{"id":"908d75a9-44a1-46f2-87b4-7115b8d57aab","arxiv_id":"2505.24682","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A dual-phase xenon TPC with an infrared-sensitive PMT observes coincident IR emission with both scintillation (S1) and electroluminescence (S2) signals, with a distinct time structure and field-dependent yield.","lead":"Researchers operated a small dual-phase xenon detector that reads out both ultraviolet and infrared light, and found infrared flashes arriving together with the usual xenon scintillation and electroluminescence signals. The work is the first dual-phase TPC measurement spanning 170 nm to 1700 nm, and could eventually help separate signal from background in dark matter and double-beta-decay searches.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"IR identification is not yet secured: the PMT's sub-1 µm sensitivity and operation at −1 V bias (rather than −4.75 V) could make the dominant 30 µs tail and field-dependent rate non-IR in origin.","rationale":"The reader's weakest assumption identifies the same load-bearing concern: the IR PMT's broad spectral sensitivity and sub-design bias voltage leave unresolved whether the observed signals, particularly the dominant ~30 µs component and the field-dependent rate, are genuinely xenon IR emission or contamination from shorter wavelengths and PMT artifacts. This concern is central because the paper's new claims—the distinct time response and the field dependence of IR EL—depend on the signal origin. The authors themselves list these alternative explanations in Section 5 and do not resolve them. However, the paper is an honest first measurement with transparent limitations, and the qualitative observation of some IR light is credible. A conditional verdict with a clear path to verification is appropriate, so I do not recommend changing the reader's verdict.","tokens_in":11997,"tokens_out":5089,"duration_ms":62220,"concrete_test":"Repeat the S2-correlated measurement (rate vs. Egas and arrival-time spectrum at 12 kV/cm) with (a) the photocathode biased at the design −4.75 V and (b) a long-pass filter with cutoff ≥1 µm, or an InGaAs photodiode insensitive below 1 µm, in front of the sensor. If the ~30 µs tail and the monotonic rate increase persist with amplitudes consistent with the known IR QE, the central claim is supported; if they disappear or drop to the UV/visible contamination level, the IR interpretation fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that IR photons from xenon electroluminescence and primary scintillation are observed rests on the assumption that the R5509-73 PMT's output is dominated by xenon IR scintillation. Two documented weaknesses undermine this. First, the PMT is sensitive from 300 nm to 1650 nm (Fig. 2), so third-continuum, neutral-bremsstrahlung, and material-fluorescence photons can all contribute. The authors dismiss third-continuum as subdominant using literature sub-percent levels and nominal QEs, but the actual QE at the photocathode bias used (~−1 V instead of the design −4.75 V) is unknown; if the short-wavelength QE degrades less than the 1.3 µm QE, visible/NUV contamination could be comparable. Second, the dominant new feature—the ~30 µs component—overlaps the expected time scale for PMT afterpulsing or delayed photoelectron extraction from a mis-biased InP/InGaAs photocathode, and the paper explicitly cannot rule out a PMT-related effect. Because this component dominates the S2-correlated IR signal, the claimed time response and the field-dependent IR rate (Fig. 8) could be partly or wholly instrumental. The qualitative detection of some IR is plausible, but the quantitative support for 'IR emission from electroluminescence' and the suggested particle-ID application is not established until the wavelength and PMT-artifact ambiguities are resolved.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports measurements with a dual-phase xenon TPC (HeXe) instrumented with both a UV-sensitive and an IR-sensitive PMT, covering 170 nm to 1700 nm. Using internal alpha events, the authors observe IR-channel pulses coincident with S1UV and S2UV signals, study the time structure of the IR signals relative to each, and measure the dependence of the IR rate following S2UV on the gas amplification field. They report a fast (~30 ns) and a slow (~20 µs) component after S1UV, and for S2UV a three-component structure with a dominant ~30 µs decay. They also find the IR rate per alpha event increases with field, and they discuss possible origins including xenon IR scintillation, third