{"id":"a67bc29d-c82d-49a4-b0fd-5947882f6cfe","arxiv_id":"2607.18378","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A 32.4 g·d PbWO4 cryogenic detector reports the first direct limits on inelastic dark matter at mass splittings up to ~510 keV (SHM) and ~780 keV (LMC).","lead":"This paper uses a lead-tungstate crystal made from archaeological lead to search for inelastic dark matter, which can scatter only after jumping to a heavier dark state. A 32.4 gram-day dataset already excludes some previously untested mass splittings, and larger exposures could reach the last viable electroweak WIMP candidates.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The δ≃510 (780) keV exclusion relies on an energy-independent NR light-yield centroid calibrated only to ~200 keV, while the iDM signal at large δ peaks near 500–800 keV.","rationale":"The reader's verdict identifies the same load-bearing assumption. I agree: the paper is transparent about using Yellin's method, the projection treatment, and the LMC benchmark, and the observed limit at high cross sections is plausible. However, the extension beyond previously probed splittings rests on the detector's response to nuclear recoils at 400–1000 keV, a region that cannot be calibrated with an AmBe source because the maximum Pb recoil energy is ~200 keV. The light detector's suboptimal performance and the lack of an energy-dependent quenching validation make this the weakest point. A binned re-analysis of the existing calibration data, or a measurement at higher recoil energies, would settle whether the acceptance is reliable. Since this is a missing validation rather than a demonstrated error, the appropriate verdict remains CONDITIONAL, unchanged.","tokens_in":9746,"tokens_out":16681,"duration_ms":148024,"concrete_test":"Re-analyze the AmBe calibration data by binning the nuclear-recoil events in recoil energy (0–100, 100–200, 200–300, 300–500, 500–1000 keV) and fitting µ_NR in each bin with a 3σ-clipped Gaussian. If bins above 200 keV contain too few events to constrain the centroid (as expected from AmBe kinematics), or if µ_NR drifts significantly with energy, then the energy-independent acceptance is unsupported. Then recompute the Yellin limit with an acceptance-efficiency curve derived from the binned fits (or from a Lindhard model with plausible parameter values) and check whether the 90% C.L. exclusion still extends beyond δ=330 keV in the SHM. If it does not, the central claim is not robust.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central result—RES-NOVA's 32.4 g·d limit reaching δ≃510 (780) keV—depends entirely on the nuclear-recoil acceptance region in Fig. 1. That region is defined by a single energy-independent NR-band centroid µ_NR=0.073, obtained from an AmBe neutron calibration. AmBe neutrons (E_n≲10 MeV) produce Pb recoils of at most ~190 keV, so the calibration cannot directly validate the band above ~200 keV. Yet for δ=510 keV the inelastic recoil spectrum is concentrated at E_R≈δ µ_A/m_A≈430 keV (and ~660 keV for δ=780 keV), far above the calibrated range. If the true quenching factor q(E) deviates from 0.073 at these energies—e.g., following a Lindhard-type energy dependence—the acceptance efficiency at the signal energies changes, altering the set of dark-red candidate events and the Yellin limit. The paper's own statement that the light detector was 'not optimized' and that no light-channel quality cuts were applied further weakens confidence in the band. The quoted correction of order 10^-4 concerns the band resolution, not the centroid, so it does not test constant quenching.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports a direct-detection search for inelastic dark matter using a 13 g PbWO4 cryogenic calorimeter (archaeological Pb) operated at LNGS, analyzing a 32.4 g·d exposure over 2.5 keV–1 MeV. Events are selected in an a priori nuclear-recoil band defined by AmBe neutron calibration in the light-yield–energy plane. The authors apply Yellin’s optimum interval method to the observed selected events and derive 90% C.L. upper limits on the spin-independent cross section σ_SI as a function of mass splitting δ for a benchmark mχ = 1.1 TeV, under both the SHM and an LMC-motivated velocity distribution. They claim that this prototype exposure already extends the excluded region to δ ≃ 510 (780) keV for SHM (LMC) at σ_SI above about 10^-32 cm^2, beyond the reach of current Xe and bubble-chamber searches. Projected sensitivities for 0.17 and 2.4 tonne·years exposures are also presented, using the simulated RES-NOVA background, and interpreted for complex electroweak multiplets.","tokens_in":10037,"tokens_out":3878,"duration_ms":35593,"significance":"If the technical assumptions hold, this is a valuable new probe of the