{"id":"9e3941ac-3d8d-4530-a082-bf9038993cf1","arxiv_id":"2507.15732","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"NTL-amplified germanium light detectors in a 10-crystal cryogenic array achieve a signal-to-noise gain near 9, sub-millisecond rise time, and projected pile-up background close to the CUPID goal.","lead":"Ten germanium light detectors that amplify faint scintillation signals by drifting charge carriers through an electric field were tested in a millikelvin array, reaching a signal-to-noise gain near nine and noise around ten eV. The results indicate this amplifier technology can help future neutrinoless double beta decay experiments reject background well enough to approach their sensitivity goals.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Central extrapolation to full NTL electrode coverage relies on an unvalidated linear area-scaling law (Eq. 5.1); if this overestimates the gain, the CUPID pile-up projections weaken.","rationale":"The experimental core is credible and well documented: eight NTL-LDs operated at 80 V, sub-millisecond rise times, and baseline noise as low as 10 eV in the far calibration geometry. The reader's CONDITIONAL verdict is appropriate. I agree with the reader's first identified weakness—the linear coverage-scaling law of Eq. 5.1—as the most load-bearing assumption for the central CUPID projection. I am less convinced by the second identified weakness (combining close light yield with far noise): Section 5.2 explicitly assumes flipping the LD so the electrode side faces the close crystal, and under that assumption the relevant baseline noise is indeed the far-calibration value measured with the higher NTL gain, so the combination is internally consistent. The principal unresolved risk is that G_NTL per unit area may change with the electrode geometry, making f_scale = 1.7–2.0 optimistic. A direct measurement of a full-coverage prototype would settle this. Because the extrapolation is flagged as an assumption rather than a hidden error, and the measured results support the technology direction, the conditional verdict stands without change.","tokens_in":24525,"tokens_out":8773,"duration_ms":96524,"concrete_test":"Measure the NTL gain of the existing full-coverage electrode prototypes (semi-square or dual-spiral, Fig. 12a,b) in the same CROSS pulse-tube cryostat at 22 mK and 80 V, with the electrode side facing a Li2MoO4 crystal, and compare the effective SNR gain with G_full = [G_CROSS − (1−ξ_CROSS)]/ξ_CROSS ≈ 14.7 predicted from Eq. 5.1. If the measured full-coverage gain is below 14.7 (i.e., f_scale < 1.7), recompute the Table 6 background indices with the empirical gain. A faster auxiliary check is a focused-LED scan across covered and uncovered areas of the present 56%-coverage wafer to test directly whether the energy-weighted gain follows the linear area average assumed by Eq. 5.1.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing step is the extrapolation from the measured 56%-coverage gain to full electrode coverage. Eq. 5.1 assumes G_eff = (1−ξ) + ξ·G_NTL with a single uniform G_NTL, so the 80-V far-side gain G_CROSS = 8.66 is converted into G_full ≈ 14.7 and f_scale ≈ 1.7 (and up to 2.0 for the CUPID configurations with the higher light yield). This is an area-weighted linear model, but the paper itself states that NTL is a surface phenomenon whose efficiency depends on field lines and trapping centers, and the close/far gain difference (3.94 versus 8.66) shows strong sensitivity to where charge is generated. The proposed full-coverage designs (semi-square and dual-spiral, Fig. 12) are not simple enlargements of the circular 56% pattern; they change the field geometry, so the per-area NTL gain need not remain constant. If the true full-coverage gain is lower—say f_scale = 1.2 instead of 1.7—the mean SNR in Table 6 drops and the simulated pile-up background index moves from about 6×10^-5 ckky toward 8×10^-5 ckky, further from the 0.5×10^-4 ckky goal quoted in the Conclusions. The central claim therefore rests entirely on an unvalidated scaling assumption. The measured detector performance itself is credible; the weakness is in the extrapolation step.