{"id":"e0b0c449-7e08-4dd9-8249-c24896d41d59","arxiv_id":"2505.08067","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"The delayed photon noise that follows scintillation pulses in liquid xenon detectors is mostly ultraviolet-induced fluorescence from quartz photosensor windows, not from the xenon or the detector walls.","lead":"This paper identifies delayed 'afterglow' photons in liquid xenon dark matter detectors as ultraviolet-induced fluorescence from the quartz windows of the light sensors. It matters because this previously unexplained background limits searches for low-mass dark matter and solar neutrinos, and identifying the source points toward detector designs that can reduce or avoid it.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The quartz attribution rests on a single two-sensor comparison: if the windowless S13370 differs from the S13371 in photon detection efficiency, package, or optical coupling, the ×20 reduction in delayed photons is not uniquely attributable to the missing quartz window.","rationale":"The reader's weakest assumption already identifies the sensor-comparison confound, and my read agrees that this is the single most load-bearing concern. The Cherenkov vacuum test and material-substitution controls rule out PTFE, electrodes, xenon purity, and many other candidates, but they cannot distinguish quartz from other photosensor materials. The Fig. 4 comparison is the only evidence specifically pointing to quartz, and it uses two different Hamamatsu device models with different packages, dark rates, and likely different photon detection efficiencies. A direct same-sensor control or a quartz-plate addition to the windowless sensor would settle whether quartz is truly the dominant source. The paper also contains internal caveats: Sec. III states that the key question of whether quartz fluorescence remains dominant at 5–1000 ms is still open, and the abstract's unqualified wording reaches beyond the directly measured t < 5 ms regime. However, the strongest claim as framed is specifically about t < 5 ms, and within that regime the sensor-comparison confound is the critical weakness. Because the reader's conditional verdict already reflects this uncertainty, no change to the verdict is needed.","tokens_in":8446,"tokens_out":4384,"duration_ms":47393,"concrete_test":"Repeat the optically isolated two-half measurement using the same sensor model in both halves: either remove the quartz window from an S13371 (or use a bare S13371 die) and compare it with an intact S13371, or place a quartz plate in front of the windowless S13370 and verify that the delayed-photon rate rises to the level seen with the intact S13371. Calibrate the relative photon detection efficiency at 175 nm for both sensors and use identical optical geometry. Additionally, directly test the ceramic hypothesis by masking the S13370 ceramic package with opaque material or black paint and checking whether the k ≈ 1/4 residual component disappears; if it persists, the ceramic attribution is unsupported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that quartz window fluorescence dominates t < 5 ms delayed photon noise rests primarily on the Fig. 4 comparison between a quartz-windowed Hamamatsu S13371 and a windowless S13370. These are different device families: the S13370 has a different ceramic package with roughly 4× more ceramic facing the xenon, two of the four tested S13370 sensors were excluded for elevated leakage current, and the remaining two differed by 53% in dark count rate. The authors note the S13371's random photon background is an order of magnitude lower than the S13370's, so dark rate alone cannot explain the higher delayed rate in the S13371, but the two sensors almost certainly have different photon detection efficiencies, spectral responses, and optical geometries at 175 nm. No control isolates quartz: the vacuum Cherenkov test (Sec. II.B) demonstrates that quartz can fluoresce under UV, but does not show that it dominates in liquid xenon; the material-substitution tests narrow the suspects to 'the photosensors' but cannot distinguish the window from the ceramic, package, or sensor face. The residual k ≈ 1/4 component in the windowless S13370 is assigned to ceramic fluorescence without a direct measurement. The argument also assumes the aluminum divider provides complete optical isolation, which is not demonstrated. Thus the dominant-source attribution is supported by one non-isolated sensor comparison.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper addresses the long-standing puzzle of delayed photon noise in liquid xenon particle detectors, which follows a power-law time profile after scintillation pulses and currently limits low-threshold dark matter and neutrino experiments. The authors describe a series of tests in a small liquid xenon testbed at LBNL, in which they systematically remove or substitute materials suspected of fluorescing (PTFE, electrodes, sensor types) and show that the delayed photon rate is insensitive to all such changes except the photosensor itself. They then use a vacuum Cherenkov measurement to demonstrate that quartz fluoresces under UV excitation with a similar magnitude and time profile, and, in a key comparison, report that a windowless Hamamatsu S13370 SiPM observes a factor-20 smaller delayed photon rate than a quartz-windowed S13371 SiPM. On this basis they conclude that the dominant component of the delayed photon noise is UV-induced fluorescence of quartz photosensor windows.","tokens_in":8670,"tokens_out":2990,"duration_ms":31483,"significance":"If the central attribution is correct, this is an important result: it identifies a background mechanism that has become the dominant limitation in searches for low-mass dark matter and in 8B solar neutrino