inelastic dark matter parameter space. The paper’s strengths include the use of a real dataset, an a priori definition of the nuclear-recoil acceptance region, a background-model-independent Yellin analysis for the observed limit, and an explicit treatment of both SHM and LMC velocity distributions. The heavy Pb target indeed provides a kinematic advantage for large mass splittings, and the high-energy acceptance up to 1 MeV is a genuine complement to xenon-based searches. The projected sensitivity to the Higgsino and other Y=1/2 electroweak multiplets is also interesting. However, as discussed below, the central observed-limit claim depends on an extrapolation of the nuclear-recoil light-yield calibration to energies far above those populated by the AmBe calibration source, and this is not validated in the manuscript.","major_comments":[{"comment":"The central claim that the 32.4 g·d exposure excludes new parameter space up to δ ≃ 510 (780) keV depends on the nuclear-recoil acceptance region defined by the energy-independent NR-band centroid μ_NR = 0.073. The AmBe neutron calibration populates Pb recoils only up to about 190 keV, while for δ = 510 (780) keV the recoil energy that minimizes vmin is E_R ≈ (μ_A/m_A) δ ≈ 430 (660) keV. The manuscript states that the energy-dependent correction to the NR-band resolution is only O(10^-4), but this does not address the centroid (quenching factor). If the light-yield centroid varies with recoil energy above the calibrated range, the ±3σ NR band at the signal energies changes, the set of selected events changes, and the Yellin limit shifts. The authors need to demonstrate that the constant-quenching assumption is valid up to 1 MeV, or to propagate an energy-dependent quenching uncertainty i","section":"Event Discrimination and Acceptance Region; Results"},{"comment":"The observed-limit curve in Fig. 2 is not accompanied by the number of selected nuclear-recoil candidate events as a function of energy, the exposure, or the detection efficiency after the 2.5 keV–1 MeV and ±3σ NR-band cuts. Without these data, the reader cannot assess how the high-δ exclusion is obtained, especially since Yellin’s optimum interval method is sensitive to the local spacing of events. The energy scale is anchored to 2615 keV and 46 keV γ lines, but no systematic uncertainty on the energy scale at 400–700 keV is given. This is a reproducibility gap for the key observed result.","section":"Data analysis; Statistical treatment; Results"}],"minor_comments":[{"comment":"The notation δ_RN^max (and later δ_max^CDD) is used without definition; please define these symbols at first use.","section":"Results"},{"comment":"The red curve is an observed limit, while the green and orange curves are projections; the caption says “Current sensitivity and the projected 90% C.L. upper limit.” Please distinguish observed limit from projected sensitivity in the caption.","section":"Fig. 2"},{"comment":"The phrase “nuclear recoils with calibrated efficiency up to 1 MeV” is misleading: the AmBe calibration extends only to about 190 keV. Please rephrase to indicate the nominal acceptance region, not a calibrated efficiency.","section":"Event Discrimination and Acceptance Region"},{"comment":"The text states that no quality cuts were applied to the light channel to avoid systematics, yet the light channel is used for event selection. Please clarify how the absence of cuts affects the light-yield resolution and the NR-band definition.","section":"Data analysis"},{"comment":"Reference [30] is listed as “in press” with a 2026 date; this should be updated with a published journal reference and volume/page if available.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The paper presents an interesting experimental result, but the observed-limit claim rests on an energy-independent nuclear-recoil light-yield centroid that is only calibrated at energies below ~200 keV, while the claimed new exclusion is driven by signal energies of 400–700 keV. This is a load-bearing issue that should be resolved before acceptance. Additionally, the lack of an event-count table for the observed analysis will make it difficult for the community to reproduce the limit. I recommend major revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The thing to know: this is a real measurement, not a recast or a projection. The RES-NOVA prototype, 32.4 g·d of PbWO4, produces a new 90% C.L. exclusion for inelastic dark matter that reaches δ ≈ 510 keV (SHM) and 780 keV (LMC) at σ_SI above ~10^-32 cm^2. That is a genuine extension beyond the ~330 keV where xenon runs out. The analysis is straightforward and honest: the nuclear-recoil acceptance is defined a priori, the observed limit comes from Yellin's optimum interval without a background model, and the ton-scale curves are clearly simulated projections.