LZ: xenon signals, calibration and a high-energy search

An unexplained event in a wider search

LUX-ZEPLIN, or LZ, reported a single unexplained event on Sep 1, 2026 at TeV Particle Astrophysics in Tendo, Japan. The extended-energy search used 220 live days and 2.84 tonne-years of exposure, reaching nuclear recoil energies of approximately 270 keV. The event is consistent with a 248 ± 23 (statistical) ± 23 (systematic) keV nuclear recoil.

The analysis has a global significance of 2.6 sigma after accounting for searches across models, versus a maximum local significance of 3.4 sigma. Neither is the probability that dark matter caused the event. The collaboration calls the analysis non-blind because artificial test events did not adequately cover the high-energy signal region. Selections and models were finalized before four artificial events were revealed. Pulse-shape analysis also cannot conclusively distinguish an electron recoil from a nuclear recoil. The result remains inconclusive.

A xenon target deep underground

LZ operates in the Davis Cavern at the 4,850-foot level of the Sanford Underground Research Facility in Lead, South Dakota, United States. Its central instrument is a liquid-xenon time projection chamber, or TPC, sensitive to low-energy nuclear recoils. The experiment searches directly for interactions of hypothetical weakly interacting massive particles, or WIMPs, with its target. Two surrounding systems enhance background rejection: a xenon skin detector and an outer detector containing gadolinium-loaded liquid scintillator. The latter captures and tags neutrons, which can otherwise produce recoil signals that resemble the interactions being sought.

Fields within the liquid and gas

High-voltage electrodes establish electric fields in the bulk liquid xenon, across its surface and in the gas above. LZ uses four woven-mesh wire grids for this purpose. Their design and construction are important to the operation of a dual-phase TPC, which must sustain fields in these different regions. The electron-extraction region is a particular focus of the grid design: it is where the liquid and gas portions of the signal path meet.

The grid engineering programme covers the production process and validation before integration into the chamber. The electrodes are physical components with fabrication and testing requirements, rather than numerical settings applied only during analysis. Their woven-wire construction and the field requirements in liquid, at the surface and in gas belong to the same detector design. The grid work addresses these hardware requirements for dark matter searches and other rare-event measurements with a dual-phase xenon chamber.

From light sensors to stored waveforms

LZ's FADR data-acquisition system samples signals from 745 photomultiplier tubes distributed across three detector subsystems. Its DDC-32 digitizers have 32 channels and sample at 100 MHz. On-board field-programmable gate arrays perform basic waveform analysis, extracting information about scintillation and electroluminescence signals. That information determines which digitized waveforms are preserved for subsequent offline analysis. The system therefore performs event selection while acquiring the sensor signals, before the retained data reach the later reconstruction and physics-analysis stages.

FADR stands for FPGA-based Architecture for Data acquisition and Realtime monitoring. The hardware is built around Kintex-7 FPGAs, with associated software and firmware. Its programmable electronics also monitor the detector and the acquisition system in parallel with ordinary data collection. Monitoring is an additional function alongside digitization and real-time event selection. The 745-sensor figure belongs to this implemented acquisition architecture; it should not be substituted with a sensor count from an earlier conceptual design.

Monitoring the outer detector's light response

The outer detector is designed to identify neutrons that scatter once in the xenon and could mimic a WIMP interaction. Its optical calibration system monitors the photomultiplier tubes that detect light in this surrounding veto system. The 2021 optical-system design specifies approximately 17 tonnes of gadolinium-loaded liquid scintillator in ten acrylic tanks around the xenon cryostat, with 228,000 litres of water forming the outermost layer. These are the quantities specified for that dated design.

The same design uses 120 inward-facing photomultiplier tubes, each eight inches across, to monitor the outer detector. A dedicated optical system was designed and built to calibrate and track their response. The light sensors, scintillator tanks and calibration equipment serve the neutron-veto function together. Their calibration concerns the surrounding detector's light response and required veto efficiency, rather than establishing that an individual event inside the xenon target was caused by dark matter.

