{"id":"2c2730f1-f16a-46b4-84be-66a36dbdc540","arxiv_id":"2508.19117","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":7,"one_line_summary":"Flow-dependent tagging, built from paired radon-polonium decays, identifies 63% of 214Pb beta-decay backgrounds in LZ's fiducial volume at 9.0% exposure cost.","lead":"The LZ dark matter experiment shows it can track individual radioactive lead atoms drifting through its liquid xenon detector, and use that tracking to recognize and remove 63% of a leading background to dark matter signals, at the cost of 9% of its data. A smart generalist might read this because it turns an ordinary detector plumbing problem, liquid flow, into a new way to clean up dark matter data and to calibrate the detector in real time.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Charged-branch transport model extrapolated from 15-min 218Po+ data to 81-min 214Pb+/214Bi+ searches without direct validation; this gap is the likely source of the 75%-vs-63% efficiency shortfall and leaves the tag's mechanism partially unsupported.","rationale":"The spectral measurement of ϵtot is a real strength and is independent of the transport model, so the headline number is credible. However, the claim's causal narrative — that the tag targets each 214Pb atom out to 81 min — rests on a charged-ion drift model fit to 218Po+ data with ΔT ≤15 min and extrapolated to 214Pb+/214Bi+ out to 64–81 min. The paper itself (Sec. V.C) identifies this extrapolation as the likely cause of the 75% vs 63% gap. The quoted systematic does not include this model uncertainty. Thus the efficiency is measured, but its decomposition into neutral/charged/semi-charged contributions and its portability to other flow states or drift fields are not. The reader's CONDITIONAL verdict is appropriate; no change to the verdict is needed.","tokens_in":27737,"tokens_out":14487,"duration_ms":158365,"concrete_test":"Extend the charged-branch validation to late ΔT using the 218Po–214BiPo sample: for pairs with ΔT>15 min, compare the observed vertical displacement relative to charged streamlines (Eq. 5–7) with the Eq. (4) prediction. If the mean residual exceeds the window width (i+ or j+) or shows a ΔT-dependent trend, the extrapolation is unsupported. An alternative check: re-derive the tag efficiency using only neutral streamlines (charge-blind tag) and compare ϵtot; if it drops by more than the 75%-vs-63% gap, the charged-branch contribution is the limiting factor.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The headline efficiency ϵtot = 63% ±6% ±7% is a direct spectral measurement (Sec. V.B.1) and is not invalidated by transport-model imperfections. However, the paper's central mechanism — that tag volumes 'target the decay of each 214Pb atom up to 81 minutes after production' — depends on the Eq. (4) charged-ion drift model being applicable to 214Pb+ and 214Bi+ on timescales 4–5× longer than the 218Po+ data used to fit it. Section V.C explicitly attributes the gap between modeled (75%) and observed (63%) efficiency to this extrapolation, and notes the charged ΔZs window is tuned on 222Rn–218Po pairs with ΔT ≤15 min. Because the same transport model is also used to define the neutral and charged streamlines for the 218Po–214BiPo-validated windows, any bias in the charged branch changes the composition of the tagged sample (e.g., preferentially missing semi-charged paths) and thus affects the interpretation of ϵtot and its portability to other flow states or drift fields. The quoted ±7% systematic covers only the 214Pb-content prior, not the model-extrapolation uncertainty. This is the load-bearing weak point.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a method for tagging 214Pb beta decays in the LZ liquid-xenon TPC by exploiting the low 222Rn activity and stable, slow liquid flow in the detector. Sequential alpha decays (222Rn–218Po) are paired to construct a flow map and a charged-ion drift model; the resulting transport model is used to define time-evolving volume selections around predicted 214Pb trajectories. Applying these volumes costs 9.0% of exposure and removes 63% ± 6%(stat) ± 7%(sys) of the 214Pb ground-state decays in the 5.5 t fiducial volume, as measured directly by re-fitting the 18–75 keVee spectrum after removing tagged events. The tag has already been used in LZ's 2024 WIMP search, and the paper also demonstrates a concurrent low-energy ER calibration sample from tagged 214Pb decays. The central uncertainty in the method is the extrapolation of the charged-ion drift model, fitted to 218Po+ data out to ~15 min, to 214Pb+ and 214Bi+ trajectories out to 81 