{"id":"f1389ab3-67a2-410e-82b0-2a2183cd62c0","arxiv_id":"2412.10830","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The R12699 PMT reaches ~0.08 mBq/PMT 60Co and ~0.06 mBq/PMT 238U late chain radioactivity, 15-fold lower than PandaX-4T's R11410, with cryogenic gain of 4.23e6 and 2.5 Hz dark count rate.","lead":"The PandaX collaboration and Hamamatsu have built a new four-channel photomultiplier tube, the R12699, with about 15 times lower radioactivity from key isotopes than the PMTs currently used in PandaX-4T. The tube also works well at liquid xenon temperatures and is a candidate sensor for next-generation dark matter and neutrino experiments.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claimed '>15-fold reduction vs PandaX-4T R11410' is derived by dividing measured R12699 activities by R11410-23 upper limits (Table 5); upper limits cannot establish a lower bound on the reduction factor, so the headline comparison is not supported as stated.","rationale":"The paper is a solid engineering and characterization report: the v2 PMT absolute activities in Table 4 (60Co 0.08(1) mBq/PMT, 238U(l) 0.06(2) mBq/PMT) and the cryogenic performance (gain 4.23±1.37e6, DCR 2.5±0.9 Hz) are directly measured and internally consistent. The reader's surface-210Po concern is real but modest, because that quantity is explicitly reported as an upper limit (<18.4 µBq/cm2) and the cleaning-transfer assumption is discussed openly. The more load-bearing weakness is the abstract's superlative quantitative claim: the '15-fold reduction compared to R11410 used in PandaX-4T' rests on comparing measured R12699 activities to the R11410-23 upper limits in Table 5. Statistically, an upper limit on the reference activity provides only a one-sided bound and cannot prove that the reference activity is at least 15 times larger. This is a logical flaw in the headline comparison, although it does not invalidate the absolute measurements or the demonstrated low-background performance. The fix is straightforward: use the measured R11410-23 values from Ref. [34], or explicitly downgrade the claim to 'at least 15 times below the R11410-23 screening upper limit.' Because the underlying data and the conditional verdict are otherwise sound, the reader's CONDITIONAL verdict should remain unchanged.","tokens_in":14933,"tokens_out":7992,"duration_ms":64271,"concrete_test":"Retrieve the R11410-23 activity central values from the PandaX-4T screening reference (Ref. [34], Qian et al., JHEP 06 (2022) 147) and recompute the 60Co, 232Th(l), and 238U(l) reduction factors using those central values. If only upper limits are available, explicitly relabel the claim in the abstract and Section 5 as 'at least X times below the R11410-23 screening upper limit' rather than a measured reduction. If the true R11410-23 activities are close to the limits, the current wording overstates the improvement and must be corrected.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The abstract's central quantitative claim is that the R12699 PMT achieves a '15-fold reduction compared to R11410 PMT used in PandaX-4T'. In Table 5, the R11410-23 (PandaX-4T) row lists 60Co <0.073, 232Th(l) <0.095, and 238U(l) <0.12 mBq/cm2, all as upper limits, while the R12699 v2 values are measured: 0.004(1), 0.004(1), and 0.003(1) mBq/cm2. Dividing a measured value by an upper limit yields only a one-sided bound, not a lower bound on the true ratio. If the actual R11410-23 activity were, for example, 0.02 mBq/cm2 for 60Co, the real reduction factor would be about 5, not >15. Section 5 repeats this comparison without acknowledging that the R11410-23 entries are limits. The absolute v2 activities (Table 4: 60Co 0.08(1) mBq/PMT, 238U(l) 0.06(2) mBq/PMT) remain valuable, but the headline factor should be reworded as 'at least 15 times below the R11410-23 screening upper limit' or supported by actual measured values from Ref. [34].","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports the development and characterization of the R12699-406-M4 photomultiplier tube, a four-channel 2-inch square PMT produced jointly by the PandaX collaboration and Hamamatsu. The authors describe the iterative replacement of two radio-contaminated components (Kovar and a sealing glass), and present HPGe screening results for the bulk materials, radon emanation measurements, and surface 210Po cleaning tests. They also describe a cryogenic test setup and give electrical results for a batch of 54 PMTs: an average gain of 4.23×10^6 at -1000 V and -100 °C, a dark count rate of 2.5 Hz per channel, and an after-pulse probability of 0.5%. The central claims are that the R12699 achieves a roughly 15-fold radioactivity reduction compared with the R11410 used in PandaX-4T, and that it combines low background with robust cryogenic performance.","tokens_in":15196,"tokens_out":4496,"duration_ms":39111,"significance":"If