Checking the scintillator's own radioactivity

The LZ Screener is a smaller detector containing approximately 23 kilograms of gadolinium-loaded, linear-alkylbenzene-based liquid scintillator. It operated inside the former LUX water shield in the Davis Laboratory to assay scintillator intended for LZ's outer detector. Carefully selected materials and photomultiplier tubes with very low backgrounds allowed the Screener to measure radioactive impurities. Its pulse-shape discrimination separated alpha and beta events, permitting isotope concentrations to be inferred by fitting their collected energy spectra. The measured carbon-14 to carbon-12 ratio was (2.83 ± 0.06 statistical ± 0.01 systematic) × 10⁻¹⁷.

The design requirement placed radioimpurities below approximately 0.07 millibecquerels per kilogram, corresponding to roughly 50 events per second above 100 keV. The Screener analysis instead estimated an outer-detector impurity rate of 97.9 ± 6.4 per second above that threshold, including 65.5 ± 1.9 per second from alpha decays. Those are predictions from the assay, not a claim that the original rate requirement was met. The measurements identified broken decay-chain equilibrium and a significant contribution from the actinium-227 subchain of the uranium-235 series.

Material selection and surface contamination

LZ's radioactivity-control programme included six years of material assays to inform construction choices and supply inputs to the background model. Materials with inherently low radioactivity were selected alongside the experiment's active inner and outer veto systems. The programme also examined contamination deposited on component surfaces, including dust and radon daughters. Controlling the material before installation and maintaining cleanliness during construction address different routes by which radioactive backgrounds can enter the assembled instrument.

Photomultiplier tubes underwent assays for fixed contaminants, while major subsystems were tested for radon emanation. Quality-control and quality-assurance procedures extended through fabrication. Separate detector systems assessed the scintillator. The programme therefore covered bulk components, surfaces and substances that could emerge from installed equipment, rather than relying on a single material-purity specification. Its historical sensitivity projection assumed a 5.6-tonne fiducial volume and 1,000 days of exposure; those design assumptions are distinct from the exposure and selected volume of the 2026 high-energy analysis.

Distributing known sources through the xenon

LZ introduces gaseous calibration isotopes through its xenon circulation system, overcoming the target's shielding against external radiation and providing more spatially uniform calibration. Observing their mixing also helps characterize xenon flow. Tritium and carbon-14 are supplied in labelled methane mixed with xenon carrier gas, while krypton-83m, xenon-131m and radon-220 come from generator sources that retain the parent isotope. Parent retention prevents long-lived contamination of the source-injection and circulation plumbing. The noble-element daughters can escape the generator material and pass through the getter during injection. Krypton-83m calibrates position-dependent light and charge collection and measures electron lifetime, which characterizes electron survival during drift and gauges xenon purity. Its 9.4 and 32.1 keV decays also support electric-field mapping through their relative light signals. Xenon-131m serves similar purposes but is preferred when mixing takes days rather than hours. These uses are distinct:

• Light and charge maps correct position-dependent collection.
• Electron-lifetime measurements track ionization-electron survival.
• Dispersed activity traces circulation and mixing within the target.

Testing reconstruction with simulated signals

LZ's simulation framework includes event generators for background radiation and models of the detector's physical response. It produces raw signals that can be converted into digitized waveforms resembling those from the operating detector. Mock data then support validation of reconstruction and analysis software. The simulation therefore extends beyond a list of expected particle energies: the response model carries the simulated event through to the kind of waveform that the acquisition and reconstruction chain handles.

Because the detector response is included, simulated data can be processed with the same analysis routines developed for experimental data. Radioactivity measurements also informed the framework's historical sensitivity projection. These are separate uses of the modelling: forecasting sensitivity from assumed conditions and checking software with artificial data. The 2020 simulation framework supported a projected spin-independent WIMP search at a mass of 40 GeV/c²; it was not itself a measurement of an unexplained event in the later science data.

Measuring backgrounds in the operating instrument

LZ's initial-science-run background analysis compared radioactivity measured within the detector with expectations from assays performed outside it. The in-situ determinations of bulk and fixed radioactive backgrounds were consistent with those external measurements. After WIMP-search selections, the observed low-energy background rate was (6.3 ± 0.5) × 10⁻⁵ events per keV of electron-equivalent energy, per kilogram, per day. That was approximately sixty times lower than the equivalent rate reported by the preceding LUX experiment.

