{"id":"e7687bad-e27e-4da2-8388-7717c71a3308","arxiv_id":"2505.04045","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Spalling bulk BaTiO3 yields single-crystal films with measured Pockels coefficients up to 160 pm/V in small single-domain regions and projected r42 up to 1980 pm/V.","lead":"Researchers peeled thin films off a bulk barium titanate crystal using a stressed nickel layer, a trick borrowed from semiconductor manufacturing. The peeled crystals retain a strong electro-optic response, potentially enabling compact optical modulators that outperform thin-film lithium niobate.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claimed r33 = 160 pm/V in 'likely single-domain' regions rests on an unverified local domain assignment and a Teng-Man analysis that assumes a smooth, homogeneous c-axis film, while the 20-μm film has RMS roughness >200 nm; this is the weakest link.","rationale":"I read the paper in good faith: the spalling fabrication, transfer, and Teng-Man calibration on bulk BTO are described in detail, and the bulk calibration value of 96.4 pm/V is close to the literature value, so the setup itself is credible. The weakest link is exactly the one the reader identified: the r33 = 160 pm/V result is the key evidence that spalled films 'preserve bulk electro-optic properties,' but it is taken in a 'likely single-domain' region whose local domain state is not directly verified, and it is extracted with a reflection analysis that assumes a smooth homogeneous film. The manuscript even flags the nonideal roughness and partial polarization rotation, so the concern is internal to the presented evidence rather than an external ideological objection. The 160 pm/V value is well above the bulk r33 reference, which makes it especially load-bearing: if it is an artifact of roughness, scattering, or a mixed-domain spot, the remaining measured r33 = 55 ± 5 pm/V is only about half the bulk value, and the headline claim of preserved bulk performance loses its quantitative support. The r42 projection of 1980 pm/V is derivative and would also fail if the 160 pm/V measurement fails. I do not think this warrants rejection, because the observed issue is a missing verification and a possible systematic bias, not a demonstrated error. A CONDITIONAL verdict is therefore appropriate, and the proposed co-localized EBSD/PFM plus planarization remeasurement is a concrete way to settle whether the concern actually lands.","tokens_in":16016,"tokens_out":6236,"duration_ms":67394,"concrete_test":"Perform a spatially co-registered measurement on the actual (001) device from Fig. 4: first map the exact configuration-(2) probe region with EBSD or confocal PFM to establish whether it is a single c-domain, and measure the topography there. Then planarize or index-match that same region (e.g., polish to RMS < 20 nm or spin-coat a transparent layer) and repeat the Teng-Man measurement of r33 at the same spot. If the value does not reproduce ~160 ± 40 pm/V under verified single-domain, low-roughness conditions—or if the local map shows multi-domain or mixed orientation—the 160 pm/V attribution fails and the headline claim should be downgraded.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that spalling preserves bulk BTO electro-optic performance rests on the r33 = 160 ± 40 pm/V value assigned to a 'likely single-domain' region. For this number to validate the claim, the optically probed spot must (i) actually be a single c-domain with out-of-plane polarization and (ii) satisfy the Teng-Man reflection model of Eqs. (4)-(5): a smooth, homogeneous, c-axis film with bulk indices and ζ = 0.1. Neither condition is established for the Fig. 4 device. EBSD in Fig. 2e was performed on a (100)-oriented spalled film, not on the (001) device measured in Fig. 4, and configuration (2) only selects 'regions with fewer domains' by optical inspection; no local EBSD or PFM verifies the polarization state at the tens-of-micron probe spot. The film is ~20 μm thick with RMS roughness >200 nm (SI Part 7, consistent with Fig. 2d). At λ = 1500 nm, 200 nm roughness is a large fraction of a wavelength, and the authors themselves invoke scattering and partial polarization rotation to explain the asymmetric |Im| peaks in Fig. 4c. Those same effects can bias the fitted Γm or δΨsp, especially in the focused-beam configuration (2), where the collimated-beam, single-angle formula is less controlled. The 160 pm/V value also exceeds the bulk r33 ≈ 105 pm/V used as the calibration reference, so it is doing substantial work: without it, the measured r33 is 55 ± 5 pm/V, about half the bulk value, which does not support 'preserve bulk properties.' Thus the concern is not disagreement with consensus but that the measurement as presented does not yet exclude roughness and domain artifacts as the source of the headline number.