{"id":"22a4957d-7df9-44c6-9e9d-3df28aa239aa","arxiv_id":"2507.07119","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"First dual heat-and-light readout of a GaAs cryogenic calorimeter, showing alpha-particle events with roughly 10x the light yield of beta/gamma events.","lead":"Scientists cooled a 4.3 g gallium arsenide crystal to 10 mK and recorded both its heat and scintillation light signals for the first time. Alpha particles produced about ten times more light than beta or gamma events, an unusual behavior that could improve dark matter searches.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Alpha light yield at 1 MeV is a cubic extrapolation from data above 2 MeV, so the factor-of-ten ratio is not yet quantitatively settled.","rationale":"The reader's weakest_assumption focuses on heat-channel linearity above 60 keV and uncalibrated alpha energy. My concern is adjacent but more specific: even granting a perfectly linear heat scale, the alpha LY at 1 MeV is not directly measured; it is an extrapolation of a cubic fit from 2-4.5 MeV. This is a standard extrapolation problem. Since the central claim in the abstract is a specific ratio at 1 MeV, this should be flagged. The reader's CONDITIONAL verdict already captures the need for better calibration, so I keep the verdict unchanged. The qualitative separation and detector performance claims are supported by the spectra; the numerical ratio is the fragile part.","tokens_in":11186,"tokens_out":3564,"duration_ms":40297,"concrete_test":"Re-fit the alpha band using the same binned data with at least three alternative models (e.g., linear with intercept, power law EL = A EH^k, and a spline) over the range 2-4.5 MeV, and also over a restricted 2.5-4.5 MeV range; if the predicted LY at 1 MeV varies by more than 0.3 keV/MeV (i.e., the alpha/beta ratio falls below 5 or rises above 20), the quantitative factor-of-ten claim should be downgraded pending a dedicated alpha measurement at ~1 MeV. The data are not public, so this test requires the authors to provide the fitted event distributions or their fit covariance.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 4.1 reports the alpha band between 2 and 4.5 MeV in Fig. 4 and fits it with EL = a EH^3 + b EH^2 + c EH, with a = -0.06(2), b = 0.54(1), c = 0.4(2), where EL is in keV and EH in MeV. Evaluating at EH = 1 MeV gives EL ≈ 0.88 keV, which is the basis of the headline alpha LY of 0.9 ± 0.2 keV/MeV. However, no alpha events are actually fitted below 2 MeV; the 1 MeV value is an extrapolation over a region where the cubic is not constrained by data. The polynomial is introduced as a convenient way to 'highlight the non-linearity' and is not derived from a scintillation model. A different model passing through the same 2-4.5 MeV data can change the 1 MeV prediction by more than the quoted statistical uncertainty. In contrast, the beta/gamma LY of 0.07 ± 0.01 keV/MeV comes from a linear fit over measured events down to 6 keV. Therefore the factor-of-ten alpha-to-beta/gamma ratio is set by an unvalidated extrapolation, not by a direct measurement at 1 MeV. The qualitative conclusion that alpha events produce more light than beta/gamma is likely robust, but the quantitative ratio is not. The authors acknowledge a 10-20% alpha energy scale uncertainty, but do not propagate the model/range uncertainty of the fit; this is the more serious systematic.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports the first operation of GaAs crystals as cryogenic scintillating calorimeters with dual heat and light readout within the DAREDEVIL project. Two GaAs wafers (4.3 g and 3.5 g) are read out by Ge-NTD thermistors for the phonon channel and by a Ge light detector with NTL amplification for the scintillation channel. The authors achieve baseline resolutions of 121 ± 2 eV and 44.5 ± 0.8 eV, resolve 55Fe Kα/Kβ lines, and, using simultaneous X-ray and 238U/234U alpha sources, observe two separated populations in the light-versus-heat scatter plot. They report light yields at 1 MeV of 0.07 ± 0.01 keV/MeV for beta/gamma events and 0.9 ± 0.2 keV/MeV for alpha events, interpreting the factor-of-ten ratio as an unusual alpha-induced light enhancement similar to ZnSe and ZnO.","tokens_in":11396,"tokens_out":3698,"duration_ms":47393,"significance":"If the quantitative light-yield values are correct, this is a valuable first demonstration of GaAs as a dual-readout scintillating calorimeter, with direct relevance to low-mass dark-matter searches via electron recoils and to particle identification in rare-event experiments. The paper's strengths are its clear experimental setup, the resolved low-energy X-ray spectra, the improved baseline resolution relative to the authors' previous work, and the explicit acknowledgment of energy-scale limitations. The