min; the authors explicitly identify this as the likely cause of the gap between the modeled (75%) and observed (63%) tagging efficiency.","tokens_in":27982,"tokens_out":3718,"duration_ms":38162,"significance":"If the result holds, this is an important advance for rare-event searches in liquid xenon: 214Pb is the leading low-energy ER background in LZ, and a demonstrated 63% tagging efficiency at 9% exposure cost is a significant background-mitigation tool. The paper's strongest point is that the headline efficiency is a direct spectral measurement (Section V.B.1), not a prediction from the transport model, so the central number does not collapse even if some model details are imperfect. The paper is also valuable for its detailed characterization of flow states and ion-neutralization times in a large LXe detector, and for introducing a concurrent calibration side band that has already been used in a published WIMP search. However, the transport-model extrapolation is a genuine limitation that affects the interpretation, portability, and the completeness of the quoted systematic uncertainty.","major_comments":[{"comment":"The charged-branch tag windows are built from an ion drift model, Eq. (4), fitted to 222Rn–218Po pairs with ΔT ≤ 15 min, then applied to 214Pb+/214Bi+ trajectories out to 64–81 min. The authors state in Section V.C that this is the likely source of the 75% vs 63% gap between modeled and observed tagging efficiency. Because the tag volumes themselves are defined by this model, systematic errors in the charged branch affect the composition of the tagged sample and the portability of the quoted efficiency to other flow states or drift fields. Yet the ±7% systematic on ϵtot in Section V.B.1 includes only the 214Pb-content prior, not the transport-model extrapolation. I would like a quantitative robustness test: vary VI, VL, and λ in Eq. (4) within the ranges allowed by the 218Po+ data and the 218Po–214BiPo pair distributions, and propagate the resulting changes to ϵtot. Absent this, the clai","section":"Section V.C; Eq. (4), Eq. (7)"},{"comment":"The charge branching fractions and neutralization time τn used to interpret the tag efficiency are measured from the same neutral search windows that define the tag. This self-calibration loop, combined with the assumptions that progeny charge state is independent of progenitor charge and that all charged progeny neutralize with the same mean τn, means that the modeled efficiency of 75% is not an independent validation of the transport model. In particular, the 218Po–214Pb charge branching fraction of 0.48 ± 0.12 is derived with the same assumed τn and is then replaced by the 222Rn–218Po value in the model, so the quoted ±0.12 is not propagated. The paper is transparent about these assumptions, but the systematic impact of this circularity should be assessed—for example, by re-deriving τn and the branching fractions with an independent dataset (e.g., 218Po–214BiPo pairs at late ΔT) or by","section":"Appendix A; Table III"},{"comment":"The spectral fit measures ϵtot and fexp simultaneously, but assumes all non-214Pb components are reduced by fexp uniformly. The paper notes that this is an approximation because background populations are not spatially uniform. The proxy-based fexp of 9.3% ± 0.6% is consistent with the direct integration, which is reassuring, but the systematic uncertainty on ϵtot does not include any contribution from this approximation. A short quantitative statement of how large the non-uniformity effect could be (e.g., by comparing the spectral-fit ϵtot with the shifted-search ϵpair in the same energy window) would strengthen the claim.","section":"Section V.B.1; Figure 12"}],"minor_comments":[{"comment":"The caption is dense and the phrase 'complimentary ϵ' appears to be a typo for 'complementary.' Also, the columns for ground-state vs excited-state measurements could be clearer about which uncertainties are statistical and which are systematic.","section":"Table II"},{"comment":"The one-sided relation ϵtot ≳ ϵpair + (1−ϵpair)fexp is useful, but the direction of the approximation could be stated more explicitly in the text, since the subsequent estimates use '≲' and '≳' interchangeably.","section":"Eq. (8)"},{"comment":"Reference [38] contains a typo: 'Elesvier' should be 'Elsevier.'","section":"References"},{"comment":"The parameters in Eqs. (1)–(3) are described as being set 'by hand'; a short explanation of the sensitivity of the final tagging efficiency to these choices would be helpful, even if the final tag windows are tuned separately.","section":"Section