the quantitative claims are appropriately revised, this is a valuable engineering result for next-generation liquid xenon detectors. The paper provides a detailed material screening of individual PMT components, a clear demonstration that replacing Kovar and sealing glass reduces the dominant 60Co and 238U(l) contributions, and a statistically meaningful characterization of 216 channels from 54 PMTs. The reported absolute v2 activities for 60Co (~0.08 mBq/PMT) and the cryogenic gain and DCR results are useful inputs for background modelers. The work is reproducible in the sense that the screening methods, the charge-spectrum fit function, and the test procedures are described in enough detail to allow independent verification. The main weaknesses are in the interpretation of upper limits and in the surface-cleaning extrapolation, which currently overstate the low-background claim.","major_comments":[{"comment":"The headline claim of a '15-fold reduction compared to R11410 PMT used in PandaX-4T' is not supported by the data shown. In Table 5, the R11410-23 (PandaX-4T) entries for 60Co, 232Th(l), and 238U(l) are upper limits (<0.073, <0.095, <0.12 mBq/cm2), while the R12699 v2 entries are measured values (0.004(1), 0.004(1), 0.003(1) mBq/cm2). Dividing a measured value by an upper limit yields a one-sided bound on the ratio, not a lower bound on the actual reduction factor. The true R11410-23 activities could be substantially below the quoted limits, so the actual reduction factor is unknown; the wording should be changed to 'lower than the R11410-23 upper limit by a factor of at least 15' or supported by measured (not limit) values from Ref. [34].","section":"Abstract, Table 5, Section 5"},{"comment":"The v2 average for 238U(l) is listed as 0.06(2) mBq/PMT, but the individual v2 entries are 0.10(3), <0.11, <0.08, and <0.17 mBq/PMT. Averaging a single measured value with three upper limits as if they were measurements is statistically invalid; it biases the average downward and underestimates the uncertainty. The same issue affects the v2 232Th(l) average (0.09(3) from 0.17(5) and three upper limits). The authors should report per-batch limits separately, use a proper combined upper-limit analysis, or present a conservative upper limit instead of a claimed measured value. This affects the central low-background claim for 238U(l).","section":"Table 4, Section 2.1"},{"comment":"The abstract and Section 5 state that the surface 210Po activity is less than 18.4 µBq/cm2, but the cleaning verification in Table 7 shows that the initial quartz sample has <18.4 µBq/cm2, whereas after the first cleaning the activity is 27.8±8.1 µBq/cm2 and after the second cleaning it is <22.5 µBq/cm2. Thus the claimed upper limit corresponds to the uncleaned sample, not to the cleaned PMT surface. Moreover, the measurement is on a quartz sample, not on an assembled PMT, so the transfer of the cleaning procedure to the actual PMT window is an assumption. The surface-radioactivity claim should be corrected to reflect the post-cleaning values, or the limitation should be explicitly acknowledged.","section":"Section 2.3, Table 7, Abstract"}],"minor_comments":[{"comment":"The radon emanation upper limits in Table 6 are not assigned a confidence level in the text, although the abstract quotes a 90% confidence level. Please state the confidence level used for all limits in Table 6 and briefly describe how the limits were computed.","section":"Section 2.2, Table 6"},{"comment":"In the charge-spectrum fit function, the notation 'iσ1^2' in the denominator of the double-PE Gaussian is ambiguous: it could mean the index i multiplied by σ1^2 or an indexing error. Please clarify the expression, e.g., write (i σ1^2 + σ0^2) explicitly with a definition of i.","section":"Section 4.1, fit function"},{"comment":"The two rightmost columns of Table 9 are labeled 'Avg gain (optimized distribution ratio)' and 'Avg gain (original distribution ratio)', but the text says the optimization improved the gain; the numbers in the table appear to be consistent with that, yet the column order could be confused. Please verify that the labels match the data.","section":"Table 9"},{"comment":"The y-axis label 'Counts/(5 keV s)' in the right panel of Figure 3 appears to lack a clear normalization; if the spectra are normalized per second or per 5 keV bin, the axis title should state this explicitly.","section":"Figure 3"},{"comment":"When comparing DCR per unit area in Table 10, the R12699 has four independent channels within a 56 mm square footprint; it may be helpful to state how the per-channel and per-area normalization is defined so that the comparison with the R11410 is unambiguous.","section":"Section 5, Table 10"},{"comment":"Reference [30] is an untitled URL; please provide a full bibliographic entry, including a collaboration name and year, or cite a published equivalent.