapping is assumed, not measured. Methods states the temperature gradient was 'assumed to be one-dimensional'; §2.1 concedes it 'may not be strictly linear.' No thermocouple profile along the ribbon is reported. The assignment of the AEE peak (x = 70–80 mm) to a regime 'well before' the first DSC exotherm (711 K at 10 K/min, Fig. S2) rests entirely on this uncalibrated gradient. The XRD crystallization onset at x ≈ 110 mm anchors only one point, and only if non-isothermal DSC kinetics transfer to the 15-min isothermal furnace hold; isothermal crystallization sets in at a lower temperature, so if the gradient is steeper toward the hot zone (plausible with fan + wet-cloth cooling), T(70–80) could approach the effective crystallization threshold. This matters because the paper's own premise — that lab XRD and STEM cannot resolve the relevant heterogeneity (§2.4: 'nonperiodic features are difficult to resolve directly by STEM') — means XRD-amorphous appearance at the peak cannot exclude incipient nanocrystallization there. The single APT volume at x = 80 (Cu nanoclusters, uniformly distributed Fe/Si/Nb/B) supports the 'pre-crystallization, atomic-heterogeneity' assignment only if T(80) is known to lie well below the crystallization temperature. If Ta(x) is steeper or non-monotonic than assumed, the peak is an incipient-crystallization effect, the abstract's 'well before crystallization' claim fails in its own terms, and the claimed correspondence to the discrete-sample optimum window (523–673 K, §2.1) is ungrounded.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports a functionally graded Fe79Si4B14Nb2Cu1 amorphous ribbon produced by annealing a melt-spun ribbon under a one-dimensional temperature gradient in a tube furnace. Using lock-in thermography (LIT), the authors map the field-odd component of the lock-in amplitude (Aodd) along the ribbon and observe a non-monotonic profile that peaks at x = 70–80 mm, before the XRD crystallization onset near x = 110 mm. STEM and APT on the peak region show an amorphous matrix containing Cu nanoclusters, which the authors interpret as the source of enhanced anomalous Ettingshausen effect (AEE). They further convert Aodd to an anomalous Nernst conductivity αANE using standard transport relations and compare three representative positions (TL, TM, TH). The central claim is that transverse thermoelectric conversion is exceptionally sensitive to subtle structural heterogeneity and that the FGM+LIT approach is a high-throughput route to optimize such materials.","tokens_in":17571,"tokens_out":7559,"duration_ms":76877,"significance":"If the central claim is correct, the work is significant: it demonstrates that a single gradient-annealed ribbon, combined with LIT imaging, can replace discrete annealing-series studies for transverse thermoelectric materials, and it identifies AEE as a sensitive probe of sub-crystallization structural heterogeneity. The experimental protocol has clear strengths: Joule heating is separated by zero-offset square-wave current, the AEE is isolated by ±H odd-field symmetrization (Eqs. 2-1 and 2-2), and uniform magnetization is verified through the constant field-odd phase. The imaging provides a large statistical dataset. However, the quantitative translation from Aodd to αANE contains an internal inconsistency, and the x-to-annealing-temperature mapping underlying the phrase 'well before crystallization' is not calibrated. These issues must be resolved before the claims as stated can be accepted.","major_comments":[{"comment":"There is an internal inconsistency between Aodd and the reported αANE. Under R > 25 (Eq. (7)), αANE ≈ σxx ΠAEE/T, and ΠAEE ∝ κ ΔT_AEE/(j_c t) with ΔT_AEE = 2Aodd. From TM (x = 80) to TH (x = 125), σxx increases by ~1.85×, κ (from D) increases by at most ~1.3–1.7×, while ΔT_AEE drops from ~13.3 mK to ~4.6 mK. The product therefore decreases, yet Fig. 5(c) reports αANE increasing monotonically from 1.83 to 2.39 A/mK. Please provide the full numerical table for κ, σxx, σAHE, Sxx, ΠAEE, and both terms of Eq. (6) at TL, TM, and TH. If αANE actually peaks at TH, then the abstract’s 'maximum AEE before crystallization' applies only to the raw temperature modulation and not to the fundamental transport coefficient; the manuscript must state this distinction explicitly.","section":"§2.5, Eqs. (3)–(7), Fig. 