IV.B"}],"recommendation":"major_revision","confidential_remarks":"The core measurement, ϵtot = 63%, is direct and not invalidated by the transport-model shortfalls. The paper is honest about the gap between modeled and observed efficiency. However, the quoted systematic uncertainty excludes the dominant model-validity limitation, and the self-calibration of the charged-branch parameters deserves a robustness analysis. I believe the authors can address this with additional tests within the scope of the existing data, so I do not recommend rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear [Colleague],\n\nThe short version: this paper delivers a real, first-of-its-kind capability — flow-based tagging of 214Pb decays in a xenon TPC — and the headline number, 63% ± 6% ± 7% efficiency at 9% exposure cost, is measured directly by a spectral fit, not derived from the transport model. That measurement holds up. The paper is also unusually candid about its own weak spots, which is rare and refreshing.\n\nWhat's new: using 222Rn–218Po sequential alpha decays to map the liquid xenon flow and ion drift, then projecting single-atom trajectories forward up to 81 minutes to define tag volumes around each 218Po decay. The 214Pb tag was already deployed in LZ's 2024 WIMP search (PRL 135, 011802). They also measure, for the first time, the 218Po–214Pb charge branching fraction (0.48 ± 0.12) and a 214Pb–214Bi charge branching fraction (0.74 ± 0.05). The calibration side band from tagged events is a nice bonus.\n\nThe central measurement — ϵtot — comes from re-fitting the 18–75 keVee spectrum after removing tagged events. That is model-independent. The systematic uncertainty (±7%) is dominated by the 214Pb branching ratio prior, which is honest. The tag windows themselves are defined by hand on pair-excess images, but that's a tuning choice, not a flaw.\n\nThe soft spots: the charged-ion transport model (Eq. 4) is fit to 218Po+ drift data out to only ~15 minutes, then applied to 214Pb+ and 214Bi+ out to 81 minutes. The authors say the gap between modeled (75%) and observed (63%) efficiency is likely from this extrapolation. That's the main weakness — it means the mechanism is not fully validated at long times, and the ±7% systematic does not cover this model-extrapolation uncertainty. Also, the charge branching fractions and neutralization time are measured on the same pair searches whose efficiency the model then predicts — a disclosed circularity that does not touch the measured ϵtot, but does limit the model's independent predictive power. Minor: the whole thing is validated in one flow state at one drift field.\n\nVerdict: this deserves a serious referee. The central result is solid, the paper is honest about its limitations, and the technique is likely to propagate to XENONnT, PandaX, and XLZD. My recommendation: engage with it, get it reviewed, and don't let the model-data gap stop it — just make sure the limits of the transport model are stated as clearly in the final version as they are here.","headline":"Directly measured 63% 214Pb tagging efficiency holds up; the charged-ion transport model extrapolation to 81 minutes is the main soft spot, but the paper is honest and deserves serious review.","tokens_in":29799,"tokens_out":2657,"would_cite":true,"duration_ms":24005,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"LZ tracks single radon-chain atoms through its liquid xenon to tag 63% of 214Pb decays — the detector's leading low-energy background — for a 9% exposure cost.","keywords":["liquid xenon TPC","214Pb tagging","radon chain decay pairs","flow mapping","ion drift model","dark matter direct detection","background rejection","LZ experiment"],"falsifier":"Measure the drift of 214Pb+ and 214Bi+ directly: the semi-charged 218Po–214BiPo pairs (where one of the two intervening ions is neutral) form a population whose vertical displacement at late times (ΔT > 40 min) depends on the true drift parameters. If their late-time vertical deviations fall where the 218Po+-based extrapolation predicts, the charged-branch assumption holds; if they systematically deviate — or if the 75% minus 63% efficiency gap persists once species-specific drift is incorporated — the transport model, not the neutralization or branching assumptions, is the limiting piece.","tokens_in":27448,"feed_emoji":"🎯","tokens_out":9258,"duration_ms":73217,"temperature":0.7,"pith_summary":"The paper claims that the slow, stable circulation of liquid xenon inside the LZ dark matter detector can be mapped precisely enough to predict where individual radioactive atoms will be minutes later. Pairs of sequential alpha decays (222Rn then 218Po) act