The energy convention matters: electron-equivalent energy in that background rate is not the nuclear recoil energy used to characterize the 248 keV candidate. The initial run set a 90% confidence-level limit of 9.2 × 10⁻⁴⁸ square centimetres for spin-independent WIMP interactions at a mass of 36 GeV/c². Its background measurements also support other analyses, including effective-field-theory interpretations and searches for neutrinoless double-beta decay. The rate, interaction limit and confidence level describe different results from that initial exposure.

When electron-capture backgrounds change the light–charge balance

Electron-capture decays of xenon-125 and xenon-127, together with double-electron-capture decays of xenon-124, generate backgrounds in dual-phase xenon detectors. At equal energy, these decays produce more light and less charge than beta decays, increasing their overlap with WIMP-like signals. Measurements in LZ examined charge yields from M-shell captures at 1.1 keV, L-shell captures at 5.2 keV and K-shell captures at 33.2 keV. The measurements used drift fields of 193 and 96.5 volts per centimetre.

Double capture from two L shells in xenon-124 produces an even stronger shift in light and charge. A model developed for these decays relates ionization density to electron–ion recombination, while the researchers identify a need for more accurate calculations. These measurements address how known radioactive processes populate the detector's signal space. Their reduced charge yield is a background-modelling issue; the measurements do not establish that an electron-capture decay caused the particular high-energy event reported in 2026.

Calibrating recoils at the low-energy end

A 2025 LZ calibration analysis used an in-situ photoneutron source to probe the response to xenon nuclei recoiling with energies of a few keV. Such measurements address a different energy regime from the extended search reaching approximately 270 keV. Under the simple elastic-scattering assumption, many dark matter models predict recoil spectra that rise approximately exponentially as the energy transferred to the target decreases. Calibrating the instrument at those small energy transfers is therefore part of the low-energy search programme.

The same programme anticipates coherent scattering of boron-8 solar neutrinos from nuclei. Such a signal resembles the expected response to dark matter with a mass of about 5.5 GeV/c² and typically produces nuclear recoils below 5 keV. The calibration analysis characterizes detector response in this regime. Its description of an expected solar-neutrino signal is not, by itself, a claim that this calibration measurement detected those neutrinos or resolved the origin of the separate high-energy candidate.

Separating small ionization signals from electrode emission

LZ can measure ionization down to a single electron, extending its potential measurements below its principal GeV-to-TeV WIMP-search range. Scattering by dark matter lighter than a GeV/c² could produce such small signals. The corresponding challenge is to understand and suppress ionization backgrounds in the few-electron regime. A separate characterization examined delayed electrons following energy deposits under different detector conditions, rather than assuming that every small ionization pulse represented a new particle interaction.

Photons emitted in coincidence with electron emission from the high-voltage grids were also quantified. A coincident-photon tag identified and rejected spontaneous grid-electron emission with high efficiency in the researchers' demonstration. The method addresses a source of very small ionization signals and provides a tool for future searches. It is distinct from the waveform and background tests applied to the high-energy event, even though both investigations rely on understanding the detector's response to ordinary processes.

Keeping the earlier search and the new result distinct

The earlier conventional WIMP analysis combined 280 live days, giving an exposure of 4.2 ± 0.1 tonne-years. Of that total, 220 live days and 3.3 ± 0.1 tonne-years were new. It introduced active tagging of electronic recoils from lead-214 beta decays and identified enhanced recombination in xenon-124 double-electron capture as a background. After artificial signal-like events used to mitigate analyst bias were removed, that analysis found no excess above expected backgrounds.

Its strongest spin-independent exclusion was 2.2 × 10⁻⁴⁸ square centimetres at 90% confidence, while its best median sensitivity was 5.1 × 10⁻⁴⁸ square centimetres, both for a 40 GeV/c² WIMP. An exclusion limit and a median sensitivity are different quantities. Those published results belong to the earlier search and its stated exposure. The separately selected higher-energy analysis examines a wider recoil window and uses a different fiducial volume, with its own background tests and inconclusive event.