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript reports the fabrication, underground operation, and characterization of ten 45x45x0.3 mm Ge light detectors with Neganov-Trofimov-Luke (NTL) thermal signal amplification, mounted with Li2MoO4 and TeO2 bolometers in the CROSS mechanical structure in a pulse-tube cryostat at the Canfranc underground laboratory. At 22 mK and 80 V electrode bias, the authors measure a mean SNR gain of about 9 for light entering the electrode side, a baseline noise of about 10 eV RMS, and sub-millisecond rise times around 0.5 ms. Using synthetic pile-up pulses injected into the measured noise baselines, they simulate the 2νββ pile-up background index for four CUPID-related configurations and conclude that, assuming full NTL electrode coverage, the devices would suppress pile-up close to CUPID's design goal of 0.5×10^-4 counts/keV/kg/yr.","tokens_in":24847,"tokens_out":13957,"duration_ms":145464,"significance":"The demonstrated batch-level reproducibility of NTL-assisted light detectors in an underground pulse-tube cryostat is a substantial technical step, and the use of measured noise baselines in the pile-up simulations is a genuine strength. If the extrapolations are validated, the paper would provide a concrete route toward rejecting one of the dominant backgrounds in CUPID. The calibration strategy (external X-ray peaks for the absolute energy scale plus scintillation-light coincidences for the NTL gain) and the internal consistency of Tables 2 and 4 give confidence in the measured detector parameters. The main limitations are the unvalidated linear scaling law for the NTL gain and the optimistic combination of noise and light-yield figures, both of which directly affect the central Conclusion.","major_comments":[{"comment":"The scaling factor fscale rests on a linear area-weighted model of the NTL gain, G_eff = (1-ξ) + ξ·G_NTL, with a single uniform G_NTL over the covered area. This model is not validated by the data. The authors themselves note in Section 4.4 that NTL is a surface phenomenon whose efficiency depends on field lines and trapping centers, and the measured effective gains, 3.94 for close illumination versus 8.66 for far illumination, show a strong dependence on where the charge is generated. The semi-square and dual-spiral electrode patterns of Fig. 12 are not simple enlargements of the circular 56%-coverage pattern, so the per-area NTL gain cannot be assumed constant. Because fscale multiplies the SNR values used in Table 6 and therefore directly determines the predicted background index, the central projection is not yet established. The authors should provide a sensitivity scan of BI versus fscale or validate the scaling with measurements of the new electrode geometries.","section":"Section 5.3, Eqs. (5.1)-(5.5)"},{"comment":"The projected SNR of about 90 at Qββ combines the 'close' light yield (0.30 keV/MeV) with the 10 eV baseline noise quoted from this work. Table 4 reports a mean σ_baseline of 23.0 eV for the close calibration and 9.5 eV for the far calibration; the 10 eV value is the far-geometry result, obtained when the scintillation light enters the electrode side and the NTL gain is high. Flipping the LD so that the electrode side faces the crystal is a reasonable design choice, but the specific combination of close light collection with the high-gain, low-noise configuration has not been measured, and the electrode pattern and antireflective coating on the entrance side could modify both the light collection and the pulse shape. The SNR entries in Table 6 should therefore be treated as an optimistic projection, and the associated systematic uncertainty should be quantified or the claim softened.","section":"Section 5.2 and Table 4"},{"comment":"The Conclusions state that the devices would enable efficient discrimination of pile-ups close to CUPID's background goals, defined as a pile-up contribution below 0.5×10^-4 ckky. The simulated values in Table 6 are all above that goal: mean BI values are 6.7×10^-5, 7.0×10^-5, and 6.1×10^-5 ckky for configurations (II), (III), and (IV), respectively, and the best individual value is 5.3×10^-5 ckky. The claim is therefore quantitatively stronger than the simulation supports. The paper should quote the simulated BIs and the residual gap relative to the goal, or explicitly present the technology as approaching but not yet meeting the CUPID target.","section":"Conclusions and Table 6"}],"minor_comments":[{"comment":"The sentence 'we obtained a gain of around 9 in the signal-to-noise ratio' should specify that this mean gain of 8.66 is for the far-illumination geometry; the close geometry gives a mean gain of 3.94 (Table 4), and the distinction matters for the CUPID projection.","section":"Abstract"},{"comment":"The terms 'Close' and 'Far' should be defined in the caption as referring to the side of the Ge wafer on which the scintillation light is incident, since the difference between the two cases is central to the gain and noise argument.","section":"Table 4 and Fig. 11"},{"comment":"The formula for the background index would benefit from a derivation or a unit check; the roles of P, A, m, Time, and Δt_r=100% are stated in prose, but the typesetting of the equation is ambiguous about the division by the crystal mass.","section":"Section 5.3, Eq. (5.7)"},{"comment":"Several citations refer to documents marked 'submitted', 'in preparation', or conference proceedings (e.g., [7], [10], [42], [48], [65]); these should be updated to published versions if they become available.","section":"References"},{"comment":"There are minor typographical errors, such as 'at at Physik-Institut' in footnote 5 and some duplicated location text in the author affiliations; a final proofread is recommended.