as tracers: each pair records how far a daughter atom drifted in the liquid between its birth and its own decay. From roughly one tracer pair per 27 cubic centimeters, LZ builds a flow map plus an ion-drift model, then projects each 214Pb atom's path forward for up to 81 minutes. Temporally evolving selection volumes around those projected paths catch 63% ± 6% (stat) ± 7% (sys) of 214Pb beta decays — the leading low-energy electron-recoil background in the WIMP search — while discarding only 9.0% of exposure. The tag was already used in LZ's 2024 WIMP search, and the tagged events also form a calibration sample collected simultaneously with science data.","feed_headline":"LZ tracks single atoms to erase 63% of its top background","feed_subtitle":"Radon-pair flow maps let LZ predict where each 214Pb atom will decay and veto it — already running in its 2024 WIMP search.","key_machinery":"The load-bearing objects are the flow map, the ion-drift model, and the evolving selection volume. The flow map is a vector field interpolated from neutral 222Rn–218Po pair displacements using a 10-cm Gaussian kernel over the 11 nearest sample vectors, integrated forward with an adaptive Runge-Kutta streamline integrator; the ion-drift model, Eq. (4), gives the instantaneous drift velocity V+Z = VL + (VI − VL)(1 − ΔT/λ)exp(−ΔT/λ), with VI = −0.55 mm/s, VL = −0.38 mm/s, λ = 115 s, fitted to the charged 222Rn–218Po band. The tag volume itself is a cylinder around each projected streamline whose radii grow with time as ΔXYs < g + h(1 − exp(−ΔT/κ)) and |ΔZs| < i + j(1 − exp(−ΔT/λ)), with paramet","core_discovery":"The central claim is that 214Pb beta decays can be individually anticipated and rejected, not just statistically subtracted. The demonstration rests on a transport model: 222Rn–218Po decay pairs provide displacement vectors for both neutral atoms (bulk flow) and singly charged ions (flow plus electric-field drift), and these vectors are interpolated into a three-dimensional flow field with streamlines projected forward in time. A charged-ion drift model, fitted to the 218Po+ population with initial drift velocity −0.55 mm/s relaxing to −0.38 mm/s on a 115-s time constant, extends the trajectories of charged progeny; selection windows around both neutral and charged streamlines, tuned on pure","pith_inferences":["I read the method as transferable: any liquid xenon (or other liquid noble) detector with slow, stable circulation and radon low enough that decay pairs are unambiguous could build the same pair-based flow map; the High Mixing state described here, where flow speeds approach the 0.4 mm/s ion drift speed and the neutral/charged pair populations blur, is the natural failure boundary.","A direct measurement of 214Pb+ and 214Bi+ drift velocities — for instance from the semi-charged 218Po–214BiPo population that falls between the neutral and charged bands — would test whether the 75-to-63% efficiency gap is an ion-drift model error; if it is, species-specific drift parameters should push the tag toward the modeled 75%.","The concurrent-calibration property may end up mattering beyond WIMP searches: every tagged 214Pb decay is a known low-energy beta in science data, so any electron-recoil-band analysis could use the tag as a live detector-response check without interrupting data taking for source injections."],"forward_implications":["The leading low-energy electron-recoil background in LZ's WIMP search is reduced by 63% at 9.0% exposure cost; the voxelized version of the tag was already deployed in the 2024 WIMP search with ϵtot as a floating parameter in the background model.","Tagged 214Pb events form the first low-energy ER calibration dataset taken concurrently with science data; its spread around the predicted beta-decay band matches the LZ-tuned response model (KS p-value 0.319) and was used to confirm the origin of reduced-charge-yield 124Xe double-electron-capture events.","The same transport model pairs 218Po with its granddaughter 214Po (via 214BiPo events) out to 81 minutes, providing a pure, high-statistics validation channel for any extension of the tag.","The 12-point gap between modeled (75%) and observed (63%) tag efficiency is attributed to extrapolating the 218Po+-only drift calibration to 214Pb+ and 214Bi+; the paper identifies backward-projected streamlines from 214BiPo events as a concrete upgrade that would also recover neutralized-ion paths.","Separately from the tag, the paper shows the detector's mixing state itself is a background control: the Minimal Mixing state reduces 218Po activity in the 5.5-tonne fiducial volume by nearly 20% relative to