continuum, neutral bremsstrahlung, delayed electrons, material fluorescence, and PMT artifacts. The paper explicitly acknowledges that the IR PMT was operated at a suboptimal photocathode bias and that PMT-related or wavelength-contamination effects cannot be ruled out.","tokens_in":12257,"tokens_out":2249,"duration_ms":28673,"significance":"If the observed signals are genuine xenon IR scintillation, the work would be the first dual-phase TPC measurement with UV-to-IR sensitivity and would open a new readout channel for particle identification and background rejection. The time-structure results, particularly the ~30 µs component, would be a new property of electroluminescence-induced IR emission. The paper is also candid about its limitations, which is commendable. However, because the central identification rests on a PMT operated outside its design bias and with sensitivity down to 300 nm, the quantitative conclusions are not yet secured; the significance of the claimed new physics therefore remains conditional on additional verification.","major_comments":[{"comment":"The IR PMT was operated at a photocathode bias of about -1 V instead of the design value of -4.75 V, so the quantum efficiency at the xenon IR wavelength is unknown. The paper acknowledges this but does not provide any calibration or cross-check. Since all quantitative statements about the IR signal, including the relative fractions of the decay components in Fig. 6 and the field-dependent rates in Fig. 8, scale directly with the quantum efficiency, the central quantitative claims are not robust until the actual QE is measured or the bias is corrected.","section":"Section 2.2"},{"comment":"The R5509-73 PMT is sensitive from 300 nm to 1650 nm, so the observed coincidence signals could include third-continuum, neutral-bremsstrahlung, or material-fluorescence photons rather than xenon IR scintillation. The paper argues that the third continuum is subdominant based on literature sub-percent levels in electroluminescence and on the nominal QE curve, but the QE at the actual operating bias is unknown; if the short-wavelength response degrades less than the 1.3 µm response, the contamination could be comparable to the IR signal. A wavelength-filter measurement or a direct spectral diagnostic is needed to support the IR interpretation.","section":"Figure 2 and Section 5"},{"comment":"The dominant ~30 µs component after S2UV overlaps the expected time scale for PMT afterpulsing or delayed photoelectron extraction from a mis-biased InP/InGaAs photocathode, and the authors state that they cannot rule out a PMT-related effect. This component accounts for more than half of the total IR signal integral, so the claimed time response and the field dependence of the IR rate could be partly or wholly instrumental. A control measurement with the correct bias voltage, a different PMT, or a pulsed light source would be required to establish that this component is intrinsic to xenon.","section":"Section 5 (Discussion) and Figure 6"},{"comment":"The field-dependence result is presented as supporting IR electroluminescence, but the rate is integrated only over a window from -1 to 34 µs relative to S2UV, and the paper notes that about 15% of IR pulses at the nominal field fall outside this window and that this fraction may vary with field. If the slow component is partly instrumental, as the previous comment argues, the observed monotonic rise in Fig. 8 could reflect a field-dependent PMT artifact or field-dependent contamination from the UV/EL light, rather than true IR EL emission. The extraction-efficiency correction does not address this possibility.","section":"Section 4.2, Figure 8"}],"minor_comments":[{"comment":"The sentence 'This was implemented by stepping down the supply voltage using a Zener diode in series with a potentiometer' could be clarified by stating whether the potentiometer adjusts the bias directly and how the -4.75 V value was calibrated.","section":"Section 2.2"},{"comment":"The liquid level is given as (4.1 ± 0.5) mm above the gate, while the nominal level in typical HeXe operation is 2.5 mm. The paper should state explicitly whether this difference affects the extraction field or the drift time, or refer to the detailed field simulation for this run.","section":"Section 2.1"},{"comment":"The description of the timing filter amplifier and how its shaping affects the IR pulse time and integrated area is missing; since the IR time response is a central result, the expected distortion from the shaping should be quantified or at least discussed.","section":"Section 3"},{"comment":"The inset shows a 'PMT dark count background' line, but it is not clear how the dark-count rate was measured at the operating temperature and bias; a brief explanation would help.","section":"Figure 5"},{"comment":"The reference list contains a private communication (Pollmann 2025) and a dataset entry for COMSOL; these would be more standard as footnotes or in a dedicated acknowledgments/software section.