Search for New Physics via Low-Energy Electron Recoils with a 4.2 Tonne\times Year Exposure from the LZ Experiment

Technical Summary: Search for New Physics via Low-Energy Electron Recoils with the LZ Experiment

Problem and Motivation
While dual-phase xenon time projection chambers (TPCs) like LUX-ZEPLIN (LZ) are primarily designed for nuclear recoil (NR) searches targeting Weakly Interacting Massive Particles (WIMPs), they possess high sensitivity to electronic recoil (ER) signals. These ER signals offer a pathway to probe a broad spectrum of new physics beyond standard WIMP candidates. This paper reports on searches for new physics models that produce ER signals, utilizing data from the first two science runs of the LZ experiment (WS2022 and WS2024). The analysis targets specific scenarios: solar axion-like particles (ALPs), bosonic dark matter (DM) absorption, mirror dark matter (MDM), and electromagnetic interactions of solar neutrinos (magnetic moment and millicharge). A key challenge addressed is the mitigation of dominant ER backgrounds, specifically the $\beta$ decay of $^{214}\text{Pb}$ originating from radon ($^{222}\text{Rn}$) contamination.

Methodology
The analysis utilizes a total exposure of 4.2 tonne-years (0.9 from WS2022 and 3.3 from WS2024) with a fiducial mass of $5.5 \pm 0.2$ tonnes. The detector operates at the Sanford Underground Research Facility (SURF) and employs a dual-phase xenon TPC surrounded by active veto systems to suppress backgrounds.

• Signal Modeling:

  • Solar ALPs: Models include Primakoff conversion (photon coupling $g_{a\gamma\gamma}$), ABC processes (atomic recombination, bremsstrahlung, Compton scattering; electron coupling $g_{ae}$), and $^{57}\text{Fe}$ nuclear de-excitation (effective nucleon coupling $g_{aN}^{\text{eff}}$). Notably, this analysis retains mass dependence in both flux and cross-section models and explores the inverse Primakoff process for $^{57}\text{Fe}$ ALPs for the first time.
  • Bosonic DM: Searches for keV-scale pseudoscalar (ALP) and vector (hidden photon) DM absorption in liquid xenon, producing monoenergetic ER lines.
  • Mirror Dark Matter: Models mirror electrons scattering with xenon electrons via photon-mirror photon mixing, including annual/diurnal modulation.
  • Solar Neutrinos: Investigates electromagnetic interactions, specifically the neutrino magnetic moment ($\mu_\nu$) and millicharge ($\delta Q$).

• Data Selection and Analysis Strategy:

  • Two Regions of Interest (ROIs) are defined. The 2D ROI uses the standard $(S1_c, S2_c)$ observable space with an upper bound of $S1_c < 80$ phd, suitable for mirror dark matter scattering and solar neutrino electromagnetic interaction signals. The 1D ROI uses reconstructed energy ($E_{\text{rec}}$) with an upper bound of $E_{\text{rec}} < 20$ keV. This 1D approach is applied to solar ALPs and bosonic DM absorption to improve efficiency for monoenergetic signals (e.g., 14.4 keV $^{57}\text{Fe}$ ALPs) where the 2D ROI's $S1_c$ cut causes efficiency roll-off.
  • Radon Tagging: A novel technique is applied to the WS2024 dataset to identify $^{214}\text{Pb}$ events. By mapping liquid xenon flow and tracking $^{222}\text{Rn}$-$^{218}\text{Po}$ decay pairs, a "tag volume" is defined. Events within this volume and time window are treated as a $^{214}\text{Pb}$-rich sideband to constrain background rates and tagging efficiency within the likelihood fit.
  • Statistical Treatment: A profile likelihood ratio (PLR) approach is used. The analysis incorporates a signal-dependent treatment and includes radon-tagged events as a sideband to constrain the dominant $^{214}\text{Pb}$ background. Synthetic "salt" events were used to validate the analysis against bias, with the 1D ROI region remaining unsalted only after all analysis choices were frozen.

Key Contributions

1. First Application of Radon Tagging in LZ ER Searches: This work introduces the radon-tagging technique to low-energy electron recoil searches, significantly improving the ability to constrain the dominant $^{214}\text{Pb}$ background.
2. Signal-Dependent Analysis: The paper adopts a dual-ROI strategy (1D energy-only and 2D $S1_c$-$S2_c$) tailored to the spectral characteristics of different signal models, optimizing sensitivity for both scattering and absorption processes.
3. New Physics Models: This is the first experimental search for the inverse Primakoff process for $^{57}\text{Fe}$ solar ALPs. It also provides the first experimental constraints on the product coupling $g_{a\gamma\gamma}g_{a}^{\text{eff}}$ for $^{57}\text{Fe}$ ALPs.
4. Expanded Exposure: The analysis combines 4.2 tonne-years of data, representing a significant increase in exposure over previous LZ ER studies.

Results
The observed data are consistent with the background-only hypothesis, with no significant excess found. Consequently, the paper sets upper limits (90% C.L.) on various new physics parameters:

• Solar ALPs:
  • Electron coupling ($g_{ae}$): $< 1.8 \times 10^{-12}$ (15% improvement over leading terrestrial constraints).
  • Photon coupling ($g_{a\gamma\gamma}$): $< 3.5 \times 10^{-10} \text{ GeV}^{-1}$ (most stringent for masses $\sim 1 \text{ eV}/c^2$ and above, 30% improvement).
  • $^{57}\text{Fe}$ ALP rate: $< 14$ events tonne$^{-1}$ year$^{-1}$ (30% improvement on $g_{ae}g_{a}^{\text{eff}}$).
• Solar Neutrinos:
  • Magnetic moment ($\mu_\nu$): $< 8.3 \times 10^{-12} \mu_B$ (a factor of two stronger than the previous WS2022 result).
  • Millicharge ($\delta Q$): $< 1.6 \times 10^{-13} e_0$ (current leading limit).
• Dark Matter:
  • Mirror Dark Matter: Excludes temperatures $T_{\text{DM}} > 0.29 \text{ keV}$; sets kinetic mixing $\epsilon < 4 \times 10^{-12}$ at $T_{\text{DM}} = 0.5 \text{ keV}$ (10% improvement over previous best).
  • Bosonic DM: Limits are comparable to leading constraints from XENONnT and PandaX-4T.

Significance and Claims
The authors claim these results represent the most stringent constraints to date on keV-scale Primakoff and $^{57}\text{Fe}$ solar ALPs, mirror dark matter, and neutrino millicharge. For bosonic dark matter, the constraints are comparable to leading constraints, while for solar neutrino magnetic moment and ABC solar ALPs, the results remain highly competitive with other leading experiments (XENONnT, PandaX-4T). The overall sensitivity is slightly stronger in several channels due to the larger exposure and radon tagging.

The paper emphasizes that the introduction of radon tagging improved the median sensitivity to signal events by approximately 10–25% in toy datasets, though the observed gain was more modest (5–10%) due to systematic uncertainties in tagging efficiency. The work demonstrates that liquid xenon TPCs are powerful probes for keV-scale electron-coupling new physics. The authors project that with next-generation baseline designs and improved background control, sensitivity could improve by an order of magnitude in event rate, potentially reaching astrophysical constraints for neutrino magnetic moments.