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The measured detector performance is credible and valuable, and the paper should not be rejected on those grounds. The main risk is that the Conclusions go beyond what the current data and simulations establish. Major revision is appropriate: the authors can address the concerns by tempering the CUPID projections, adding a sensitivity analysis of fscale and of the SNR assumptions, and explicitly reporting the residual gap in Table 6."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a solid, useful engineering paper from the CROSS/BINGO/CUPID line. The real content is the measured performance of ten NTL-amplified Ge light detectors in a pulse-tube cryostat: eight devices operated at 80 V without leakage, mean rise time 0.54 ms, and effective SNR gain roughly a factor 4-12 depending on which side faces the scintillator. Those numbers are credible and internally consistent. The pile-up projections for CUPID are the value-added part, but they rest on a linear area-scaling law (Eq. 5.1) that is not validated, and on combining the best measured noise (far-side, 10 eV) with the best light yield (close-side, 0.30 keV/MeV). That makes the CUPID numbers somewhat optimistic. The central conclusion that NTL-LDs are a viable route to approaching CUPID's pile-up goal probably survives, but only if the extrapolation holds. What is actually new: the 10-device batch in the CROSS structure, the electrode-coverage surface effect (large close/far gain difference), sub-millisecond rise times, and pile-up simulations on real noise baselines for four CUPID configurations. The measurements are careful: X-ray calibration at 0 V, scintillation-light calibration in NTL mode, per-channel working points, and an explicit check that their simplified injection method agrees with the earlier approach to about 1%. The paper also flags that the current circular electrode mask only covers 56% of a square wafer, and it does not hide the spread in detector-to-detector noise. Soft spots, in proportion: (1) Eq. 5.1 is an unvalidated assumption. The authors themselves say NTL efficiency depends on field lines and trapping centers, so an area-weighted linear extrapolation from 56% to 100% coverage is a plausible first guess, not a demonstrated result. If the true gain scaling is worse, the background index moves from about 0.6-0.7 times 10^-4 ckky toward 0.8 times 10^-4 ckky, further from the 0.5 times 10^-4 goal. This should be labeled clearly as an assumption, not a measured outcome. (2) The SNR ~90 at Q_beta_beta mixes close-geometry light yield with far-geometry 10 eV noise; Table 4 shows the close-geometry mean noise is 23 eV. That is an optimistic blend, and the paper does not clearly flag the mix. (3) There is a minor typo-level inconsistency: the CUPID goal appears as both 0.5 times 10^-5 and 0.5 times 10^-4 ckky in the text. (4) One of ten LDs was lost to electronics and one showed leakage above 50 V; batch consistency is good but not perfect. None of these are fatal. The measured detector performance is directly calibrated, the pile-up simulation is documented well enough to be checked, and the conclusions are conditionally phrased. The load-bearing assumption is the only thing I would want tested before accepting the CUPID projection at face value. Who this is for: anyone in bolometric 0nu_beta_beta or low-temperature light detection. It deserves a serious referee. I would accept it for review and ask for a sensitivity scan of the BI projection to f_scale, and for a clearer separation of measured versus extrapolated numbers in the abstract and conclusions. I would cite it in my own writing on NTL light detectors.","headline":"Solid measured NTL-LD batch performance; the CUPID projections are optimistic but worth engaging, not dismissing.","tokens_in":957,"tokens_out":1157,"would_cite":true,"duration_ms":52599,"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 batch of ten voltage-amplified germanium light detectors reached the speed and sensitivity needed to reject double-beta pile-up background in a tonne-scale experiment.","keywords":["neutrinoless double-beta decay","Neganov-Trofimov-Luke effect","cryogenic light detectors","thermal signal amplification","pile-up rejection","pulse-shape discrimination","germanium bolometers","lithium molybdate scintillators"],"falsifier":"Measure the effective gain and baseline noise of a full-coverage wafer with the electrode side facing the scintillator, at the same 80 V bias: the paper's projection requires a gain about $1.7\\times$ the $8.66$ measured at 56% coverage and a baseline noise