dual-readout concept and the qualitative particle discrimination are convincingly demonstrated. However, the headline alpha light yield is obtained by extrapolating a cubic fit beyond the fitted energy range, so the quantitative factor-of-ten ratio is not yet firmly established; this limits the strength of the claims as currently written.","major_comments":[{"comment":"The headline alpha light yield of 0.9 ± 0.2 keV/MeV at 1 MeV is obtained by evaluating the cubic polynomial EL = a EH^3 + b EH^2 + c EH, with a = -0.06(2), b = 0.54(1), c = 0.4(2), at EH = 1 MeV. The text states that the alpha band lies between 2 and 4.5 MeV, so 1 MeV is outside the fitted region and the cubic is unconstrained there. Since c is the dominant term at 1 MeV and has a 50% fractional uncertainty, the quoted uncertainty does not include model or range extrapolation error. Please either report the alpha light yield at a directly measured energy, or add a robustness test showing how the 1 MeV value changes under alternative fit forms (e.g., linear and quadratic) over the same 2–4.5 MeV range, and propagate that spread into the quoted uncertainty.","section":"Section 4.1"},{"comment":"The heat-channel energy scale for beta/gamma events is calibrated with X-ray lines only up to about 60 keV, and linear response is assumed up to several MeV; the authors acknowledge this assumption but do not assign a systematic uncertainty to it. In addition, the alpha energy scale carries a stated 10–20% uncertainty because no clear alpha calibration peaks are available. These two systematics are not propagated into the quoted light-yield values or into the alpha-to-beta/gamma ratio. Please provide a systematic-error budget for the energy scales and state the resulting uncertainty on the factor-of-ten ratio.","section":"Section 4.1"},{"comment":"The conclusion states that the beta/gamma light yield was '0.07 ± 0.001 keV/MeV', while Table 2 and Section 4.1 report '0.07 ± 0.01 keV/MeV'; this is a numerical inconsistency that should be corrected. In addition, the concluding sentence repeats the factor-of-ten ratio as a settled value; given the extrapolation and energy-scale issues raised above, the conclusion should qualify this ratio as preliminary or as an extrapolated estimate.","section":"Section 5"}],"minor_comments":[{"comment":"The axis label 'events / events' is unclear; 'counts' or 'arbitrary units' would be more informative, and the inset showing the 6 keV–1.2 MeV region is too small to read in the current figure.","section":"Figure 4"},{"comment":"The label '3 RMS' should be '3σ' (three times the root-mean-square baseline resolution), to match the text.","section":"Figure 5"},{"comment":"References [38] and [39] appear to be duplicate entries for the same publication ('python package for dark matter scattering in dielectric targets'); please merge or renumber them.","section":"References"},{"comment":"The abstract states that both light yields are 'measured at 1 MeV'; for alpha events this is a nominal energy with a 10–20% uncertainty, so the abstract should either mention this caveat or phrase the alpha value as an extrapolated estimate.","section":"Abstract"},{"comment":"The text says the combination of gamma-ray and X-ray sources has a total activity of approximately 1 Bq; it would be clearer to give the activity of each source separately.","section":"Section 2"}],"recommendation":"major_revision","confidential_remarks":"This is a credible experimental first measurement that deserves publication after revision. The main technical concern is that the central quantitative claim—the alpha-to-beta/gamma light-yield ratio of about 10—rests on an extrapolation of a cubic fit beyond the fitted data range, and the relevant systematic uncertainties are not fully propagated. This is fixable by reanalysis and by qualifying the claims, so I recommend major revision rather than rejection. The duplicate references and the numerical typo in the conclusion should also be corrected."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The real news here is the proof of principle: a 4.3 g GaAs wafer operated at 10 mK with simultaneous heat and scintillation light readout, and a clear particle-discrimination band separated in the light-versus-heat scatter plot. That is new, and it matters for low-mass dark matter searches because GaAs combines a small band gap with a polar lattice. The 44.5 eV baseline resolution on the thinner crystal is also a solid step forward. The authors are honest about the calibration limits—they flag the 10–20% alpha energy scale uncertainty and the assumed linearity of the heat channel above 60 keV. The comparison to ZnSe and ZnO is appropriate and gives context to the anomalous alpha enhancement.