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The paper contains a genuine and useful development, but the quantitative claims need careful revision before publication. The most serious issue is the use of upper limits to claim a 15-fold reduction; this is not a matter of taste but of statistical validity, and it can be fixed by rewording or by obtaining measured values. The averaging of upper limits in Table 4 is a related concern that also affects the absolute 238U(l) activity quoted in the abstract. The surface 210Po claim is overstated relative to the data in Table 7. None of these issues appear to require new experimental data beyond what is already in the manuscript; they can be addressed by reanalysis and rewording. I recommend major revision, not rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The R12699 is a real step forward for radiopure PMTs, and the absolute activity numbers are the story, not the headline comparison. The v2 R12699 numbers (60Co 0.08(1) mBq/PMT, 238U(l) 0.06(2) mBq/PMT, 232Th(l) 0.09(3) mBq/PMT) are credible: HPGe at CJPL with a Geant4 efficiency model, a per-material budget, and a clear iterative path from v0 to v1 to v2. The cryogenic qualification of 54 tubes (216 channels) is also solid—gain, DCR, and afterpulse numbers at -100 °C look well measured, with a sensible fitting function for gain.\n\nThe soft spots are real but not fatal. The abstract's '15-fold reduction compared to R11410 used in PandaX-4T' is not supported as stated. In Table 5, the R11410-23 row lists 60Co, 232Th(l), and 238U(l) as upper limits; dividing the measured R12699 values by those limits gives a one-sided statement, not a lower bound on the true reduction. Reword to 'at least 15 times below the R11410-23 screening upper limit,' or quote measured values from Ref. [34]. The same caveat applies to Section 5.\n\nTwo smaller issues. Table 4 averages batches that mix measured values with upper limits; explain the averaging rule or quote the spread. And the surface 210Po number is measured on a quartz witness sample, not the assembled PMT; that's an understandable proxy but should be presented as an upper limit on the cleaning process, not on the PMT surface itself. The elevated 40K is openly acknowledged and the future plan (potassium chromate photocathode) is sensible.\n\nNone of these weaknesses touch the central claim that a compact four-channel 2-inch PMT can be made with sub-0.1 mBq/PMT activities on the critical chains while holding good cryogenic performance. That is the real contribution, and it's useful for next-generation LXe designs.\n\nFor a serious referee: yes, send it. The experimental work is thorough and the paper is honest about 40K and about the radon and 210Po limits. The required changes are wording and presentation, not new data.","headline":"A genuinely useful radiopure PMT with credible absolute activity numbers, but the headline '15-fold reduction' overstates the data because it divides measured values by upper limits.","tokens_in":15864,"tokens_out":1961,"would_cite":true,"duration_ms":17368,"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":"The R12699 photomultiplier tube cuts per-tube radioactivity by more than 15-fold through two material substitutions, while keeping full gain at liquid xenon temperature.","keywords":["photomultiplier tube","liquid xenon detector","low radioactivity","radiopurity screening","cryogenic PMT performance","dark matter search","radon emanation","surface 210Po alpha activity"],"falsifier":"Place an assembled R12699 PMT window in an $\\alpha$ spectrometer with sensitivity below $1~\\mu$Bq/cm$^2$ and count $^{210}$Po decays over several weeks; if the measured surface activity exceeds $18.4~\\mu$Bq/cm$^2$, the paper's surface-radioactivity claim is contradicted for the actual device.","tokens_in":14705,"feed_emoji":"⚛️","tokens_out":15613,"duration_ms":119841,"temperature":0.7,"pith_summary":"To make next-generation multi-ten-tonne liquid xenon detectors sensitive to dark matter and neutrinoless double $\\beta$ decay, the photomultiplier tubes that read the xenon scintillation light must themselves contribute almost no radioactivity. This paper claims the newly developed R12699 PMT achieves that by replacing two radioactive components: the Kovar metal parts (an iron-cobalt-nickel alloy) with an iron-nickel alloy, and the high-radioactivity sealing glass with a specially selected low-background glass melted in a platinum crucible. High-purity germanium measurements give about $0.08$ mBq/PMT of $^{60}$Co and $0.06$ mBq/PMT of $^{238}$U late-chain activity, more than 15 times lower than the R11410 PMTs used in PandaX-4T, with radon emanation below $3.2~\\mu$Bq/PMT. The same tube keeps an average gain of $4.23\\times10^{6}$ and a dark count rate of $2.5$ Hz per channel at $-100\\,^{\\circ}\\mathrm{C}$, so