near 10 eV in that geometry; if the noise remains near 23 eV, the SNR at the double-$\\beta$ Q-value drops from about 90 to 40 and the pile-up background misses the target.","tokens_in":24296,"feed_emoji":"❄️","tokens_out":13556,"duration_ms":137004,"temperature":0.7,"pith_summary":"This paper reports a batch of ten germanium light detectors that amplify their own thermal signals by applying a voltage across patterned electrodes, using the Neganov-Trofimov-Luke effect. Operated together at millikelvin temperatures in a pulse-tube cryostat, the detectors achieved a signal-to-noise gain of about a factor 9 at 80 V, baseline noise of order 10 eV, and sub-millisecond rise times. The authors argue this combination of speed and sensitivity is what is needed to reject random coincidences of two-neutrino double-beta decays, a leading background in searches for neutrinoless double-beta decay of molybdenum-100. Extrapolating their measured performance to full electrode coverage, they estimate the residual pile-up background would land close to the design goal of the planned tonne-scale experiment.","feed_headline":"Ninefold signal gain puts cryogenic detectors near target","feed_subtitle":"Ten germanium light sensors ran at 80 V, reached ~10 eV noise and sub-ms response, cutting pile-up background near goal.","key_machinery":"The central mechanism is the Neganov-Trofimov-Luke effect: electron-hole pairs created by absorbed scintillation light drift in an applied electric field and deposit extra heat before collection, so the thermal pulse seen by the phonon sensor is amplified by roughly $G_{NTL}\\approx e V_{NTL}\\eta/\\epsilon$, where $V_{NTL}$ is the electrode bias, $\\epsilon\\approx 3$ eV is the pair-creation energy in germanium, and $\\eta<1$ accounts for trapping losses. The devices use concentric aluminium ring electrodes covering 56% of the square wafer, with the electrode side facing away from the scintillating crystal in the present geometry. The pile-up projection is carried by an area-weighted gain formula, which assumes the effective gain of a partially covered wafer scales linearly with the covered fraction, and by simulations that apply pulse-shape-discrimination cuts to synthetic pile-up events placed on real noise traces.","core_discovery":"The paper's central claim is that NTL-assisted germanium light detectors can be produced as a uniform batch of ten and operated together in a pulse-tube cryostat, with an 80 V electrode bias improving the signal-to-noise ratio by a mean factor around 9 relative to the unbiased devices. In the best calibration geometry the baseline noise RMS reaches about 10 eV, and strong polarization of the germanium thermistors shortens the rise time to roughly 0.5 ms. From these measured values, pile-up simulations inject synthetic two-neutrino double-$\\beta$ pulses onto the recorded noise baselines and show that a full-electrode-coverage version of the same detector would suppress random-coincidence background in a tonne-scale molybdenum-100 experiment close to the design goal of $0.5\\times10^{-4}$ counts/keV/kg/yr.","pith_inferences":["The linear-scaling assumption for partial electrode coverage is the main unmeasured step: if a full-coverage wafer does not reach about 1.7 times the gain seen at 56% coverage, the projected background index is optimistic.","Pairing the 10 eV baseline noise from the 'far' calibration with the 0.30 keV/MeV light yield of the 'close' geometry is an optimistic combination; using the 23 eV mean noise of the close geometry would put the SNR at the double-beta Q-value near 40 rather than 90.","The large spread in residual background across nominally identical detectors suggests that the shape of vibration noise, not just its RMS, controls pile-up rejection, so mechanical decoupling or noise-cancellation work may buy as much as higher gain.","The same electrode-on-wafer recipe should transfer to other scintillating bolometer materials and phonon sensors, so the result is not limited to molybdenum-bearing detectors."],"forward_implications":["With full electrode coverage, the projected mean pile-up background index is about $6\\times10^{-5}$ counts/keV/kg/yr, with several of the eight detectors approaching the $5\\times10^{-5}$ design goal.","Because the thermistor readout and room-temperature front-end electronics are unchanged, existing bolometric arrays could adopt the NTL upgrade with added cabling, power supplies, and optical-fiber regeneration only.","Eight of nine operational devices held the 80 V bias without leakage current and reached similar gains, supporting reproducible production at the scale of a large array.","Operating at 22 mK with strong thermistor polarization shortens rise times to around 0.5 ms while costing only about 30% in signal-to-noise, improving the time separation available for pile-up