\n\nThe soft spot is the headline quantitative claim. The alpha light yield of 0.9 ± 0.2 keV/MeV at 1 MeV is not a direct measurement. The alpha band is fitted only between 2 and 4.5 MeV with an empirical cubic polynomial, and the 1 MeV value is an extrapolation outside the fitted range. The quoted 0.2 keV/MeV uncertainty is statistical only; it does not include model or range uncertainty, which could easily be larger than the quoted error. The beta/gamma value of 0.07 ± 0.01 keV/MeV, by contrast, comes from a linear fit over measured events down to 6 keV, so it is on solid ground. The factor-of-ten ratio is therefore an extrapolated estimate, not a direct observation at 1 MeV. What is directly observed is that in the 2–4.5 MeV band, alpha events produce several times more light than beta/gamma events of the same heat energy. That qualitative inversion is robust and is the real finding. The specific 1 MeV number should be labeled as an extrapolation and given a more conservative uncertainty.\n\nThere is also a small mechanical misprint: the abstract gives the beta/gamma light yield as 0.07 ± 0.01 keV/MeV, while the conclusions give 0.07 ± 0.001 keV/MeV. The T able 2 in the text uses 0.07 ± 0.01, so the conclusions value looks like a typo.\n\nOverall, this is a competent experimental paper with a clear first result and honest caveats. The central demonstration holds; the quantitative alpha light yield at 1 MeV needs to be reframed as an extrapolation with a larger systematic uncertainty. It deserves a serious referee, and the paper should be accepted after a moderate revision that fixes the misprint and softens the 1 MeV claim to match what the data actually constrain.","headline":"A genuine first—GaAs as a dual-readout scintillating calorimeter—but the factor-of-ten alpha/beta light-yield ratio rests on an extrapolation and should be treated as provisional.","tokens_in":12135,"tokens_out":2212,"would_cite":false,"duration_ms":26209,"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 gallium arsenide crystal operated at about 10 mK separates alpha particles from beta/gamma radiation by reading heat and scintillation light together, with alpha events producing about ten times more light per unit energy.","keywords":["low-temperature calorimeters","gallium arsenide scintillator","dual heat-light readout","particle discrimination","alpha light-yield anomaly","quenching factor","dark matter searches","NTL amplification"],"falsifier":"Run the same dual-readout setup with a mono-energetic alpha source whose full-energy peaks appear in the heat spectrum, plus high-energy gamma calibration lines above 1 MeV to anchor the heat scale; if the alpha band then maps to a different true deposited energy, the reported ratio of about ten will not survive unchanged.","tokens_in":10917,"feed_emoji":"💡","tokens_out":12689,"duration_ms":125465,"temperature":0.7,"pith_summary":"This paper reports the first operation of a gallium arsenide (GaAs) crystal as a cryogenic scintillating calorimeter that reads out both the heat (phonon) signal and the scintillation light from every event. The central result is a clear separation between $\\alpha$ particles and $\\beta$/gamma radiation in the light-versus-heat plane, with a light yield at 1 MeV of $0.9 \\pm 0.2$ keV/MeV for $\\alpha$-induced events and $0.07 \\pm 0.01$ keV/MeV for $\\beta$/gamma events, a ratio of about ten. That ratio is anomalous: in most scintillators heavy particles produce less light than electrons of the same energy, not more. The authors argue the anomaly is worth pursuing because it may make low-energy nuclear recoils from light dark matter scattering easier to see, and they also report improved heat-channel resolution, with a 3.5 g crystal reaching a 44.5 eV baseline resolution and a 133.5 eV analysis threshold. A sympathetic reader would take the paper as establishing the first demonstration of this detector concept and a quantitative first look at the unusual luminescence behavior, with the ratio measurement itself flagged as preliminary.","feed_headline":"GaAs crystal splits alpha from beta/gamma by light yield","feed_subtitle":"At 10 mK, alpha events shine about ten times brighter than beta/gamma at 1 MeV, a reversal of typical scintillators.","key_machinery":"The load-bearing mechanism is the dual-readout bolometer itself: a GaAs wafer (4.3 g or 3.5 g) instrumented with a germanium neutron-transmutation-doped thermistor records the heat channel, while a thin germanium plate mounted 10 mm away records scintillation photons, with an applied bias providing NTL (voltage-assisted) amplification that improved the light detector's baseline resolution from 60.4 eV to 5.2 eV. Particle identification comes from comparing the light energy $E_L$ to the heat energy $E_H$ event by event: $\\beta$/gamma events lie on a straight line $E_L = a' E_H$ with $a' = 