{"id":"c3437b6f-4ab1-4101-a80a-3a6c8cf65710","arxiv_id":"1908.00518","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"A liquid xenon TPC calibrated with fixed-angle neutron scattering reports nuclear-recoil ionization yields from 0.3 to 6 keV, finding a strong downward trend below 1 keV and a 10-15% rise with drift field.","lead":"Using tagged neutron scattering, researchers measured how much ionization xenon nuclei produce when struck by low-energy recoils, down to 300 eV, the lowest such calibration yet. The result matters for dark matter and neutrino experiments that hunt for tiny nuclear recoil signals in liquid xenon detectors.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The sub-keV yields are model-inferred: the 0.296 and 0.442 keV points depend on an assumed Poisson width and an extrapolated normalization A; a floated-width refit is required before those values can be taken as quantitative.","rationale":"The reader's weakest assumption identifies the same load-bearing issue: the two lowest-energy yields are inferred from spectra modeled with a Poisson ionization distribution and an extrapolated normalization A. I agree that this is the most consequential assumption in the paper, because the headline sub-keV claim rests on it. The paper does provide several mitigating elements: a in-situ single-electron calibration, a demonstrated single-electron trigger efficiency, a background-subtraction procedure, and a model-independent cross-check that found agreement within statistical uncertainties. These give real support to the measurement and are properly credited here. Nevertheless, the cross-check inherits the same A prior and the same simulated recoil-energy spectra, so it cannot independently validate the Poisson shape. The unpublished extraction-efficiency reference [23] and the incomplete citation [20] compound the difficulty of independent verification, and the absence of reported ω values from BD1-8 prevents checking the consistency of the Poisson assumption with the measured widths at 1 keV. The concern does not invalidate the paper; the qualitative decline below 1 keV is likely to survive, and the quoted systematics are in the right direction. But because the specific values at 0.296 and 0.442 keV are model-inferred, the paper should either provide the floated-width re-fit or release the code and data needed for an independent check. Since the reader's verdict is already CONDITIONAL and the same concern is reflected there, no further verdict shift is needed.","tokens_in":16751,"tokens_out":17541,"duration_ms":201898,"concrete_test":"Re-run the BD9 and BD10 MCMC fits with ω (or an explicit Fano-factor parameter) free rather than fixed to the Poisson value, using the same Gaussian prior on A; report the posterior median and 68% interval for Qy at 0.296 and 0.442 keV and compare directly with Table I. If the floated-width fits shift Qy by more than the quoted 11% modeling systematic, the current sub-keV points are model-dominated rather than data-dominated. As a second check, repeat the BD10 fit with the center of the A prior shifted by ±1σ and by the quoted ±6.4% left/right scaling systematic, and show the resulting Qy; this directly quantifies how much of the claimed decline below 1 keV depends on the normalization extrapolation.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The two lowest-energy results (BD9 at 0.442 keV and BD10 at 0.296 keV) are not measured directly from the observed spectra alone; they are inferred by an MCMC fit in which the true ionization count is drawn from a Poisson distribution with mean Qy×E (Sec. IV A), and in which the absolute event normalization A is constrained by a Gaussian prior taken from the mean of the A values fitted at BD1-8 (Sec. IV B). At 0.296 keV the best-fit mean is about 1.0 ionized electron, so the Poisson zero-electron fraction is roughly 37%; this is a substantial correction that enters through the model rather than through data. If the true ionization-count distribution at these energies is Fano-narrowed or otherwise different in shape from Poisson, the relation between the observed 1e/2e/3e peak ratios and the mean changes, and the inferred Qy would shift. The paper assigns an 11% modeling systematic to this choice and describes a model-independent cross-check in Sec. IV D, but that cross-check still uses the high-energy A and the same simulated recoil spectra, so it does not independently validate the Poisson shape. The fitted width parameter ω from BD1-8 is never reported, making it impossible to check whether the Poisson assumption (ω=1) is consistent with the trend of ω at 0.93-1 keV before extrapolating downward. A bias in A, for example from an unmodeled energy-dependent selection or from relying on the unpublished extraction-efficiency reference [23], would propagate nearly linearly into Qy at these two energies. The broad conclusion that the yield declines below 1 keV is likely robust, but the numerical values at 0.296 and 0.442 keV, and thus the precise shape of the sub-keV trend, are less secure than the statistical-only error bars in Table I suggest.