5"},{"comment":"The x→annealing-temperature mapping is not measured. Methods states that heat transfer and the resulting temperature gradient were 'assumed to be one-dimensional,' and §2.1 concedes the gradient 'may not be strictly linear.' No thermocouple profile along the ribbon is reported. The claims that the Aodd peak at x = 70–80 mm corresponds to a particular annealing temperature 'well before crystallization' depend on this uncalibrated mapping. XRD, STEM, and APT anchor the structural state at x = 5, 80, and 125 mm, which supports a structural gradient, but they do not establish the continuous temperature axis. Please either measure Ta(x) along the ribbon or rephrase the interpretation in terms of the directly observed structural states, and quantify how the uncertainty in Ta(x) affects the inferred optimum annealing temperature.","section":"§2.1, Methods; Fig. 1; Fig. 3"},{"comment":"The thermal conductivity κ entering Eq. (3) is not independently reported or defined. The manuscript measures thermal diffusivity D, but Eq. (3) requires κ. If κ is obtained via κ = ρCpD, the density and specific heat values used must be given; if κ is estimated from previous work, this is an unstated free parameter in the central αANE and R analysis. The αANE values in Fig. 5 and the claim R > 25 are therefore not fully determined by the reported measurements. Please list κ at TL, TM, and TH with provenance and uncertainty.","section":"Eq. (3), Methods"},{"comment":"The mechanistic attribution to Cu nanoclusters is based on a single APT volume at x = 80 mm. No comparative APT or cluster statistics are shown for TL (x = 5 mm) and TH (x = 125 mm), so the claim that the AEE enhancement in TM is 'associated with the mechanisms of Cu nanoclusters,' and the abstract’s statement that the heterogeneity 'accounts for the enhanced response,' go beyond the data presented. Please provide APT results from TL and TH (or at least Cu cluster number density and size distributions across the three regimes), or explicitly frame the Cu-cluster mechanism as inferred from prior discrete-sample studies rather than established in this FGM sample.","section":"§2.4, Fig. 4; Abstract"}],"minor_comments":[{"comment":"The legend refers to 'input charge current C', while the text and equations use j_c or I. Please unify notation.","section":"Fig. 1"},{"comment":"The symbol l in dφ_lag/d_l is not defined in the main text; clarify that it is the distance from the thermal excitation center and specify the fitting range used for the slope.","section":"Eq. (1)"},{"comment":"References [41] and [42] are listed as 'accepted' without DOIs. If they are public, provide DOIs; if not, this hampers verification of the comparison with previous discrete-sample results.","section":"References [41], [42]"},{"comment":"The flexibility test is qualitative. Please state the bending radius at which the intact (x < 110 mm) region survives and clarify that the fracture occurred during handling/detachment rather than during bending.","section":"§2.6, Fig. S4"}],"recommendation":"major_revision","confidential_remarks":"The core imaging observation is plausible and the LIT protocol is careful, but the quantitative αANE analysis in Fig. 5 conflicts with the raw Aodd data unless additional information (especially κ) is supplied. The uncalibrated temperature gradient is also a substantive concern for the 'well before crystallization' wording. I recommend major revision rather than rejection because the experimental methodology and the raw Aodd profile are valuable and the issues appear addressable with additional data and revised claims."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The genuinely new thing here is the single-sample platform: a gradient-annealed ribbon that encodes a continuous annealing-temperature series in one strip, read out by LIT mapping of the AEE. That is a real throughput advance over the same group's discrete-sample studies, and the non-monotonic Aodd(x) with a pre-crystallization peak is a plausible, internally consistent observation. The sign of the signal, the odd-field symmetrization, and the constant φodd give me reasonable confidence that what they're mapping is AEE, and the XRD/σ/D plateau before x≈110 mm does support their claim that conventional characterizations would miss the heterogeneity window.