LUX-ZEPLIN Dark Matter Experiment

• LUX-ZEPLIN is a multi-tonne dual-phase xenon TPC designed for detecting dark matter interactions and probing rare new physics events.
• It integrates advanced calibration systems, stringent cleanliness controls, and layered veto detectors to achieve ultra-low background rates.
• The experiment’s robust simulation and analysis pipelines validate its sensitivity to WIMP signals, solar neutrinos, and rare decay channels.

The LUX-ZEPLIN (LZ) experiment is a multi-tonne dual-phase xenon time-projection chamber (TPC) designed for direct detection of dark matter, precision low-background physics, and rare-event searches. Situated 4850 feet underground at the Sanford Underground Research Facility (SURF), Lead, South Dakota, LZ provides a large (7 t active, 5.6 t fiducial) ultrapure LXe target with advanced background rejection via multiple layers of active veto detectors and stringent material selection. The principal scientific goals include probing weakly interacting massive particles (WIMPs) as well as neutrinoless double-beta decay and additional new-physics channels, reaching WIMP-nucleon cross-section sensitivities at the $10^{-48}$ cm$^2$ level (Collaboration et al., 2019, Akerib et al., 2018, Akerib et al., 2020).

1. Detector Architecture and Signal Formation

LZ employs a dual-phase xenon TPC of 1.46 m diameter and height, containing 7 t of active LXe, of which 5.6 t define the standard WIMP-search fiducial volume. Events in the fiducial volume are characterized by prompt VUV scintillation (S1) and delayed electroluminescence (S2) signals. The S1 pulse is generated by direct excitation and recombination following an interaction, detected by two arrays of Hamamatsu R11410 PMTs (253 top, 241 bottom). Ionization electrons drift upward under an applied field and are efficiently extracted into a gas layer above the liquid where they produce the S2 pulse. Three-dimensional event reconstruction uses drift time (z) and the S2 hit pattern (x, y). The S1 and S2 signals undergo spatial correction using in situ $^{222}$Rn–$^{218}$Po mapping and $^{83m}$Kr calibration (Collaboration et al., 2019, Aalbers et al., 2 May 2024).

Surrounding the TPC, a 5–10 cm LXe "skin" instrumented with 131 PMTs provides an active gamma-ray veto. An additional 17 t Gd-loaded liquid scintillator outer detector (OD), viewed by 120 8’’ PMTs, tags neutron captures by the characteristic 8 MeV gamma cascade, with performance ≥95% at a 200 keV threshold and <5% dead time, as validated by extensive optical calibration (Turner et al., 2021). The entire instrument is housed in a 238 t ultrapure water Cherenkov shield instrumented for muon vetoing (Collaboration et al., 2019).

2. Background Reduction and Cleanliness Control

Achieving the LZ sensitivity goal demands total electron-recoil background rates ≲$10^{-5}$ events/keV$_{ee}$/kg/day (Aalbers et al., 2022). Background sources are categorized as (a) fixed, "external" backgrounds from detector materials and (b) "bulk," internal backgrounds from radioisotopes dissolved in the xenon. Rigorous ex situ radioassays (HPGe, ICP-MS, GDMS, NAA) on all components, combined with radon emanation measurements, constrain U, Th, K, and Co contamination, guiding the selection of low-radioactivity titanium for the cryostat (A(U)<5 mBq/kg, A(Th)<3 mBq/kg) and qualifying Hamamatsu R11410 PMTs (A(U)=5.0±2.1 mBq/PMT) (Akerib et al., 2020, Aalbers et al., 2022).

Strict dust and $^{210}$Pb plate-out controls are enforced, with PTFE and metal surfaces neutralized under ionizers, and assembly in radon-reduced cleanrooms (C$_{Rn}$<0.5 Bq/m$^3$); final PTFE $^{210}$Pb plate-out is 158±13 μBq/m$^2$ (Akerib et al., 2020). The GdLS for the OD was subjected to additional purification, reducing $^{175}$Lu by ×37, complying with rate requirements (<10 Hz) (Akerib et al., 2020, Turner et al., 2021).