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports a spalling-based method for fabricating single-crystal BaTiO3 (BTO) thin films, with demonstrated thicknesses from 100 nm to 15 μm and lateral dimensions up to several millimeters, and a transfer process onto SiO2/Si using UV-curable resin. XRD and EBSD are used to show crystallinity and local domain structure. Electro-optic characterization is performed with a Teng-Man reflection setup calibrated on bulk BTO, yielding 96.4 pm/V, close to literature values. In spalled films, the authors report r33 = 55 ± 5 pm/V in multi-domain regions after poling and r33 = 160 ± 40 pm/V in smaller, 'likely single-domain' regions probed with a focused configuration. They then project r42 = 680 pm/V and, for the 160 pm/V region, up to 1980 pm/V by assuming the bulk r42/r33 ratio, and conclude that spalled BTO films preserve bulk electro-optic properties and exceed thin-film lithium niobate performance.","tokens_in":16372,"tokens_out":4821,"duration_ms":47149,"significance":"If the central claim is reliably established, the paper would be significant: spalling would provide a low-cost, scalable route to large-area single-crystal BTO films with bulk-like electro-optic coefficients, attractive for integrated photonic modulators. The work has several concrete strengths: the spalling and transfer procedures are described in detail; the Teng-Man apparatus is calibrated against bulk BTO with a value consistent with literature; the frequency dependence of the modulation signal is reported up to 1 MHz; and the authors explicitly acknowledge surface roughness and domain switching as limitations. However, the evidence for the headline conclusion currently rests on a single local electro-optic value whose domain assignment and optical-model assumptions are not independently verified, while the multi-domain value alone (55 ± 5 pm/V) is about half the bulk r33 and would not by itself support the preservation claim.","major_comments":[{"comment":"The claim that spalled BTO preserves bulk electro-optic properties hinges on r33 = 160 ± 40 pm/V assigned to 'likely single-domain regions', but the domain state at the optically probed spot is not established. The EBSD map in Fig. 2e was taken on a (100)-oriented spalled film, whereas the electro-optic device in Fig. 4a is a (001)-oriented film; configuration (2) merely selects 'regions with fewer domains' by optical inspection. Without local EBSD or PFM on the actual device at the probe location, the 160 pm/V value cannot be attributed specifically to a single c-domain, and the abstract's phrasing 'r33 = 160 pm/V in single-domain regions' overstates the evidence. This is load-bearing because the poled multi-domain value is 55 ± 5 pm/V, roughly half the bulk r33 = 105 pm/V used as reference, so the preservation claim stands or falls on the unverified 160 pm/V measurement.","section":"Electro-optic characterization; Fig. 4d and following paragraph"},{"comment":"The Teng-Man analysis embodied in Eqs. (4)-(5) assumes a smooth, homogeneous, c-axis oriented film with bulk refractive indices and ζ = 0.1. The measured device is ~20 μm thick with RMS roughness greater than 200 nm (SI Part 7, consistent with Fig. 2d), which is a substantial fraction of the probing wavelength λ = 1500 nm. The authors themselves attribute the asymmetric |Im| peaks in Fig. 4c to scattering and partial polarization rotation. Under these conditions, the single-angle, collimated-beam formula can be biased, and no control experiment or roughness-aware transfer-matrix model is provided to quantify the resulting error in r33. The concern is stronger for configuration (2), where the focused beam replaces the irises and the planar-wave assumption underlying the Teng-Man derivation is even less controlled.","section":"Eqs. (4)-(5) and Fig. 4c-d"},{"comment":"The projected r42 values (680 pm/V after poling and 'up to 1980 pm/V' in the abstract) are not measured quantities: they are obtained by assuming the bulk r42/r33 ratio holds in the spalled film, as stated in the main text. Because the r42/r33 ratio could be affected by clamping, roughness, or