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The paper is interesting and the experimental effort is substantial, but the central claim of observing xenon IR emission is currently underdetermined by the PMT bias and wavelength-response ambiguities. The authors themselves list the unresolved alternative explanations. I would not reject the manuscript, because the qualitative coincidence and field-dependence are suggestive and the limitations are fixable with additional measurements (proper bias, wavelength filter, or a control PMT). However, in its present form the quantitative time-response and field-dependence claims cannot be considered established. I recommend major revision, asking for either new data that remove the PMT-bias ambiguity or a substantial weakening of the conclusions to purely phenomenological observations with explicit caveats in the abstract and title."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a genuine first — simultaneous UV and IR readout in a dual-phase xenon TPC — and the qualitative claim (IR photons coincident with both S1 and S2, increasing with electroluminescence field) is supported by the time-correlation data. It deserves referee time. But treat the quantitative time structure as provisional: the dominant ~30 µs component, and the field-dependent rate, are not yet cleanly separated from PMT bias artefacts and non-IR sources.\n\nWhat's new: the authors got an IR-sensitive PMT into the same TPC as the standard UV PMT and saw coincident IR pulses with both S1 and S2 for alpha events from 222Rn daughters. Their Fig. 4, the drift-time vs IR-S1 time scatter, is a clean way of showing those coincidences. The field dependence in Fig. 8 is a monotonic increase, consistent with electroluminescence, and they corrected for extraction efficiency. That is a real step beyond the prior gas-only and liquid-only measurements.\n\nWhat's solid: the paper is transparent about its own limitations. Section 5 explicitly lists third continuum, neutral bremsstrahlung, PTFE/quartz fluorescence, and a PMT-related effect as possible contributors to the slow component; it states plainly that the absolute light yield cannot be quantified because the PMT QE is unknown at the ~−1 V bias actually used. That is the right way to report a first measurement. The decay fits are descriptive, and the systematic uncertainties on the rate are estimated.\n\nWhere I'd push back: the stress-test note is right that the IR identification is not fully secured. The R5509-73 is sensitive from 300 nm to 1650 nm, the actual photocathode bias was −1 V instead of −4.75 V, and the resulting QE is unknown. The dominant ~30 µs tail overlaps the timescale expected for delayed photoelectron extraction or afterpulsing from a mis-biased InP/InGaAs photocathode, and the authors cannot rule that out. So the specific claim that the ~30 µs component is xenon IR emission, and the pulse-shape-discrimination suggestion built on it, are not established. That said, the qualitative detection of some IR from electroluminescence is plausible: you see a population of IR pulses lining up with S2, and the rate grows with field. The paper itself does not oversell — it says the data \"support\" the observation and lists the follow-up needed (correct electronics, a wavelength filter, higher purity).\n\nCitation pattern looks fine; prior gas-phase work is properly cited, and the self-citations are to their own earlier measurements where they are the relevant prior result.