Material screening input feeds into the simulation-based background model, cross-validated in situ: ER spectra, $\alpha$ rates, and cosmogenic activation tails agree with radioassay-based predictions at the 20–50% level (Aalbers et al., 2022). Internal backgrounds (e.g., $^{222}$Rn, $^{220}$Rn, $^{85}$Kr, $^{136}$Xe 2$\nu\beta\beta$) are controlled through gas-handling (charcoal chromatography) and measured with in situ decay tagging and energy spectrum fits.

3. Simulations, Event Reconstruction, and Analysis

The main LZ simulation framework is BACCARAT, a GEANT4-based toolkit extended to support component-centric geometries, customizable radioactivity assignment, and fine-grained deposit recording (Collaboration et al., 2020). Event generators model full decay chains, radon emanation, surface plate-out, ($\alpha$,n) neutron spectra (SOURCES4A), spontaneous fission (FREYA), and cosmogenics (MUSIC/MUSUN), connecting directly to NEST yield models and detector-response functions (Collaboration et al., 2020).

Mathematical descriptions fundamental to LZ physics analysis include the NR differential rate per unit mass,

$$\frac{dR}{dE_R} = \frac{\rho_0}{m_\chi} \frac{\sigma_0}{2\mu_N^2} F^2(E_R)\int_{v>v_{\min}}\frac{f(v)}{v} dv,$$

and detector observables,

$$n_\gamma = N_{\rm ex} + r N_{\rm i};\quad n_e = (1-r) N_{\rm i};\quad S1 = g_1 n_\gamma;\quad S2 = g_2 n_e,$$

where $g_1$ and $g_2$ are position-dependent detection gains (Collaboration et al., 2020).

Full pipeline ensures all simulated events pass through raw photon/electron response, PMT impulse modeling, digitization, and trigger simulation, matching real data formatting (Collaboration et al., 2020). This supports side-by-side validation of reconstruction algorithms and sensitivity calculations, informing both fast (NEST) and full-detector chains for background/signal studies and mock data challenges (MDCs).

Statistical inference relies on (extended) unbinned profile-likelihood-ratio (PLR) fits in (S1, log$_{10}$S2), with nuisance parameters (background yields, energy scales, efficiencies) constrained by calibration data and radioassays, and results cross-checked by toy MC (Aalbers et al., 2022, Aalbers et al., 22 Oct 2024). Signal and background models integrate NEST-based ER/NR partitioning, full position/spatial corrections, and the latest efficiency measurements (Aalbers et al., 22 Oct 2024, Aalbers et al., 2022).

4. Detector Calibration Systems and Performance

Comprehensive detector calibration leverages both internal and external sources (Aalbers et al., 2 May 2024). Internally dispersed radioisotopes include $^{83m}$Kr (position/field mapping), $^{131m}$Xe, $^3$H and $^{14}$C (low-energy ER), and $^{220}$Rn (liquid flow, $\alpha$, $\beta$ tagging). Three rod-inserted tubes provide deployment of $\gamma$ and neutron sources (AmLi, $^{88}$YBe) with mm-level position repeatability.

Low-energy nuclear recoil calibration is anchored by a $^{88}$Y/Be photoneutron source, providing monochromatic 152 keV and 950 keV neutrons (yield: (34±1) n/s and (0.22±0.03) n/s at deployment, respectively) (Aalbers et al., 18 Sep 2025). Events with S1c $\in$ [0,40] phd and S2c $\in$ [7 e$^-$, $10^4$ phd] in a defined FV demonstrate NR efficiency reaching 33±5% above 10 keV$_\text{nr}$, with in situ Q$_y$(3.1 keV$_\text{nr}$) = (4.8±0.7) e$^-$/keV$_\text{nr}$, validating sub-2 keV threshold sensitivity (Aalbers et al., 18 Sep 2025). The NR and ER bands, and S2/S1 discrimination, are thereby calibrated at keV-scale energies.

The OD's optical calibration system utilizes FPGA-driven LED pulsers and multi-fiber injection, achieving dynamic range up to $1\times10^6$ photons, PMT gain stability 1%, pulse-width FWHM <20 ns, and channel-to-channel uniformity within 5% (Turner et al., 2021).