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper reports a measurement of the nuclear-recoil ionization yield Qy in liquid xenon from 0.3 to 6 keV using monoenergetic 579 keV neutrons from TUNL and an array of fixed-angle liquid-scintillator detectors. The dual-phase TPC is calibrated in situ for single-electron response, trigger efficiency, electron lifetime, and extraction efficiency; recoil spectra are compared with a Geant4-based simulation via MCMC fits that float Qy, a width parameter ω, and a normalization A. The authors report Qy at four drift fields from 220 V/cm to 6240 V/cm, observe a 10–15% increase of Qy with field between 1 and 6 keV, and find a downward trend below 1 keV, reaching Qy = 3.47 e−/keV at 0.296 keV. The two lowest-energy points are extracted using a Poisson ionization model and a Gaussian prior on A inherited from the higher-energy fits.","tokens_in":17046,"tokens_out":9281,"duration_ms":102518,"significance":"If the results hold, this is an important calibration for low-mass WIMP and CEνNS searches: it extends nuclear-recoil charge-yield data below 1 keV for the first time, with single-electron sensitivity, and it provides a field-dependent calibration relevant to many dual-phase xenon detectors. The paper is careful in its treatment of backgrounds, time-of-flight and PSD selection, in-situ calibrations, and Monte Carlo framework; the statistical precision in Table I is sufficient to make the field dependence above 1 keV clearly visible. The principal caveat is that the two sub-keV yields are inference-dependent because zero-electron events are unobservable; the Poisson shape and extrapolated A are used to convert the observed few-electron spectra into a mean yield. The authors quantify this with an 11% modeling systematic, but a free-width robustness fit would materially strengthen the quantitative claim.","major_comments":[{"comment":"The two lowest-energy yields (0.442 and 0.296 keV) are extracted from spectra in which zero-electron events are unobservable, so the mapping from the measured 1e/2e/3e ratios to Qy depends on the assumed shape of the ionization-count distribution and on the extrapolated normalization A. Please report the fitted width parameter ω for BD1–8 and add a robustness fit for BD9/10 in which the width is floated (or the distribution is generalized, e.g., to a negative binomial with a Fano factor), so the reader can see how Qy changes when the Poisson assumption is relaxed. The 11% modeling systematic in Table I is based on a cross-check that still uses the same high-energy A and simulated recoil spectra; it does not fully probe the shape-sensitivity of the fit.","section":"Section IV A/B; Table I, BD9/BD10"},{"comment":"The Gaussian prior on A for the two lowest-energy bins is taken from the mean and standard deviation of A fitted at BD1–8, and the scaling systematic is estimated from a left/right split of the data. This is one specific model of a possible A bias. Please show the fitted A (and ω) values as functions of energy for the BD1–8 fits, and test the sensitivity of the BD9/10 yields to the prior width, for instance by using a flat prior or a prior broadened by the observed scatter. Without this, the absolute normalization of the sub-keV yields rests on an assumption that A is energy-independent up to the specific corrections considered.","section":"Section IV B/D and Table I"},{"comment":"In the iterative fitting, the multiple-scatter background is modeled using an empirical yield-vs-energy function interpolated from the first-stage fits, but no systematic uncertainty is assigned to the choice of this interpolation function. If the multiple-scatter component is non-negligible in the low-energy channels, an incorrect interpolation could bias Qy. Please quantify this by repeating the fits with alternative interpolation forms or by reporting the fractional size of the multiple-scatter component in each channel.","section":"Section IV B (multiple-scatter interpolation)"}],"minor_comments":[{"comment":"The caption states a drift field of 200 V/cm, while the text and Table I use 220 V/cm; please correct this inconsistency.","section":"Figure 6 caption"},{"comment":"The expression pextr(6.24 V/cm) should be pextr(6.24 kV/cm) to match the extraction-field value quoted earlier in the paper.","section":"Section III D"},{"comment":"Reference [20] is listed as '(????), Submitted to Phys. Rev. C' and reference [23] as 'In preparation'; since the LUX comparison and the extraction-efficiency uncertainty rely on these works, please update them with published versions or provide the relevant values in the text.","section":"References [20] and [23]"},{"comment":"There is a typo in 'while the the other two cells'; please remove the duplicated article.","section":"Section II C"},{"comment":"The conclusion contains 'a new measurement of the the nuclear recoil ionization yield'; please delete the duplicated 'the'.","section":"Section VI"}],"recommendation":"major_revision","confidential_remarks":"This is a strong experimental paper with careful calibrations and a clear potential impact on the field. My main reservation is the model dependence of the two lowest-energy data points; I would like to see the floated-width robustness fit and the reported A/ω values before accepting the quantitative sub-keV yields. The incomplete references should also be fixed. With those additions, the paper would be suitable for publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"What you should know: this is the lowest-energy nuclear-recoil ionization yield measurement in liquid xenon to date, reaching 0.3 keV where the signal is on average about one electron. It also reports a statistically significant drift-field dependence in the 1–6 keV range, which contradicts Aprile et al. The paper is well executed and the central data are real; there is no reason to suspect anything shoddy at the experimental level.