\n\nThe soft spots are real but not fatal. Most important: the annealing-temperature profile Ta(x) is never measured. The whole assignment of the peak at x=70–80 mm to 'well before crystallization' rests on an assumed monotonic gradient, and the paper itself concedes the gradient 'may not be strictly linear.' If the real Ta(x) is steeper than assumed, the peak could sit much closer to the isothermal crystallization threshold, and the abstract's central phrase becomes misleading. That's fixable — a thermocouple profile or a calibration run would settle it — but without it, the central claim is underdetermined. Second, the Aodd line profile has no error bars and the peak abuts the electrode gap near x=75 mm, so the peak position is partly inferred. Third, there's a tension between the abstract's 'peak before crystallization' and Fig. 5c, where αANE increases monotonically to TH. The AEE maximum at TM can coexist with a higher αANE at TH because κ changes, but the paper doesn't explicitly reconcile that, and a reader can be forgiven for finding the framing inconsistent. Fourth, the Cu-nanocluster mechanism rests on a single APT volume; that's typical for such studies, but it should be stated as one sample.\n\nNone of these undermine the methodological value. The paper deserves a serious referee, and the right outcome is probably major revision: measure or bracket the temperature profile, add error bars, and soften the 'well before crystallization' language until the x→T mapping is actually supported. I'd be happy to read the revised version.","headline":"A promising single-sample gradient-annealing + LIT platform for transverse thermoelectric screening, but the central 'peak before crystallization' claim rests on an unmeasured temperature profile.","tokens_in":18354,"tokens_out":2300,"would_cite":true,"duration_ms":22315,"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 claims that the anomalous Ettingshausen effect—transverse charge-to-heat conversion in a magnetic material—is maximized not by full crystallization but by an intermediate state of atomic-scale heterogeneity, and that a single tem","keywords":["Functionally graded materials","Anomalous Ettingshausen effect","Lock-in thermography","Amorphous alloys","Structural heterogeneity","Transverse thermoelectric conversion","Cu nanoclusters"],"falsifier":"Measure the true annealing-temperature profile along the ribbon (e.g., with a thin thermocouple array or temperature-sensitive phase-transition markers), or anneal discrete ribbons at a series of known temperatures and check that the AEE peak appears at the temperature inferred for x = 70–80 mm. A non-monotonic Ta(x) or a shift of the peak in the discrete series would undercut the structural assignment.","tokens_in":17088,"feed_emoji":"🔥","tokens_out":4277,"duration_ms":44635,"temperature":0.7,"pith_summary":"The authors try to establish that subtle structural heterogeneity, not full crystallization, is what most strongly enhances the anomalous Ettingshausen effect (transverse charge-to-heat conversion) in a magnetic amorphous alloy. They do this by annealing one Fe-based ribbon under a one-dimensional temperature gradient so that each point along the ribbon represents a different annealing state, then imaging the AEE response with lock-in thermography. The AEE rises to a peak at an intermediate position, then falls; this peak sits well before crystallization sets in, and it is invisible in X-ray diffraction and in ordinary electrical and thermal transport measurements. If true, transverse thermoelectric measurements are uniquely sensitive probes of atomic-scale disorder, and a single graded sample can serve as a high-throughput platform for material optimization.","feed_headline":"Disorder, not crystals, maximizes a transverse thermoelectric effect","feed_subtitle":"Imaging one gradient-annealed ribbon exposes a peak invisible to X-rays and ordinary transport measurements.","key_machinery":"The central object is a temperature-gradient-annealed ribbon that serves as a one-dimensional library of structural states, with position x standing in for annealing temperature. The probe is lock-in thermography: a square-wave current drives the sample, and the field-odd component of the lock-in amplitude (Aodd) isolates the anomalous Ettingshausen effect from