the local domain state, this projection should be explicitly labeled as an assumption-dependent estimate wherever it is used to support claims of exceeding thin-film lithium niobate performance. The abstract currently presents the 1980 pm/V projection without this caveat, which makes it appear to be a measured result.","section":"Abstract and 'r42 projection' paragraph after Fig. 4d"}],"minor_comments":[{"comment":"The abstract and conclusion state 'r33 = 160 pm/V in single-domain regions', while the main text says 'likely single-domain areas' and 'likely single-domain regions'; please align the wording with the level of evidence actually available.","section":"Abstract and conclusion"},{"comment":"The comparison of the spalled-film r33 values (55 ± 5 pm/V) with the r42 values for PLD, sputtering, and other methods in Table 1 compares different tensor elements; the comparison should be made on the same coefficient or on an effective electro-optic coefficient.","section":"Table 1 and text following Fig. 4c"},{"comment":"The sentence describing the profilometer scan refers to 'middle inset' and 'right inset' without explicit labels in the figure; please label the insets or describe them by position and content more explicitly.","section":"Fig. 1c description"},{"comment":"Please clarify whether the bulk calibration value of 96.4 pm/V obtained on the (100)-oriented substrate with an in-plane field is r33 or an effective coefficient involving r42, and define the 'c-axis configuration' used for that measurement.","section":"SI Part 5"},{"comment":"The simplification from Eq. (4) to Eq. (5) with ζ = 0.1 and n_o ≈ n_e ≈ n introduces a numerical factor 10/9; showing the intermediate algebraic steps would remove ambiguity for readers.","section":"Eq. (5)"},{"comment":"The phrase 'controllable thicknesses ranging from 100 nm to 15 μm' is stronger than the demonstrated data, since the spall depth varies spatially within a given film; consider 'with demonstrated thicknesses' or a similar qualifier.","section":"Abstract"}],"recommendation":"major_revision","confidential_remarks":"The fabrication and transfer work is interesting and likely within scope, but the central evidence for preserved bulk electro-optic performance rests on one local value whose domain state is not verified and whose Teng-Man retrieval uses assumptions violated by the film's roughness. I would ask the authors to add local domain imaging at the actual probe locations and either a roughness-bias analysis or a control experiment before the claim can be accepted as stated. This is an evidentiary issue that can be addressed within the manuscript's scope, hence major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The thing to know: this paper's real contribution is the spalling process itself—applying stress-induced exfoliation to a ferroelectric oxide and transferring the flakes onto a substrate. That is new, it works, and the roughness–thickness scaling law (r ~ C·t^0.5) is a useful engineering result. The Teng-Man calibration on bulk BTO (96.4 pm/V) checks out, so the measurement chain is credible. Credit where due: the fabrication and transfer sections are careful, and the paper is honest in the text about the film's roughness and domain complications.\n\nThe soft spot is the headline electro-optic number. The r33 = 160 ± 40 pm/V comes from a region only inferred to be single-domain, and the EBSD was done on a different film, not the device actually measured. The probed film is ~20 μm thick with RMS roughness over 200 nm—hardly the smooth, homogeneous c-axis film that the Teng-Man analysis assumes. The authors themselves invoke scattering and polarization rotation to explain asymmetries in their data, and those same effects can bias the extracted coefficient. The 160 pm/V also exceeds the bulk r33 ≈ 105 pm/V used as calibration, so it is doing a lot of work in the argument. Without it, the measured value is 55 ± 5 pm/V after poling, which is roughly half bulk—fine as a demonstration, but not evidence that spalled BTO preserves bulk electro-optic properties.\n\nThe abstract overreaches: it drops the word \"likely\" before \"single-domain\" and headlines the projected r42 = 1980 pm/V, which depends on assuming the bulk r42/r33 ratio holds. That is a projection, not a measurement, and it should not lead the abstract.