\n\nThe soft spot is the NR band itself. The band uses a single energy-independent light-yield centroid, µ_NR = 0.073, fitted from AmBe neutron data. AmBe neutrons give Pb recoils up to about 190 keV, so the centroid is directly measured only below that. But the iDM signal for δ near 500–780 keV peaks at recoil energies of several hundred keV. So the headline reach rests on the assumption that the quenching factor is flat from 200 keV to 1 MeV. The paper notes that the energy-dependent correction to the band resolution is tiny, but that says nothing about the centroid. If the centroid drifts, the accepted event set changes and the limit moves. This is a real, unquantified systematic. It's probably not fatal—quenching for heavy ions often flattens at high energies—but \"probably\" is not calibration. The text's phrase \"calibrated efficiency up to 1 MeV\" overstates what the data actually constrain.\n\nSmaller issues: no event-count table or raw data, so the Yellin result can't be independently checked from the paper. The projections depend on the collaboration's own simulated background. The light detector was not optimized and no light-channel quality cuts were applied, which adds a bit more uncertainty to the band. None of this is a dealbreaker; it just limits how much the reader can verify.\n\nBottom line: this is a competent, honest letter that should go to peer review. The main requested revision should be a quantified treatment of the NR-band energy dependence—a direct calibration at higher recoil energies or a systematic that propagates into the limit. The paper is for the direct-detection and iDM communities, and it deserves a serious referee.","headline":"New PbWO4 iDM limits are real but the high-δ reach depends on an uncalibrated extrapolation of the nuclear-recoil light yield.","tokens_in":10814,"tokens_out":4493,"would_cite":true,"duration_ms":37819,"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":"The RES-NOVA prototype detector already excludes inelastic dark matter at mass splittings up to 510 keV (SHM) and 780 keV (LMC), going beyond the kinematic reach of xenon-based searches.","keywords":["inelastic dark matter","cryogenic calorimeter","PbWO4","nuclear recoil","mass splitting","direct detection","light-yield discrimination","optimum interval method"],"falsifier":"Measure the nuclear-recoil light yield in PbWO4 at cryogenic temperature at recoil energies spanning 2.5 keV–1 MeV, for example with a tagged neutron beam or monochromatic neutron sources. If the measured quenching factor deviates from 0.073 as a function of energy, or if the ±3σ band does not contain the true nuclear-recoil population at the extremes, the accepted-event set and the Yellin limits would change; a large deviation could erase the claimed extension beyond 330 keV.","tokens_in":9598,"feed_emoji":"⚛️","tokens_out":4987,"duration_ms":41888,"temperature":0.7,"pith_summary":"This paper uses a 13-gram PbWO4 cryogenic crystal, operated as a bolometer with a light detector, to search for inelastic dark matter—a scenario where the dark-matter particle must first up-scatter into a slightly heavier state before colliding with a nucleus. With only a 32.4 g·day exposure, the detector sets new 90% exclusion limits at mass splittings up to 510 keV under the Standard Halo Model and 780 keV under a Large Magellanic Cloud–motivated velocity distribution, extending beyond the roughly 330 keV reach of established xenon experiments. The advantage comes from the heavy lead target (A ≈ 207), which raises the maximum kinematically accessible splitting, combined with a nuclear-recoil acceptance that extends to 1 MeV, capturing the high-energy recoils that large splittings produce. If these limits hold, the technique already probes previously unconstrained inelastic-dark-matter parameter space, and with future tonne-year exposures it would reach cross sections characteristic of thermal-relic electroweak WIMPs such as the Higgsino.","feed_headline":"Prototype Pb detector excludes inelastic dark matter to 510 keV","feed_subtitle":"A 13-gram archaeological-lead bolometer, exposed 32.4 g·days, beats xenon's reach on large mass splittings.","key_machinery":"The central object is the inelastic up-scattering threshold: vmin(ER) = (mA·ER/µA + δ)/√(2·mA·ER), whose minimum over ER gives δ_max = ½µA·vmax². The heavy Pb nucleus (A ≈ 207) enlarges the reduced mass µA, pushing δ_max to about 530 (830) keV for SHM (LMC); the detector's acceptance of nuclear recoils up to 1 MeV keeps events with ER ≃ δ, which fall at hundreds of keV, inside the analysis window. Event selection relies on the light-yield discrimination between electron recoils and nuclear recoils, with the nuclear-recoil band defined from an AmBe