5. Science Program and Key Results

The main WIMP search uses a 5.5–5.6 t FV with 60–280+ day live-times, currently yielding world-leading spin-independent (SI) and spin-dependent (SD) WIMP-nucleon constraints: the observed 90% CL exclusion is $\sigma_{SI}<2.2\times10^{-48}$ cm$^2$ at $m_\chi=40$ GeV/c$^2$, surpassing previous direct-detection results (Aalbers et al., 22 Oct 2024). ER-background rates of $(6.3\pm0.5)\times10^{-5}$ events/keV$_{ee}$/kg/day (60× lower than LUX) enable sensitivity to rare processes (Aalbers et al., 2022).

Beyond WIMPs, LZ has demonstrated evidence for $^8$B solar neutrino CE$\nu$NS ($4.5\sigma$ in WS2025), constraining weak mixing at $Q=19.2$ MeV/c$^2$ and validating SM flux expectations (Akerib et al., 8 Dec 2025). The experiment also sets competitive limits on CR-boosted DM ($\sigma_0\sim4\times10^{-33}$ cm$^2$ at $m_\chi$ ~ 1 MeV/c$^2$) (Aalbers et al., 23 Mar 2025), atmospheric millicharged particles ($\epsilon$ ~ $2\times10^{-3}$) (Aalbers et al., 6 Dec 2024), and exotic new physics via low-energy ER spectra (Collaboration et al., 2021). Systematic studies of delayed/few-electron backgrounds and SPE-coincidence tagging enhance sub-GeV electron-recoil sensitivity (Akerib et al., 7 Oct 2025).

Dedication of 7 t of natural xenon (741 kg $^{134}$Xe, 646 kg $^{136}$Xe) enables double-beta decay searches. Projected median 90% CL exclusions, after 1000 days, are $$T_{1/2}^{2\nu,90\%}\left(^{134}\text{Xe}\right) = 1.7\times10^{24}$$ yr, $$T_{1/2}^{0\nu,90\%}\left(^{136}\text{Xe}\right) = 1.06\times10^{26}$$ yr (natural Xe), extendable to $1.06\times10^{27}$ yr with isotopic enrichment (Akerib et al., 2019, LUX-ZEPLIN et al., 2021).

6. Systematic Uncertainties and Validation

Extensive cross-validation underpins the LZ background model and sensitivity projection. Systematic uncertainties in material radioactivity, radon emanation, cosmogenics, OD performance, and modeling of single-electron pathologies are accounted for in the analysis, contributing typically at the 10–20% level to overall sensitivity (Collaboration et al., 2020, Aalbers et al., 2022). All key detector parameters (electron lifetime, gains, calibration yields) are measured repeatedly and stability is tracked to the percent level. OD neutron veto efficiency is benchmarked against reference MC, with systematic variations inducing ≤20% shifts in veto survival probability near threshold. Full simulation and reconstruction reproduce science data format, supporting real-time MDCs and calibration optimization (Collaboration et al., 2020, Aalbers et al., 2 May 2024).

7. Context, Impact, and Outlook

The LZ experiment realizes an unprecedented background-free environment for multi-tonne rare-event searches. Its background rejection and calibration methodology set one of the lowest ER background rates among tonne-scale LXe TPCs, as necessary to advance exploration of the "neutrino fog" regime in direct dark-matter searches (Aalbers et al., 2022, Aalbers et al., 22 Oct 2024). The platform's flexible calibration infrastructure, comprehensive background control, and robust analysis pipelines enable both flagship and ancillary science, including CE$\nu$NS, double-beta decay, millicharged particles, and nonstandard neutrino properties (Collaboration et al., 2021, Aalbers et al., 6 Dec 2024, Aalbers et al., 23 Mar 2025, Akerib et al., 8 Dec 2025). Future phases aim at 1000-day exposures, further background reductions, and upgrades toward even higher sensitivity (Aalbers et al., 2022).

LZ’s contributions thus delineate the present experimental frontier for WIMP and rare-event searches in liquid xenon, combining technical innovation in cleanliness, calibration, simulation, and analysis (Collaboration et al., 2019, Aalbers et al., 2022, Akerib et al., 2018, Collaboration et al., 2020).