coherent-scattering measurements, and it points to a concrete design improvement for next-generation detectors such as XLZD. The paper is commendable for combining multiple material-substitution controls, a vacuum Cherenkov fluorescence measurement, and power-law characterization with explicit acknowledgment of systematic uncertainties in the fitted parameters. The authors also openly state the limitation that the direct measurement covers t < 5 ms and that the 5-1000 ms regime remains unprobed. The main weakness is that the central attribution rests on a single comparison between two different sensor products, and the paper's abstract draws a broader conclusion than the data directly support.","major_comments":[{"comment":"The central claim that quartz window fluorescence is the dominant source rests on a single comparison between the quartz-windowed Hamamatsu S13371 and the windowless S13370. These are different sensor families with different ceramic packages (the text notes roughly 4x more ceramic facing the xenon in the S13370), different dark-count and leakage-current behavior, and likely different photon detection efficiency and spectral response at 175 nm. The factor-20 reduction is therefore not uniquely attributable to the absence of the quartz window. A direct control is needed, for example adding a quartz plate to the windowless sensor or measuring a windowed and windowless version of the same sensor, to remove these confounds.","section":"Section II.C, Fig. 4"},{"comment":"The aluminum divider is stated to optically isolate the two halves of the active region, but no test of that isolation is presented. If the isolation is imperfect, the lower rate in the S13370 could be affected by different optical coupling or by photons crossing the divider, rather than by the absence of quartz. A cross-check using a known light source in one half while monitoring the other, or instrumenting both halves with identical sensors, would make the comparison unambiguous.","section":"Section II.C"},{"comment":"The abstract states that the dominant component of the delayed photon noise is due to UV-induced fluorescence of quartz photosensor windows, without time qualification. However, the direct measurement in this work extends only to t < 5 ms, and the paper itself says that the 5-1000 ms regime remains a key question with only 'suggestive if inconclusive' evidence. The abstract should be qualified to the measured timescale, or the later-time claim should be supported by a direct measurement, since the delayed-noise background in dark matter searches extends well beyond 5 ms.","section":"Abstract and Section III"}],"minor_comments":[{"comment":"The temperature is given as '(-99 +/- 1) C' with a calibration uncertainty of '1 C'; the degree symbol should be included consistently (e.g., '1 °C').","section":"Section II.A"},{"comment":"In the definition dp(t) = α a-bar t^k, the sign convention for k is not stated explicitly. Since k is negative, it would help to note that k < 0 for a decaying power law.","section":"Section I"},{"comment":"The sentence describing the S13370 sensor says 'the measured random photon background rate in the single S13371 sensor is an order of magnitude smaller than that of the S13370'; it would be clearer to state that this is the dark/random rate in the absence of an excitation pulse, since that is the relevant comparison for the delayed-photon measurement.","section":"Section II.C"}],"recommendation":"major_revision","confidential_remarks":"The paper is well within the scope of the journal and addresses a timely problem. The main technical weakness is the uncontrolled sensor comparison in Section II.C; a direct quartz-addition control would substantially strengthen the central claim. The authors should also be asked to align the abstract with the measured time range. I see no reason to doubt the authors' integrity, but the current evidence, while suggestive, is not yet sufficient for the broad claim as stated."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is the first paper that plausibly identifies the source of the ms-scale delayed photon noise that has been eating low-threshold LXe searches for years: UV-stimulated fluorescence of quartz photosensor windows. The experimental work is well-designed, and the key control—comparing a quartz-windowed sensor to a windowless one with more ceramic facing the xenon—actually cuts against the most obvious alternative. The stress-test worry about the sensor comparison is overstated: if ceramic were the dominant fluor, the windowless sensor with 4x more ceramic should have shown more delayed light, not 20x less. That is a genuinely informative comparison.\n\nWhat is new: the attribution itself, plus the vacuum Cherenkov test showing quartz fluoresces on the right timescale with a power law. The material-substitution tests in Sec. II.A are careful and rule out PTFE, electrodes, and the R8778-vs-S13371 difference. The paper also reports what looks like a delayed photocathode after-pulse in PMTs; minor but new.\n\nSoft spots, in order of real concern. First, the two sensors in the key comparison are different models, and no PDE/spectral-response normalization is shown. If the S13370's quantum efficiency at 175 nm is much lower than the S13371's, the 20x reduction could be trivial. The excitation pulse is also detected at 175 nm, so the ¯a normalization may account for this, but the authors do not demonstrate it explicitly, and the α values in Table I are not directly comparable across sensors. Second, the abstract says \"dominant component\" without the t<5 ms qualifier that the body uses; the 5–1000 ms window is the one that matters for dark matter searches, and the paper admits it is unprobed. Third, the residual k~1/4 component is attributed to ceramic without a direct measurement—plausible but a guess. Fit parameters lack uncertainties throughout, which is annoying for a paper built on power-law fits.