rejection."],"supporting_citations":[{"why":"Defines the voltage-assisted thermal amplification that the light detectors exploit, the paper's core mechanism.","marker":"[29, 30]"},{"why":"Supplies the electrode pattern, evaporation recipe, and earlier NTL light-detector gain and noise results that the batch reproduces at 80 V.","marker":"[35]"},{"why":"Supplies the pile-up injection and pulse-shape discrimination methodology, and the SNR/rise-time point the paper compares against.","marker":"[39]"},{"why":"Establishes two-neutrino double-beta random coincidences as a background source and gives the formula used to convert rejection power into a background index.","marker":"[36]"},{"why":"Sets the tonne-scale experiment's detector geometry, light-yield assumptions, and the $0.5\\times10^{-4}$ counts/keV/kg/yr pile-up goal the extrapolations target.","marker":"[7]"},{"why":"Validates the low-mass mechanical holder and provides the 0.30 keV/MeV close-geometry light yield used for the square-wafer configurations.","marker":"[28]"},{"why":"Provides the 0.36 keV/MeV light yield for the octagonal baseline-detector configurations.","marker":"[33]"}],"fun_headline_variants":["Cryogenic light detectors gain 9x signal-to-noise via NTL","NTL boost gives Ge light detectors 9x SNR at 80 V","Pile-up background cut by NTL-boosted cryogenic detectors","Batch of Ge light detectors hit 10 eV noise with NTL","Near-target sensitivity from NTL-boosted Ge detectors"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The projection rests on two untested assumptions: that the signal gain grows linearly with the fraction of the wafer covered by electrodes (so the gain measured at 56% coverage can be scaled to full coverage), and that the best measured baseline noise can be combined with the best measured light yield.","fun_headline_variants_meta":{"raw":{"variants":["Cryogenic light detectors gain 9x signal-to-noise via NTL","NTL boost gives Ge light detectors 9x SNR at 80 V","Pile-up background cut by NTL-boosted cryogenic detectors","Batch of Ge light detectors hit 10 eV noise with NTL","Near-target sensitivity from NTL-boosted Ge detectors"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001158,"raw_usage":{"total_tokens":4839,"prompt_tokens":1031,"completion_tokens":3808,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":647,"completion_tokens_details":{"reasoning_tokens":3714}},"tokens_in":647,"tokens_out":3808,"duration_ms":27855,"temperature":1.0,"reasoning_tokens":3714,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T15:26:00.406611+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the effective gain and baseline noise of a full-coverage wafer with the electrode side facing the scintillator, at the same 80 V bias: the paper's projection requires a gain about $1.7\\times$ the $8.66$ measured at 56% coverage and a baseline noise near 10 eV in that geometry; if the noise remains near 23 eV, the SNR at the double-$\\beta$ Q-value drops from about 90 to 40 and the pile-up background misses the target.","supporting_citations":[{"cited_title":"Novati et al., Charge-to-heat transducers exploiting the Neganov-Trofimov-Luke effect for light detection in rare-event searches , Nucl","cited_arxiv_id":null,"evidence_quote":"Supplies the electrode pattern, evaporation recipe, and earlier NTL light-detector gain and noise results that the batch reproduces at 80 V."},{"cited_title":"Enhanced light signal for the suppression of pile-up events in Mo-based bolometers for the $0\\nu\\beta\\beta$ decay search","cited_arxiv_id":"2302.13944","evidence_quote":"Supplies the pile-up injection and pulse-shape discrimination methodology, and the SNR/rise-time point the paper compares against."},{"cited_title":"Chernyak et al., Random coincidence of 2ν2β decay events as a background source in bolometric 0ν2β decay experiments, Eur","cited_arxiv_id":null,"evidence_quote":"Establishes two-neutrino double-beta random coincidences as a background source and gives the formula used to convert rejection power into a background index."},{"cited_title":"CUPID, the CUORE Upgrade with Particle IDentification","cited_arxiv_id":"2503.02894","evidence_quote":"Sets the tonne-scale experiment's detector geometry, light-yield assumptions, and the $0.5\\times10^{-4}$ counts/keV/kg/yr pile-up goal the extrapolations target."},{"cited_title":"A novel mechanical design of a bolometric array for the CROSS double-beta decay experiment","cited_arxiv_id":"2405.18980","evidence_quote":"Validates the low-mass mechanical holder and provides the 0.30 keV/MeV close-geometry light yield used for the square-wafer configurations."},{"cited_title":"Innovating Bolometers' Mounting: A Gravity-Based Approach","cited_arxiv_id":"2503.04481","evidence_quote":"Provides the 0.36 keV/MeV light yield for the octagonal baseline-detector configurations."}],"review_version":1}