\n\nThe genuinely new part is the extension below 0.7 keV, where LUX stopped with ~30% uncertainty. The detector is small but purpose-built, with careful in-situ calibration of single-electron response, trigger efficiency, electron lifetime, and background subtraction. The Geant4 simulation is detailed, the fitting is iterative and transparent, and the systematic budget is unusually complete. The 220 V/cm yields agree with LUX where they overlap, so the energy scale appears sane.\n\nThe soft spots are exactly where the reader and stress-test put them. The two lowest-energy points (0.442 and 0.296 keV) are not directly measured from observed peaks; they come from a Poisson model for the ionization-count distribution, with the overall normalization A constrained by a Gaussian prior from the higher-energy fits. That is a real dependence on model assumptions. The paper assigns an 11% systematic for the Poisson choice and does a cross-check, but the cross-check uses the same A and same simulated recoil spectra, so it is not fully independent. The fitted width parameter ω from BD1–8 is never reported, so you cannot see whether the Poisson assumption at low energy is consistent with the trend at 0.93–1 keV. That is a fair referee request, not a fatal flaw. A bias in A would propagate nearly linearly into Qy at those two points, but the cross-check and the statistical errors do bracket the result.\n\nTwo smaller but real issues: the electron extraction efficiency is taken from an in-preparation reference [23], which is unavailable, and Ref [20] (the key LUX comparison) is cited without authors or title. Both are fixable and should be fixed before publication.\n\nThe broad conclusion — yield declines below 1 keV and NEST overpredicts there — is almost certainly robust. The precise numerical values at 0.296 and 0.442 keV are less secure than the statistical-only error bars suggest, but the paper says so in its own systematic discussion, which is honest.\n\nWho this is for: the low-mass WIMP and CEνNS community, plus anyone building or running dual-phase xenon detectors. They will use these numbers. I would bring it to reading group and I would cite it.\n\nSend it to peer review. Ask the authors to report the fitted ω values, ideally a floated-width fit at the two lowest energies, to release data/code, and to complete or replace Refs [20] and [23]. With those, it should be accepted.","headline":"A careful, genuinely new calibration of nuclear-recoil ionization yield in liquid xenon down to 0.3 keV, with sub-keV points that are model-inferred but not fatally so; deserves peer review.","tokens_in":17775,"tokens_out":1702,"would_cite":true,"duration_ms":21411,"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 reports the lowest-energy calibration of nuclear-recoil ionization yield in liquid xenon, down to 0.296 keV, where the average recoil produces 1.1 electrons, and a statistically significant field dependence between 1 and 6 keV.","keywords":["ionization yield","liquid xenon","nuclear recoil","low-energy calibration","single-electron sensitivity","dual-phase time projection chamber","dark matter direct detection","coherent elastic neutrino-nucleus scattering"],"falsifier":"Measure the ionization spectrum of roughly 0.3 keV xenon recoils in a detector that can count zero, one, and two electrons without relying on a high-energy normalization prior, and compare the mean to $3.47\\pm0.4$ e-/keV; a significantly different mean would show that the Poisson or normalization assumption is wrong.","tokens_in":16490,"feed_emoji":"⚛️","tokens_out":9389,"duration_ms":81978,"temperature":0.7,"pith_summary":"The paper aims to pin down how many ionization electrons a low-energy nuclear recoil produces in liquid xenon, the calibration that dark-matter and neutrino detectors need to convert measured charge into recoil energy. Using monoenergetic neutrons scattered at fixed angles and a detector sensitive to single extracted electrons, the authors measured the ionization yield from 0.3 to 6 keV, reaching 0.296 keV, the lowest nuclear-recoil calibration reported to date. At that energy the average signal is 1.1 electrons, or $3.47$ e-/keV at 220 V/cm. The yields agree with earlier results between 2 and 6 keV but with better precision, and below 1 keV they fall faster than the smooth extrapolation commonly used in the field. The paper also reports a 10-15% rise in yield between 220 V/cm and 6240 V/cm over 1-6 keV, the first statistically significant field dependence seen in this range.","feed_headline":"Just 1.1 electrons: xenon recoil yield measured at 0.3 keV","feed_subtitle":"Single-electron calibration could reshape dark-matter and neutrino search sensitivities.","key_machinery":"The load-bearing setup is a dual-phase xenon time projection chamber operated with single-extracted-electron triggering, paired with ten fixed-angle liquid-scintillator neutron detectors. Kinematics of monoenergetic 579 keV neutrons scattering on xenon nuclei fixes each event's recoil energy from the scattering angle, and time-of-flight plus pulse-shape cuts isolate single-scatter neutron events. The analysis then fits simulated ionization spectra, generated by a Monte Carlo model of the setup and re-sampled through measured electron lifetime, extraction efficiency, and single-electron resolution, to the observed charge-count spectra using a Bayesian Markov chain with the ionization yield $Q_y$ and a width parameter $\\omega$ as free parameters. For the two lowest-energy bins the ionization count is modeled as a Poisson process, which introduces the main systematic at those