\n\nThe citation pattern looks fine; the earlier PTFE hypothesis is properly referenced and the LUX/LZ numbers are engaged. No data or code are provided, which is a small disappointment for a small-scale lab measurement.\n\nWho this is for: anyone working on LXe TPCs or designing next-generation detectors (XLZD). I would bring it to a reading group. Recommendation: send it to peer review. The central claim is strong for t<5 ms and clearly worth refereeing; the authors are honest about the longer-time extrapolation.","headline":"First credible identification of quartz window fluorescence as the dominant ms-scale delayed photon background in LXe detectors; the case is strong for t<5 ms, but the abstract overreaches to longer times.","tokens_in":9237,"tokens_out":3883,"would_cite":true,"duration_ms":38919,"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 series of controlled tests in a liquid xenon cell traces the decade-old 'delayed photon noise' background to UV-stimulated fluorescence from quartz photosensor windows.","keywords":["liquid xenon detectors","delayed photon noise","quartz fluorescence","photosensor windows","power law decay","accidental coincidence background","dark matter search","scintillation afterpulsing"],"falsifier":"Take the same optically isolated liquid xenon cell, place a removable fused-quartz plate in front of the windowless sensor, and measure the delayed photon rate with the plate present and absent; if the rate does not rise by roughly the factor of 20 shown in Fig. 4 when the quartz is added, the central attribution fails. As a second check, search operating xenon detector data for the per-sensor correlation between average excitation pulse size and delayed photon rate using the event-selection conditions the paper describes; a clean null result would contradict the claim.","tokens_in":8183,"feed_emoji":"💡","tokens_out":13290,"duration_ms":120140,"temperature":0.7,"pith_summary":"Almost a decade after a power-law delayed photon noise was first observed following scintillation pulses in liquid xenon detectors, this paper reports a series of small-cell tests aimed at identifying its origin. The authors systematically replaced PTFE walls with Kapton and aluminum, removed the electrode, varied the liquid level, switched between two sensor types, and found the power law unchanged, narrowing the source to the photosensors themselves. Vacuum measurements using Cherenkov light as the excitation pulse confirmed that quartz fluoresces with a magnitude and time profile similar to the observed noise. A direct, optically isolated comparison between a quartz-windowed and a windowless silicon photomultiplier showed the delayed photon rate dropping by a factor of about 20 when the quartz was absent. The paper concludes that ultraviolet-stimulated fluorescence from quartz photosensor windows is the dominant component of the $t < 5$ ms delayed photon noise, a background that currently limits low-mass dark matter searches and coherent neutrino-nucleus scattering measurements.","feed_headline":"Quartz windows cause delayed photon noise in xenon detectors","feed_subtitle":"Tests in a liquid xenon cell trace the millisecond photon trickle to UV-stimulated quartz fluorescence.","key_machinery":"The central experimental device is a small cylindrical liquid xenon cell whose active volume is split by an aluminum divider into two optically isolated halves, one viewed by a quartz-windowed S13371 silicon photomultiplier and the other by a windowless S13370. A cascade trigger opens four successive 100 µs windows after each excitation pulse, letting the authors count individual delayed photons from about 10 µs out to several milliseconds. The decisive move is the paired comparison: with the quartz present, the delayed photon rate in the 50–100 µs bin is about a factor of 20 higher than without it, which isolates the quartz window as the dominant emitter. A separate vacuum measurement, using Cherenkov photons as the excitation pulse, confirms that quartz alone reproduces the measured magnitude and time profile.","core_discovery":"The paper claims that the delayed photon noise $d_p(t)=\\alpha \\bar a t^k$ seen after xenon scintillation pulses is dominated, on millisecond timescales, by ultraviolet-induced fluorescence of the quartz windows on the photosensors, rather than by the xenon, PTFE walls, electrodes, or impurities. The evidence is threefold: the power law survives every material substitution that leaves quartz in the light path; a similar power-law delayed photon signal is produced when PMT windows are excited by Cherenkov light in vacuum; and a windowless sensor in the same optically isolated liquid xenon cell measures a delayed photon rate about a factor of 20 lower than a quartz-windowed sensor. A residual component with a shallower exponent $k\\simeq 1/4$ remains in the windowless sensor, which the authors suspect is fluorescence of the sensor's ceramic package. This identification gives the field a concrete, removable cause for a background that had been hypothesized to come from PTFE and had gone unexplained for years.","pith_inferences":["If the mechanism is correct, the delayed photon noise amplitude in large detectors should scale with the total area of quartz exposed to scintillation light; a comparison of background rates across operating detectors with different photomultiplier models would test this scaling.","The same charge-trapping power-law fluorescence may apply to other optical window materials, so future low-background detectors should screen candidate materials under 175 nm excitation specifically on millisecond timescales rather than only at