the low background does not come at the cost of cryogenic electrical performance.","feed_headline":"New PMT cuts detector background 15-fold for xenon searches","feed_subtitle":"Material swaps keep 60Co and 238U near silicon photomultiplier levels while gain stays high at –100 °C.","key_machinery":"The load-bearing mechanism is the replacement of two specific materials inside the PMT: Kovar, an iron-cobalt-nickel alloy whose neutron-activated $^{60}$Co dominated the gamma background, is swapped for an iron-nickel alloy, and the high-radioactivity sealing glass (Glass-1) is swapped for a low-background glass made from selected raw materials and melted in a platinum crucible to avoid recontamination. The quantitative engine is the high-purity germanium screening campaign at a deep underground laboratory, with detection efficiencies computed by the Monte Carlo simulation program the paper uses, which traces each decay chain ($^{60}$Co, $^{238}$U early/late, $^{232}$Th early/late, $^{40}$K, $^{137}$Cs, $^{235}$U) to specific PMT parts. A second element is the optimized voltage divider ratio, which increases the dynode-stage gain by about 30% while lowering the cathode-to-first-dynode voltage only slightly, with less than 5% loss in electron collection efficiency.","core_discovery":"The central claim is that the R12699-406-M4 PMT, a 2-inch square quartz-windowed tube with four independent detection channels in one body, reaches radioactivity levels low enough for next-generation liquid xenon detectors without sacrificing cryogenic operation. The active steps are material substitutions: replacing the Kovar plate, pin, and pipe with a new iron-nickel alloy brings $^{60}$Co from roughly $0.66$ mBq/PMT in the original v0 tube down to about $0.08$ mBq/PMT in the final v2 version, and replacing the sealing glass (Glass-1) brings $^{238}$U late-chain activity down to $0.06$ mBq/PMT and $^{232}$Th late-chain activity to $0.09$ mBq/PMT. The paper also reports an optimized voltage divider that raises the gain by about 30% compared with the manufacturer's recommended ratio. On a batch of 54 PMTs (216 channels), the average gain is $(4.23\\pm1.37)\\times10^{6}$ at $-1000$ V and $-100\\,^{\\circ}\\mathrm{C}$, the average dark count rate is $(2.5\\pm0.9)$ Hz per channel, and the average after-pulse probability is $0.5\\%$, with $^{40}$K identified as the remaining radioactivity to be addressed in future work.","pith_inferences":["A direct alpha assay of fully assembled R12699 windows, rather than quartz witness samples, would settle whether the cleaning procedure's $^{210}$Po removal transfers to production tubes.","The optimized divider lowers the cathode-to-first-dynode voltage from 160 V to 120 V; a dedicated VUV quantum-efficiency scan at the 178 nm xenon line would confirm that single-photon response is not degraded beyond the reported $<5\\%$ collection-efficiency change.","Because four channels share one tube envelope, a single PMT failure would remove four channels at once, so next-generation detector designs would need correlated-failure modeling rather than treating channels as independent.","The remaining $^{40}$K in the photocathode is the obvious next lever; if a $^{39}$K-enriched photocathode works at 178 nm, R12699's radioactivity would approach the silicon-photomultiplier level without changing the rest of the tube."],"forward_implications":["If the R12699's radioactivity holds in mass production, next-generation multi-ten-tonne liquid xenon detectors can install denser photocathode coverage with a lower PMT-induced background than current R11410-based detectors.","The 15-fold reduction in $^{60}$Co and $^{238}$U(l) per tube directly shrinks the gamma background in the dark matter and neutrinoless-double-beta regions of interest.","The radon emanation rate below $3.2~\\mu$Bq/PMT and surface $^{210}$Po below $18.4~\\mu$Bq/cm$^2$ remove two background paths that can mimic low-energy nuclear recoils.","At $-100\\,^{\\circ}\\mathrm{C}$ the average gain of $4.23\\times10^{6}$, dark count rate of $2.5$ Hz per channel, and after-pulse probability of $0.5\\%$ are comparable to or better than the R11410 values quoted for existing detectors, so the low-background design does not require a new readout threshold.","The four-channel-per-tube format and 30% gain improvement from the optimized divider mean the same number of readout channels can be packed in a smaller radioactive footprint."],"supporting_citations":[{"why":"Provides the PandaX-4T R11410-23 radioactivity values used as the comparison baseline and the electrostatic-collection radon emanation system used to measure R12699 emanation rates.","marker":"[34]"},{"why":"Supplies the LZ R11410-10 radiopurity numbers against which the R12699 improvement is benchmarked.","marker":"[32]"},{"why":"Supplies the XENONnT R11410-20 radiopurity numbers used in the PMT