neutron calibration at µ_NR = 0.073, and limits are set with Yellin's optimum-interval method.","core_discovery":"The paper establishes that inelastic dark matter with mass splitting δ can be probed at δ up to about 510 keV (SHM) or 780 keV (LMC) with a PbWO4 cryogenic detector using just 32.4 g·d of exposure. For spin-independent scattering normalized per nucleon, the minimum incoming speed is vmin(ER) = (mA·ER/µA + δ)/√(2·mA·ER), and the maximum accessible splitting is δ_max = ½µA·vmax². Lead (A ≈ 207) gives a substantially larger δ_max than xenon (A ≈ 131), and the detector's 2.5 keV–1 MeV nuclear-recoil acceptance includes recoil energies of several hundred keV, where large-δ signals are concentrated. Using Yellin's optimum-interval method on events selected in the nuclear-recoil band, the paper rep","pith_inferences":["The limits hinge on assuming an energy-independent light-yield ratio for nuclear recoils; if the quenching factor in PbWO4 varies with recoil energy beyond the stated ±3σ band, the accepted event set changes and the reported δ reach would shift, so a dedicated calibration across 2.5 keV–1 MeV would settle this.","The same PbWO4 detector design could be applied to other inelastic or excited-state dark-matter models, such as two-step luminous dark matter, where the high-recoil-energy acceptance is equally beneficial.","The strong LMC dependence suggests that experiments could augment sensitivity by targeting directions or time windows with enhanced high-velocity halo components, e.g., through annual modulation or directional detection.","If a 2.4 tonne·year exposure returns a null result, it would exclude a significant portion of electroweak-multiplet parameter space above common thermal-relic masses, which would constrain supersymmetric extensions."],"forward_implications":["Existing 32.4 g·d data already exclude previously unconstrained parameter space in δ ≈ 330–510 keV (SHM) and δ ≈ 500–780 keV (LMC) at σ_SI above roughly 10⁻³² cm².","At large splittings the signal appears at recoil energies ER ≃ δ, so an analysis window extending to 1 MeV is essential; this is why xenon TPCs, which typically analyze only up to about 100 keV, lose sensitivity in this region.","Future exposures of 0.17 and 2.4 tonne·years would push sensitivity to σ_SI ≈ 10⁻³⁹ cm², reaching the thermal-relic cross sections of complex electroweak multiplets with hypercharge Y = 1/2, including the 1.1 TeV Higgsino.","The LMC-motivated velocity distribution roughly doubles the accessible δ relative to SHM, demonstrating that the halo velocity model strongly affects the reach.","At 2.4 tonne·years, the projected reach exceeds current limits for all thermal Y = 1/2 multiplets from n = 2 to n = 12 under the SHM, and retains discovery potential up to the 100 TeV unitarity bound."],"fun_headline_variants":["Lead bolometer beats xenon for heavy dark matter splitting","13 grams of archaeological lead probe dark matter beyond 500 keV","Inelastic dark matter constrained to 510 keV with 32 g·d","Ancient lead detector reaches 780 keV dark matter splitting"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The nuclear-recoil acceptance region is defined from an AmBe neutron calibration assuming an energy-independent light-yield ratio (µ_NR = 0.073) with a ±3σ band, and this same band is applied to dark-matter-induced recoils across 2.5 keV–1 MeV; if the quenching factor or its resolution varies with recoil energy, the set of accepted events changes and the reported limits shift.","fun_headline_variants_meta":{"raw":{"variants":["Lead bolometer beats xenon for heavy dark matter splitting","13 grams of archaeological lead probe dark matter beyond 500 keV","Inelastic dark matter constrained to 510 keV with 32 g·d","Ancient lead detector reaches 780 keV dark matter splitting"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000962,"raw_usage":{"total_tokens":3934,"prompt_tokens":747,"completion_tokens":3187,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":491,"completion_tokens_details":{"reasoning_tokens":3116}},"tokens_in":491,"tokens_out":3187,"duration_ms":18695,"temperature":1.0,"reasoning_tokens":3116,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T15:33:34.041973+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the nuclear-recoil light yield in PbWO4 at cryogenic temperature at recoil energies spanning 2.5 keV–1 MeV, for example with a tagged neutron beam or monochromatic neutron sources. If the measured quenching factor deviates from 0.073 as a function of energy, or if the ±3σ band does not contain the true nuclear-recoil population at the extremes, the accepted-event set and the Yellin limits would change; a large deviation could erase the claimed extension beyond 330 keV.","supporting_citations":[],"review_version":1}