points.","core_discovery":"The central claim is that the nuclear-recoil ionization yield in liquid xenon, measured by tagging neutron scatters and counting electroluminescence pulses at single-electron resolution, declines steeply below 1 keV: the 0.296 keV bin has yield $3.47^{+0.41}_{-0.40}$ e-/keV at 220 V/cm and the 0.442 keV bin $4.58^{+0.39}_{-0.38}$ e-/keV, both below a straightforward extrapolation of higher-energy data. Between 1 and 6 keV the yield is roughly flat at fixed drift field, and the new measurements agree with the prior fixed-angle measurement there but with smaller uncertainties. The same data show that the yield increases by 10-15% as the drift field is raised from 220 V/cm to 6240 V/cm, contrary to a recent measurement that found no field dependence in the 5-14 keV range.","pith_inferences":["If the sub-keV decline is confirmed independently, liquid-xenon searches for low-mass dark matter and coherent neutrino scattering will need to assume smaller signals at fixed threshold, which weakens projected sensitivity unless thresholds reach one or two electrons.","The reported field dependence, combined with single-electron reach, suggests that operating at high drift field could recover part of the signal lost to the low yield; a dedicated background-versus-field study would show whether that gain is usable in practice.","The Poisson-model systematic at 0.30 and 0.44 keV could be resolved by a detector that records the full zero-, one-, and two-electron counting distribution without relying on a high-energy normalization prior, or by an independent measurement of the nuclear-recoil Fano factor."],"forward_implications":["Between 2 and 6 keV the new yields agree with prior measurements but with smaller uncertainty, so existing WIMP and coherent-scattering sensitivity estimates in that range rest on firmer ground.","Below 1 keV the measured falloff means that ionization-only detectors will see fewer electrons per recoil than the common extrapolation predicts; a 0.3 keV recoil typically makes just one electron.","The 10-15% field dependence implies that raising the drift field from about 200 V/cm to several kV/cm can modestly increase the charge signal for 1-6 keV nuclear recoils, a factor experiments can weigh when choosing operating parameters.","Because the measurement reaches single-electron counting, future lower-threshold searches can calibrate at the few-electron level instead of extrapolating from higher energies."],"supporting_citations":[{"why":"Prior fixed-angle measurement of the nuclear-recoil ionization yield at a similar drift field; this work extends it below 1 keV and compares against it.","marker":"[20]"},{"why":"Recent measurement of the field dependence of the yield between 5 and 14 keV; it is the result this work's 1-6 keV field dependence contradicts.","marker":"[21]"},{"why":"Prior measurement of the electron extraction efficiency in the same detector; it supplies the extraction probability correction used throughout the analysis.","marker":"[23]"},{"why":"Monte Carlo framework used to build the detector geometry and generate simulated recoil-energy spectra for the fits.","marker":"[25]"},{"why":"Drift-velocity measurements in liquid xenon used to convert simulated event depths into drift times for the electron-lifetime correction.","marker":"[29]"},{"why":"Public simulation package whose nuclear-recoil ionization model, extrapolated below 1 keV, overpredicts the yields measured here.","marker":"[33]"}],"fun_headline_variants":[],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The two lowest-energy yields are extracted by assuming the ionization count follows a Poisson distribution and by using a normalization fixed by fits at higher energies; if the true distribution is narrower than Poisson or the normalization is biased, the reported 0.30 and 0.44 keV yields shift.","fun_headline_variants_meta":{"error":"Client error '402 Payment Required' for url 'https://api.deepseek.com/chat/completions'\nFor more information check: https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/402"},"cache_creation_input_tokens":0},"created_at":"2026-08-14T15:50:10.578169+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the ionization spectrum of roughly 0.3 keV xenon recoils in a detector that can count zero, one, and two electrons without relying on a high-energy normalization prior, and compare the mean to $3.47\\pm0.4$ e-/keV; a significantly different mean would show that the Poisson or normalization assumption is wrong.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Prior fixed-angle measurement of the nuclear-recoil ionization yield at a similar drift field; this work extends it below 1 keV and compares against it."},{"cited_title":"Aprile, M","cited_arxiv_id":null,"evidence_quote":"Recent measurement of the field dependence of the yield between 5 and 14 keV; it is the result this work's 1-6 keV field dependence contradicts."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Prior measurement of the electron extraction efficiency in the same detector; it supplies the extraction probability correction used throughout the analysis."},{"cited_title":"Akerib, X","cited_arxiv_id":null,"evidence_quote":"Monte Carlo framework used to build the detector geometry and generate simulated recoil-energy spectra for the fits."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Drift-velocity measurements in liquid xenon used to convert simulated event depths into drift times for the electron-lifetime correction."}],"review_version":1}