radioactivity comparison table.","marker":"[33]"},{"why":"Provides the Monte Carlo simulation program used to compute high-purity germanium detection efficiencies for the complex PMT geometry.","marker":"[28]"},{"why":"Documents the R13111 low-background PMT and the $^{39}$K-enriched photocathode approach cited as the future route to reduce $^{40}$K.","marker":"[35]"},{"why":"Supplies the manufacturer's PMT gain and ion-feedback model that underlies the voltage-divider optimization and the after-pulse interpretation.","marker":"[36]"},{"why":"Provides the single-photoelectron charge-spectrum fit function used to extract gain and single-PE resolution.","marker":"[39]"},{"why":"Supplies the XENON1T R11410-21 electrical performance values used to compare gain, after-pulse probability, and dark count rate.","marker":"[42]"},{"why":"Supplies the PandaX-4T R11410-23 electrical performance values used in the comparison table.","marker":"[41]"}],"fun_headline_variants":["PMT with 15x lower radioactivity improves xenon detector sensitivity","Novel PMT enables ultra-low-background cryogenic operation for xenon","15-fold radioactivity cut in new PMT for xenon dark matter detectors","Low-background PMT design for next-gen liquid xenon searches"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The surface $^{210}$Po claim is measured on a quartz witness sample, not on a fully assembled PMT, so the surface-radioactivity argument assumes the cleaning procedure removes $^{210}$Po from the real tube window just as it did from the sample.","fun_headline_variants_meta":{"raw":{"variants":["PMT with 15x lower radioactivity improves xenon detector sensitivity","Novel PMT enables ultra-low-background cryogenic operation for xenon","15-fold radioactivity cut in new PMT for xenon dark matter detectors","Low-background PMT design for next-gen liquid xenon searches"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000285,"raw_usage":{"total_tokens":1786,"prompt_tokens":1159,"completion_tokens":627,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":775,"completion_tokens_details":{"reasoning_tokens":549}},"tokens_in":775,"tokens_out":627,"duration_ms":5618,"temperature":1.0,"reasoning_tokens":549,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T15:34:26.443136+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Place an assembled R12699 PMT window in an $\\alpha$ spectrometer with sensitivity below $1~\\mu$Bq/cm$^2$ and count $^{210}$Po decays over several weeks; if the measured surface activity exceeds $18.4~\\mu$Bq/cm$^2$, the paper's surface-radioactivity claim is contradicted for the actual device.","supporting_citations":[{"cited_title":"Qian, et al., Low radioactive material screening and background control for the PandaX-4T experiment, JHEP 06 (2022) 147","cited_arxiv_id":null,"evidence_quote":"Provides the PandaX-4T R11410-23 radioactivity values used as the comparison baseline and the electrostatic-collection radon emanation system used to measure R12699 emanation rates."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the LZ R11410-10 radiopurity numbers against which the R12699 improvement is benchmarked."},{"cited_title":"Aprile, et al., Material radiopurity control in the XENONnT experiment, Eur","cited_arxiv_id":null,"evidence_quote":"Supplies the XENONnT R11410-20 radiopurity numbers used in the PMT radioactivity comparison table."},{"cited_title":"Chen, et al., BambooMC — A Geant4-based simulation program for the PandaX experiments, JINST 16 (09) (2021) T09004","cited_arxiv_id":null,"evidence_quote":"Provides the Monte Carlo simulation program used to compute high-purity germanium detection efficiencies for the complex PMT geometry."},{"cited_title":"Abe, Development of low background PMT R13111, J","cited_arxiv_id":null,"evidence_quote":"Documents the R13111 low-background PMT and the $^{39}$K-enriched photocathode approach cited as the future route to reduce $^{40}$K."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the manufacturer's PMT gain and ion-feedback model that underlies the voltage-divider optimization and the after-pulse interpretation."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the single-photoelectron charge-spectrum fit function used to extract gain and single-PE resolution."},{"cited_title":"Barrow, et al., Qualification tests of the R11410-21 photomultiplier tubes for the XENON1T detector, JINST 12 (01) (2017) P01024","cited_arxiv_id":null,"evidence_quote":"Supplies the XENON1T R11410-21 electrical performance values used to compare gain, after-pulse probability, and dark count rate."},{"cited_title":"Huang, Development of the photosensor system for PandaX-4T and the search for light-mediator-dark-matter, Ph.D","cited_arxiv_id":null,"evidence_quote":"Supplies the PandaX-4T R11410-23 electrical performance values used in the comparison table."}],"review_version":1}