\n\nAll that said, this is not a desk-reject. The spalling of BTO is a solid, reproducible fabrication result that will interest people working on BTO photonics and layer transfer. The electro-optic data are preliminary and should be treated as such. My recommendation: send it to peer review, but require major revision—ideally local PFM or EBSD on the actual device, or at minimum relegate the 160 pm/V and the r42 projection to clearly labeled tentative results and remove them from the abstract. This is a useful paper in need of disciplined claims, not a flawed one in need of rejection.","headline":"The spalling process for BTO is a genuine fabrication advance, but the headline 160 pm/V rest on an unverified single-domain assumption on a rough, thick film; it deserves peer review with major revision, not desk rejection.","tokens_in":17016,"tokens_out":1784,"would_cite":false,"duration_ms":20929,"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":"Spalling single-crystal barium titanate into thin films preserves its strong electro-optic response, with r33 up to 160 pm/V in likely single-domain regions and a projected r42 near 1980 pm/V.","keywords":["barium titanate","thin-film spalling","electro-optic coefficient","Pockels effect","Teng-Man measurement","photonic integration","single-crystal films","lithium niobate comparison"],"falsifier":"Map the domain structure of the exact (001) film used in Fig. 4 with polarization-resolved EBSD or piezoresponse force microscopy, then correlate the map with the position probed by the Teng-Man measurement; if the 160 pm/V spot is actually multi-domain, or if remeasuring with a clamped high-frequency drive drops the coefficient below the thin-film lithium niobate value, the central claim would be contradicted.","tokens_in":15790,"feed_emoji":"⚡","tokens_out":5235,"duration_ms":51935,"temperature":0.7,"pith_summary":"The paper tries to establish that spalling—peeling single-crystal barium titanate (BTO) films off a bulk wafer with a stressed nickel layer—produces films that keep the bulk material's strong electro-optic response. If true, it gives photonics a scalable route to BTO thin films without slow epitaxial growth, with measured coefficients that already beat commercial thin-film lithium niobate. The evidence combines film fabrication, transfer, and Teng-Man reflection electro-optic measurements. The paper reports $r_{33} = 55 \\pm 5$ pm/V in multi-domain films, rising to $r_{33} = 160 \\pm 40$ pm/V in smaller likely single-domain regions, and projects $r_{42}$ near 1980 pm/V if the bulk $r_{42}/r_{33}$ ratio holds. The practical payoff would be compact, low-voltage modulators needing much shorter interaction lengths than lithium niobate.","feed_headline":"Spalled barium titanate films reach 160 pm/V electro-optic coefficient","feed_subtitle":"Stress-exfoliated BTO exceeds thin-film lithium niobate, enabling compact low-voltage modulators.","key_machinery":"The central mechanism is spalling: a Ni stressor electroplated on a Ti/Au seed layer puts the BTO surface under compression, and when the residual stress exceeds a critical value, a roller-mounted tape initiates a fracture whose depth is set by the force balance between tensile stress, compressive stress, and bending moment. The second load-bearing element is the Teng-Man reflection measurement, which extracts $r_{33}$ from the voltage-induced phase retardance between s- and p-polarized reflected light using a Soleil-Babinet compensator. Together these convert a bulk single crystal into a device-ready thin film and then quantify its electro-optic response.","core_discovery":"Spalled BTO thin films are single-crystalline and retain a bulk-like Pockels response despite surface roughness and stress-induced domain switching. Using a Teng-Man reflection measurement at 10 kHz (unclamped conditions), the authors measured $r_{33} = 42 \\pm 3$ pm/V before poling, $r_{33} = 55 \\pm 5$ pm/V after out-of-plane poling in multi-domain regions, and $r_{33} = 160 \\pm 40$ pm/V when probing tens-of-micron areas with fewer domains. Taking the bulk $r_{42}/r_{33}$ ratio from BTO, the local measurement projects to $r_{42} \\approx 1980$ pm/V under unclamped excitation, exceeding the best reported MBE-grown BTO. The paper argues that these numbers show bulk electro-optic properties are preserved locally in spalled films, making them suitable for integrated photonic devices.","pith_inferences":["If the high $r_{33}$ value depends on the unclamped piezoelectric contribution, clamped MHz-to-GHz operation could show a smaller coefficient; the paper does not directly measure the clamped response.","The