\n\nWho this is for: the noble-liquid TPC community, especially people thinking about alternative readout channels or background discrimination. A serious referee should engage with it; I would expect conditional acceptance with requested revisions on the interpretation.","headline":"First dual-phase TPC with simultaneous UV/IR readout: a genuine first measurement with honest limitations, but the quantitative IR time response is not yet nailed down.","tokens_in":12813,"tokens_out":1895,"would_cite":true,"duration_ms":22379,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["29.40.Mc"],"model":"deepseek-v4-flash","headline":"A dual-phase xenon detector can read out infrared scintillation from both liquid and gas, alongside ultraviolet light, giving rare-event searches a second wavelength channel.","keywords":["liquid xenon","scintillation","infrared radiation","dark matter","noble gas detectors","time projection chamber","electroluminescence","dual-phase detector"],"falsifier":"Insert a long-pass optical filter that transmits only wavelengths above about 1 µm in front of the infrared sensor and repeat the electroluminescence-coincidence measurement; if the ~30 microsecond component disappears or shrinks dramatically, it was not xenon's 1.3 µm scintillation but shorter-wavelength light or a sensor artefact. Alternatively, operating the sensor at its design bias and observing the same time profile would argue the tail is a real xenon emission feature.","tokens_in":11802,"feed_emoji":"🔬","tokens_out":10247,"duration_ms":106074,"temperature":0.7,"pith_summary":"This paper reports the first operation of a dual-phase xenon time projection chamber whose readout covers 170 nm to 1700 nm, and shows that infrared photons arrive in coincidence with both the prompt ultraviolet scintillation from the liquid and the electroluminescence from the gas. The infrared time response contains a fast component and a dominant slow decay of about 30 microseconds that has not previously been reported for xenon, and the infrared rate from electroluminescence rises as the gas amplification field is increased from about 9 to 15 kV/cm. If correct, the finding gives xenon-based detectors an additional wavelength channel that could improve particle identification and background rejection in dark matter and neutrinoless double-beta decay searches.","feed_headline":"Xenon detector catches infrared flashes from liquid and gas","feed_subtitle":"Infrared pulses track usual UV flashes and electroluminescence, adding a second readout channel for rare-event searches.","key_machinery":"The central object is a dual-phase xenon time projection chamber that pairs the standard ultraviolet readout with an infrared-sensitive photomultiplier covering roughly 300 nm to 1.65 µm, so the same event is viewed in two widely separated wavelength bands. Alpha decays from a radon calibration source tag events, and the drift-time relation between the UV S1 and S2 pulses lets each IR pulse be assigned to either liquid scintillation or gas electroluminescence. The load-bearing measurement is the distribution of IR arrival times relative to the UV S1 and S2 peaks, fit with sums of exponentials, together with the rate of IR photons as a function of the gas-field strength.","core_discovery":"In a small dual-phase xenon time projection chamber instrumented with one ultraviolet-sensitive and one infrared-sensitive photomultiplier, the authors observe IR photons in coincidence with both the prompt liquid scintillation (S1UV) and the gas-phase electroluminescence (S2UV). The IR pulses following the electroluminescence signal are described by a fast, poorly resolved component of about one-third of the signal, a ~1.7 microsecond component, and a dominant ~30 microsecond component that makes up roughly half of the total; a similar ~20 microsecond tail follows the S1UV signal. The number of IR photons associated with electroluminescence increases monotonically with the amplification field, and the ratio of IR to UV signal may not be constant across fields. The authors conclude that xenon emits infrared scintillation during electroluminescence and that this emission, with its distinct time structure, could serve as an additional readout channel.","pith_inferences":["The paper does not establish that the 30 microsecond component is xenon emission; testing it with a long-pass filter that blocks light below 1 µm would separate xenon's 1.3 µm continuum from shorter-wavelength contaminants and sensor effects.","If the liquid-phase IR yield is genuinely two orders of magnitude below the gas-phase yield, as earlier work suggests, the practical payoff is likely to come from the gas electroluminescence channel rather than from S1 readout; the paper's own data cannot quantify the liquid yield.","A high-speed digitizer and correctly biased infrared sensor could resolve the unresolved fast component and test whether it matches the ~26 ns UV S1 decay constant, which would strengthen the case that the IR fast component is scintillation rather than an artefact.","The apparent non-constant IR-to-UV ratio across fields hints that the two emissions are produced by different excitation pathways; a precision measurement of this