the Low Mixing state used for science data."],"supporting_citations":[{"why":"Supplies the 214Pb single-scatter background model and the ER spectral decomposition used to measure ϵtot in the 18–75 keVee fit.","marker":"[10]"},{"why":"Provides the first-science-run data, single-scatter event cuts, and the ER/NR band definitions the tag operates within.","marker":"[9]"},{"why":"Provides the ion-drift behavior and charge-branching measurements (0.50 ± 0.03) against which the drift model and branching fractions are anchored.","marker":"[36]"},{"why":"Gives the 214Pb half-life and the 214Pb→214Bi ground-state branching (12.7(9)%) that set the tag window length and the dominant systematic on ϵtot.","marker":"[13]"},{"why":"The 2024 WIMP search where the voxelized binary tag was deployed with ϵtot as a constrained floating parameter.","marker":"[25]"},{"why":"The prior 214Pb tagging effort in the field that this tag exceeds in efficiency by more than an order of magnitude.","marker":"[24]"},{"why":"Supplies the DAQ live fraction used in the expected pair-selection efficiency.","marker":"[37]"},{"why":"The open-source software providing the Gaussian-kernel flow interpolation and Runge-Kutta streamline integration.","marker":"[38]"}],"fun_headline_variants":["LZ flow maps veto 63% of 214Pb background","Predicting single atom decays cuts LZ's top background 63%","Flow tagging lets LZ drop its top background by 63%","LZ tags 214Pb through flow to remove 63% of background"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The tag's charged branch assumes that an ion-drift model fitted to 218Po+ ions over the first 15 minutes — with one species-independent neutralization time of 49 minutes — correctly predicts where 214Pb+ and 214Bi+ ions go for up to 81 minutes.","fun_headline_variants_meta":{"raw":{"variants":["LZ flow maps veto 63% of 214Pb background","Predicting single atom decays cuts LZ's top background 63%","Flow tagging lets LZ drop its top background by 63%","LZ tags 214Pb through flow to remove 63% of background"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000582,"raw_usage":{"total_tokens":2561,"prompt_tokens":713,"completion_tokens":1848,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":457,"completion_tokens_details":{"reasoning_tokens":1770}},"tokens_in":457,"tokens_out":1848,"duration_ms":13050,"temperature":1.0,"reasoning_tokens":1770,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T15:56:53.348329+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the drift of 214Pb+ and 214Bi+ directly: the semi-charged 218Po–214BiPo pairs (where one of the two intervening ions is neutral) form a population whose vertical displacement at late times (ΔT > 40 min) depends on the true drift parameters. If their late-time vertical deviations fall where the 218Po+-based extrapolation predicts, the charged-branch assumption holds; if they systematically deviate — or if the 75% minus 63% efficiency gap persists once species-specific drift is incorporated — the transport model, not the neutralization or branching assumptions, is the limiting piece.","supporting_citations":[{"cited_title":"Design and production of the high voltage electrode grids and electron extraction region for the LZ dual-phase xenon time projection chamber","cited_arxiv_id":"2106.06622","evidence_quote":"Supplies the 214Pb single-scatter background model and the ER spectral decomposition used to measure ϵtot in the 18–75 keVee fit."},{"cited_title":"Delaunay et al","cited_arxiv_id":null,"evidence_quote":"Provides the ion-drift behavior and charge-branching measurements (0.50 ± 0.03) against which the drift model and branching fractions are anchored."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the 214Pb half-life and the 214Pb→214Bi ground-state branching (12.7(9)%) that set the tag window length and the dominant systematic on ϵtot."},{"cited_title":"Radon Emanation Techniques and Measurements for LZ","cited_arxiv_id":"2211.11857","evidence_quote":"The prior 214Pb tagging effort in the field that this tag exceeds in efficiency by more than an order of magnitude."},{"cited_title":"Perry, Background Mitigation in LZ and Next Gener- ation Rare Event Searches, Ph.D","cited_arxiv_id":null,"evidence_quote":"Supplies the DAQ live fraction used in the expected pair-selection efficiency."},{"cited_title":"Measurements of the ion fraction and mobility of alpha and beta decay products in liquid xenon using EXO-200","cited_arxiv_id":"1506.00317","evidence_quote":"The open-source software providing the Gaussian-kernel flow interpolation and Runge-Kutta streamline integration."}],"review_version":1}