steady state.","The paper's own note that an anecdotal per-sensor correlation search reported no effect should be read as a test still to be done with the proper event selection; a clean null result would be a genuine falsifier, while a positive result would confirm the mechanism.","A simple follow-up experiment could place a removable fused-quartz plate in front of a windowless sensor and toggle it in and out; if the factor-of-20 difference does not track the presence of the plate, the central attribution would be in doubt."],"forward_implications":["Replacing fused-quartz windows with low-fluorescence materials, or removing the window entirely with silicon photomultipliers, should reduce the delayed photon noise and hence the accidental-coincidence background in future xenon detectors.","The amplitude and slope of the power law should track the amount of quartz and ceramic surface exposed to scintillation light, so different sensor layouts should show different delayed-noise parameters even for identical excitation pulses.","A per-sensor correlation between the average excitation pulse size and the delayed photon rate should be observable, provided events are selected carefully to avoid the wash-out the paper describes.","Because a residual ceramic-fluorescence component exists, windowless designs will still need to minimize exposed ceramic to fully suppress the background.","The paper's comparison of its millisecond-timescale data with later-time data from large detectors offers suggestive evidence that the same quartz mechanism continues to dominate out to hundreds of milliseconds, which is the regime relevant for dark matter searches."],"supporting_citations":[{"why":"Identifies the delayed photon noise and distinguishes it from delayed electron noise, defining the phenomenon this paper explains.","marker":"[12]"},{"why":"Detector background study providing the power-law delayed photon measurement with photomultiplier tubes and the comparison fit used in Table I.","marker":"[13]"},{"why":"Direct limits on PTFE photoluminescence, ruling out the previously leading candidate material and narrowing the search to the photosensors.","marker":"[14]"},{"why":"A doctoral thesis providing the power-law delayed photon fits and the continuous-data fitting method used for large detectors.","marker":"[15]"},{"why":"Earlier testbed work establishing that the measured power law is unbroken to at least 5 ms and is insensitive to xenon purity and applied electric fields.","marker":"[24]"},{"why":"Measurements of ceramic fluorescence that support attributing the residual component to the sensor's ceramic package.","marker":"[28]"},{"why":"Explanation of power-law luminescence by charge trapping on point defects, supporting the suggested mechanism for quartz fluorescence.","marker":"[30]"},{"why":"Early observation of fused quartz phosphorescence following ultraviolet exposure, consistent with the measured effect.","marker":"[32]"}],"fun_headline_variants":["Quartz fluorescence identified as xenon detector noise source","UV glow from quartz windows explains xenon detector background","Xenon detector's mystery photon noise traced to quartz","Quartz windows blamed for xenon detector's delayed photons","UV fluorescence of quartz windows culprit in xenon detectors"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The conclusion rests on the assumption that the quartz-windowed and windowless sensors differ in their delayed photon rates only because of the quartz window, with the aluminum divider giving complete optical isolation; if the windowless sensor's lower rate were instead due to its different package, leakage current, or dark rate, the attribution to quartz would weaken, and the residual $k\\simeq 1/4$ component is assigned to ceramic fluorescence without a direct control measurement.","fun_headline_variants_meta":{"raw":{"variants":["Quartz fluorescence identified as xenon detector noise source","UV glow from quartz windows explains xenon detector background","Xenon detector's mystery photon noise traced to quartz","Quartz windows blamed for xenon detector's delayed photons","UV fluorescence of quartz windows culprit in xenon detectors"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000565,"raw_usage":{"total_tokens":2628,"prompt_tokens":847,"completion_tokens":1781,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":463,"completion_tokens_details":{"reasoning_tokens":1703}},"tokens_in":463,"tokens_out":1781,"duration_ms":11962,"temperature":1.0,"reasoning_tokens":1703,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T22:04:30.238928+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the same optically isolated liquid xenon cell, place a removable fused-quartz plate in front of the windowless sensor, and measure the delayed photon rate with the plate present and absent; if the rate does not rise by roughly the factor of 20 shown in Fig. 4 when the quartz is added, the central attribution fails. As a second check, search operating xenon detector data for the per-sensor correlation between average excitation pulse size and delayed photon rate using the event-selection conditions the paper describes; a clean null result would contradict the claim.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Detector background study providing the power-law delayed photon measurement with photomultiplier tubes and the comparison fit used in Table I."},{"cited_title":"Gibbons and P","cited_arxiv_id":null,"evidence_quote":"Measurements of ceramic fluorescence that support attributing the residual component to the sensor's ceramic package."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Early observation of fused quartz phosphorescence following ultraviolet exposure, consistent with the measured effect."}],"review_version":1}