ratio could probe the underlying electroluminescence mechanism."],"forward_implications":["The infrared channel can act as a second, wavelength-separated measure of electroluminescence, since its rate grows with the gas amplification field.","If the fast IR component fraction differs between alpha-induced and electron-induced signals, IR pulse shape could help identify particle types in the gas.","Comparing IR and UV signals could help reject accidental coincidences, such as distinguishing single-electron S2-like signals from true S1 signals.","The dominant ~30 microsecond IR tail, if genuine xenon emission, sets a timescale that future IR-readout designs must accommodate."],"supporting_citations":[{"why":"Measured IR light yield in xenon gas and its growth with electric field, the direct predecessor of the field-dependence result.","marker":"Belogurov et al. (2000)"},{"why":"Prior measurement of IR scintillation time response in gaseous xenon for alpha particles, providing the fast and slow component values used for comparison.","marker":"Hammann et al. (2024)"},{"why":"First time-resolved IR scintillation measurement in gaseous xenon, establishing the few-nanosecond fast component and microsecond-scale slow decay.","marker":"Piotter et al. (2023)"},{"why":"Measured the infrared emission spectrum of xenon excimers, locating the continuum near 1.3 µm and giving the spectral basis for the IR interpretation.","marker":"Borghesani et al. (2007a)"},{"why":"Detected IR scintillation in liquid xenon with yield two orders lower than gas, the liquid-phase prior for the S1-coincident IR signals.","marker":"Bressi et al. (2000)"},{"why":"Measured UV scintillation decay constants in liquid xenon, the reference for comparing the fast IR component's timescale.","marker":"Cichon et al. (2022)"},{"why":"Characterized the third-continuum UV emission with about 8 ns decay, used to estimate possible shorter-wavelength contamination in the IR sensor.","marker":"Millet et al. (1978)"},{"why":"Documented delayed-electron tails and single-electron signals in a xenon TPC, considered as an alternative source of the slow IR tail.","marker":"Aprile et al. (2014)"},{"why":"Provided the electron extraction efficiency model used to correct the field-dependent IR rates.","marker":"Xu et al. (2019)"},{"why":"Reported a slow IR component in argon, cited by the paper as evidence that slow noble-gas IR scintillation processes exist.","marker":"Buzulutskov et al. (2011)"}],"fun_headline_variants":["Xenon detector sees infrared echoes of ultraviolet flashes","Infrared light from xenon offers second readout channel","Dual-phase xenon emits infrared alongside ultraviolet","Xenon TPC detects infrared from scintillation and electroluminescence","UV and IR photons both detected in xenon detector"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The conclusions assume the infrared sensor's pulses are really xenon's infrared light, even though the sensor was run below its design voltage and also responds to shorter wavelengths; if the slow 30-microsecond tail or much of the coincident signal comes from the sensor itself, from surrounding materials, or from shorter-wavelength xenon emission, the time-response claims weaken.","fun_headline_variants_meta":{"raw":{"variants":["Xenon detector sees infrared echoes of ultraviolet flashes","Infrared light from xenon offers second readout channel","Dual-phase xenon emits infrared alongside ultraviolet","Xenon TPC detects infrared from scintillation and electroluminescence","UV and IR photons both detected in xenon detector"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000863,"raw_usage":{"total_tokens":3717,"prompt_tokens":893,"completion_tokens":2824,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":509,"completion_tokens_details":{"reasoning_tokens":2743}},"tokens_in":509,"tokens_out":2824,"duration_ms":23527,"temperature":1.0,"reasoning_tokens":2743,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T12:15:04.626842+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Insert a long-pass optical filter that transmits only wavelengths above about 1 µm in front of the infrared sensor and repeat the electroluminescence-coincidence measurement; if the ~30 microsecond component disappears or shrinks dramatically, it was not xenon's 1.3 µm scintillation but shorter-wavelength light or a sensor artefact. Alternatively, operating the sensor at its design bias and observing the same time profile would argue the tail is a real xenon emission feature.","supporting_citations":[],"review_version":1}