single-domain attribution should be tested directly: EBSD was performed on a (100) film, while the high $r_{33}$ came from a (001) spall, so the domain state at the exact probe spot remains unverified.","Reducing roughness and scattering could push measured values closer to bulk $r_{33} = 105$ pm/V or allow a direct angled-electrode test of the $r_{42}$ projection.","The same stressor and fracture mechanics may transfer to other perovskite oxides, since the spalling control does not depend on BTO-specific chemistry."],"forward_implications":["Films 100 nm to 15 µm thick and millimeters across can be transferred onto Si/SiO2, so the process is compatible with established photonic platforms.","At $r_{33} = 160$ pm/V, spalled BTO exceeds thin-film lithium niobate's $\\sim 30$ pm/V, which implies shorter modulators and lower drive voltages.","The projected $r_{42} \\approx 1980$ pm/V would put spalled BTO above MBE-grown BTO (923 pm/V), the previous thin-film benchmark.","Bulk substrates can be repolished and reused, lowering the material cost per device by up to roughly 20-fold.","Poling and lower-stress spalling improve both roughness and the measured coefficient, indicating a clear path toward better films.","The measured values are unclamped (10 kHz) and limited by an RC roll-off near 18 kHz, so faster devices will require different electrode geometries."],"supporting_citations":[{"why":"Supplies the controlled-spalling mechanics and stressor-thickness relationship used to set spall depth and film thickness.","marker":"[35]"},{"why":"Establishes wafer-scale spalling feasibility for other material systems, supporting the scalability claim for BTO.","marker":"[38]"},{"why":"Provides the Teng-Man reflection technique used to extract electro-optic coefficients from the spalled films.","marker":"[44]"},{"why":"Extends the reflection measurement formalism and supports the analysis of reflected-beam phase retardance.","marker":"[46]"},{"why":"Supplies bulk BTO electro-optic and dielectric reference data and the unclamped frequency context used to interpret the 10 kHz measurements.","marker":"[27]"},{"why":"Provides the bulk BTO Pockels coefficients and the lithium niobate benchmark that the spalled films are compared against.","marker":"[4]"},{"why":"Gives the MBE-grown BTO $r_{42} = 923$ pm/V result that the paper's projected $r_{42}$ is meant to exceed.","marker":"[28]"},{"why":"Contextualizes the frequency dependence of BTO Pockels coefficients and the distinction between clamped and unclamped response.","marker":"[48]"}],"fun_headline_variants":["Spalled BTO hits 160 pm/V Pockels coefficient","Exfoliated BTO films beat lithium niobate for modulators","Stress-spalled BTO retains bulk electro-optic response","BTO spalling yields 160 pm/V, better than TFLN"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The central claim rests on the assumption that the spot where $r_{33} = 160 \\pm 40$ pm/V was measured really is single-domain and that bulk refractive indices, $\\zeta = 0.1$, and the bulk $r_{42}/r_{33}$ ratio remain valid in the rough spalled film.","fun_headline_variants_meta":{"raw":{"variants":["Spalled BTO hits 160 pm/V Pockels coefficient","Exfoliated BTO films beat lithium niobate for modulators","Stress-spalled BTO retains bulk electro-optic response","BTO spalling yields 160 pm/V, better than TFLN"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000685,"raw_usage":{"total_tokens":3097,"prompt_tokens":925,"completion_tokens":2172,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":541,"completion_tokens_details":{"reasoning_tokens":2096}},"tokens_in":541,"tokens_out":2172,"duration_ms":16978,"temperature":1.0,"reasoning_tokens":2096,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T23:38:33.817571+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Map the domain structure of the exact (001) film used in Fig. 4 with polarization-resolved EBSD or piezoresponse force microscopy, then correlate the map with the position probed by the Teng-Man measurement; if the 160 pm/V spot is actually multi-domain, or if remeasuring with a clamped high-frequency drive drops the coefficient below the thin-film lithium niobate value, the central claim would be contradicted.","supporting_citations":[{"cited_title":"S., Sluka, T., Tagantsev, A","cited_arxiv_id":null,"evidence_quote":"Provides the bulk BTO Pockels coefficients and the lithium niobate benchmark that the spalled films are compared against."}],"review_version":1}