0.069(1)$ keV/MeV, while $\\alpha$ events follow $E_L = a E_H^3 + b E_H^2 + c E_H$ with $a = -0.06(2)$, $b = 0.54(1)$, $c = 0.4(2)$, whose slope near 1 MeV gives the ten-fold higher light yield. The physical expectation that GaAs would behave this way comes from its direct band gap of 1.42 eV and its previously reported cryogenic scintillation of about 2 photons per keV, which in principle yields a large photon count per low-energy recoil.","core_discovery":"On its own terms, the paper claims that GaAs can serve as a dual-readout cryogenic scintillating calorimeter: the same energy deposition produces a phonon pulse in the GaAs wafer, sensed by a neutron-transmutation-doped germanium thermistor, and a scintillation pulse in a nearby germanium light detector operated with NTL (voltage-assisted) amplification. Calibration runs with an X-ray source and a uranium $\\alpha$ source produced a light-versus-heat scatter plot with two well-separated bands. Fitting those bands gives a light yield for $\\beta$/gamma events of $0.07 \\pm 0.01$ keV/MeV (about 0.05 photons/keV at the 840 nm GaAs emission line) and for $\\alpha$ events of $0.9 \\pm 0.2$ keV/MeV, both evaluated at 1 MeV of heat energy, i.e., an $\\alpha$-to-$\\beta$/gamma ratio of about ten. The paper explicitly notes that this ratio carries an additional 10--20% uncertainty because the $\\alpha$ energy scale rests on an extrapolation from X-ray calibration lines below 60 keV with no distinct $\\alpha$ peaks. It also reports improved heat-channel performance over its earlier measurement: $140 \\pm 8$ eV resolution at 5.9 keV for the 4.3 g crystal and $59 \\pm 1$ eV for the 3.5 g crystal, with baselines of $121 \\pm 2$ eV and $44.5 \\pm 0.8$ eV respectively.","pith_inferences":["If the alpha enhancement is real, the standard explanation of quenching as saturation of luminescence centres under high ionisation density is incomplete for GaAs; a density-dependent recombination channel involving defects or excitonic states would be needed.","Because the alpha energy scale carries a 10--20% uncertainty, the ratio is best read as a preliminary value; a moderate shift in the assigned alpha energies would change the ratio, though a ratio well above one would likely remain.","The same light-versus-heat technique could be applied to other small-band-gap polar semiconductors, such as indium phosphide or aluminium arsenide, that are predicted to be sensitive to dark-photon absorption.","A direct testable extension is to expose GaAs to a neutron source to produce nuclear recoils; if recoil events fall in the alpha-enhanced band rather than on the beta/gamma line, the dark-matter motivation is confirmed."],"forward_implications":["Event-by-event alpha/beta-gamma separation is demonstrated in GaAs, so alpha-emitting contamination can be tagged and rejected as background in a future dark-matter or rare-event search.","The anomalous alpha-enhanced light yield implies that nuclear recoils, which also produce dense ionisation tracks, may scintillate more than electron recoils, potentially improving sensitivity to light dark matter scattering on gallium or arsenic nuclei.","Thresholds of 360 eV (4.3 g crystal) and 133.5 eV (3.5 g crystal) bring sub-keV recoil detection within reach, the regime relevant for sub-GeV dark matter and dark-photon absorption.","The NTL-amplified germanium light detector improves light-channel baseline resolution by a factor of 12, showing that the dual-readout approach can handle very faint scintillation signals.","If the anomaly survives a proper alpha calibration, GaAs joins ZnSe and ZnO as another cryogenic scintillator with a quenching factor above one, a small family with reversed alpha/beta response."],"supporting_citations":[{"why":"It supplies the GaAs cryogenic scintillation baseline, about 2 photons/keV under X-rays at 10 K and 840 nm emission, used to convert measured light to photons/keV and to justify the dual-readout design.","marker":"[50]"},{"why":"It is the authors' earlier GaAs low-temperature calorimeter measurement, whose 1.5 keV threshold this work improves to 360 eV and whose setup is extended with light readout.","marker":"[35]"},{"why":"It defines the optimum-filter algorithm used to estimate event energies from the digitised pulse streams.","marker":"[31]"},{"why":"It documents the neutron-transmutation-doped germanium thermistors used as thermal sensors on both the GaAs absorbers and the light detector.","marker":"[34]"},{"why":"It establishes voltage-assisted amplification of the light detector signal without adding noise, the effect used to improve light-channel resolution.","marker":"[41]"},{"why":"It underpins the NTL amplification mechanism that improved the light-channel baseline resolution by a factor of 12.","marker":"[43]"},{"why":"It provides the