Peltier and Joule contributions. Supporting decomposition of the anomalous Nernst conductivity (with an assumed Onsager relation) separates the dominant ANE-related term from the Hall-deflected Seebeck term, attributing the peak to the former.","core_discovery":"On the authors' own terms, annealing a melt-spun Fe79Si4B14Nb2Cu1 ribbon under a temperature gradient produces a functionally graded material whose local annealing state varies continuously along its length. Lock-in thermography maps the field-odd, current-induced temperature oscillation at every point, isolating the anomalous Ettingshausen effect. The resulting AEE amplitude profile is non-monotonic: it rises from 4.90 mK at the cold end, peaks at approximately 6.65 mK at x = 70–80 mm, and then declines, with a steep drop near the crystallization onset at x ≈ 110 mm. Structural analysis by STEM and APT finds Fe-based crystalline alloys and Cu nanoclusters (average 1.7 nm diameter) embedded","pith_inferences":["The same FGM-plus-LIT methodology could be extended to map other transverse transport coefficients—anomalous Hall effect, spin Peltier effect—across structural gradients, turning a single sample into a continuous phase diagram.","Because the position-to-temperature mapping is not calibrated, a direct thermocouple measurement of the annealing-temperature profile along the ribbon would tighten the structural interpretation and could be used to assign an absolute temperature scale to every x-position.","If the pre-crystallization peak is a general feature of magnetic amorphous alloys, device design could deliberately target the disorder-enhanced window instead of avoiding it, preserving flexibility while maximizing conversion."],"forward_implications":["A single heterogeneity-graded sample mapped by lock-in thermography can replace discrete annealing series for locating the optimum transverse thermoelectric response.","Transverse thermoelectric measurements can detect atomic-scale structural heterogeneity that lab-scale X-ray diffraction and longitudinal electrical/thermal transport measurements miss.","The AEE peak appears in the pre-crystallization (atomic-heterogeneity) regime rather than at full crystallization, pointing to cu-nanocluster-driven interfacial scattering as a design lever.","If the curvature is representative, the high-performing region of the ribbon retains mechanical flexibility, making it suitable for flexible transverse thermoelectric devices."],"fun_headline_variants":["Crystals lose: disorder maximizes a transverse thermoelectric effect","Disorder beats crystals for a transverse thermoelectric boost","Imaging reveals disorder as the secret to a transverse effect","Peak transverse heat-to-charge conversion from atomic disorder"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The claim leans on the assumption that each position x along the 130-mm ribbon corresponds to a distinct, monotonically increasing annealing temperature; the actual Ta(x) profile was never measured, so the structural meaning of the peak at x = 70–80 mm rests on an inferred mapping rather than a calibrated one.","fun_headline_variants_meta":{"raw":{"variants":["Crystals lose: disorder maximizes a transverse thermoelectric effect","Disorder beats crystals for a transverse thermoelectric boost","Imaging reveals disorder as the secret to a transverse effect","Peak transverse heat-to-charge conversion from atomic disorder"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000235,"raw_usage":{"total_tokens":1333,"prompt_tokens":738,"completion_tokens":595,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":482,"completion_tokens_details":{"reasoning_tokens":529}},"tokens_in":482,"tokens_out":595,"duration_ms":6332,"temperature":1.0,"reasoning_tokens":529,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T12:16:02.187086+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the true annealing-temperature profile along the ribbon (e.g., with a thin thermocouple array or temperature-sensitive phase-transition markers), or anneal discrete ribbons at a series of known temperatures and check that the AEE peak appears at the temperature inferred for x = 70–80 mm. A non-monotonic Ta(x) or a shift of the peak in the discrete series would undercut the structural assignment.","supporting_citations":[],"review_version":1}