theoretical framework for sub-GeV dark matter detection with scintillating targets that motivates the choice of small-band-gap GaAs.","marker":"[26]"},{"why":"It supplies the comparison case of ZnO scintillating bolometers with a quenching factor above one, the closest precedent for the anomalous alpha/beta ratio reported here.","marker":"[11]"},{"why":"It gives the typical quenching-factor range of 0.2--0.6 for alpha versus beta/gamma in low-temperature scintillators, the baseline against which the measured ratio of about ten stands out.","marker":"[47]"},{"why":"It is an earlier study of cryogenic scintillation properties of n-type GaAs that supports the material's expected light response and low-temperature behaviour.","marker":"[24]"}],"fun_headline_variants":["GaAs scintillator flips light ratio: alpha shines 10x brighter","First GaAs calorimeter sees alpha light 10x beta/gamma","GaAs dual-readout crystal: alpha glow beats beta/gamma by 10x","GaAs detector reverses light yield: alpha outshines beta/gamma"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The ten-to-one alpha-to-beta/gamma light-yield ratio rests on a heat energy scale calibrated only up to 60 keV and assumed linear to several MeV, with alpha-particle energies inferred from that scale at a stated 10--20 percent uncertainty; if the scale is wrong for alphas, the ratio changes.","fun_headline_variants_meta":{"raw":{"variants":["GaAs scintillator flips light ratio: alpha shines 10x brighter","First GaAs calorimeter sees alpha light 10x beta/gamma","GaAs dual-readout crystal: alpha glow beats beta/gamma by 10x","GaAs detector reverses light yield: alpha outshines beta/gamma"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000396,"raw_usage":{"total_tokens":2182,"prompt_tokens":1156,"completion_tokens":1026,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":772,"completion_tokens_details":{"reasoning_tokens":943}},"tokens_in":772,"tokens_out":1026,"duration_ms":9218,"temperature":1.0,"reasoning_tokens":943,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T19:29:33.522327+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same dual-readout setup with a mono-energetic alpha source whose full-energy peaks appear in the heat spectrum, plus high-energy gamma calibration lines above 1 MeV to anchor the heat scale; if the alpha band then maps to a different true deposited energy, the reported ratio of about ten will not survive unchanged.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It supplies the GaAs cryogenic scintillation baseline, about 2 photons/keV under X-rays at 10 K and 840 nm emission, used to convert measured light to photons/keV and to justify the dual-readout design."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It is the authors' earlier GaAs low-temperature calorimeter measurement, whose 1.5 keV threshold this work improves to 360 eV and whose setup is extended with light readout."},{"cited_title":"Gatti and P","cited_arxiv_id":null,"evidence_quote":"It defines the optimum-filter algorithm used to estimate event energies from the digitised pulse streams."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It documents the neutron-transmutation-doped germanium thermistors used as thermal sensors on both the GaAs absorbers and the light detector."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It establishes voltage-assisted amplification of the light detector signal without adding noise, the effect used to improve light-channel resolution."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It underpins the NTL amplification mechanism that improved the light-channel baseline resolution by a factor of 12."},{"cited_title":"Direct detection of sub-gev dark matter with scintillating targets","cited_arxiv_id":null,"evidence_quote":"It provides the theoretical framework for sub-GeV dark matter detection with scintillating targets that motivates the choice of small-band-gap GaAs."},{"cited_title":"Zno-based scin- tillating bolometers: new prospects to study double beta decay of 64zn","cited_arxiv_id":null,"evidence_quote":"It supplies the comparison case of ZnO scintillating bolometers with a quenching factor above one, the closest precedent for the anomalous alpha/beta ratio reported here."},{"cited_title":"Scintillation in low-temperature par- ticle detectors","cited_arxiv_id":null,"evidence_quote":"It gives the typical quenching-factor range of 0.2--0.6 for alpha versus beta/gamma in low-temperature scintillators, the baseline against which the measured ratio of about ten stands out."},{"cited_title":"Derenzo, E","cited_arxiv_id":null,"evidence_quote":"It is an earlier study of cryogenic scintillation properties of n-type GaAs that supports the material's expected light response and low-temperature behaviour."}],"review_version":1}