Rare-Event Search Experiments

Updated 4 December 2025

Rare-event search experiments are precision measurements in particle and astroparticle physics focused on detecting processes with extremely low event rates using specialized detector technologies.
They employ advanced methodologies such as phonon-based calorimetry, ionization/scintillation yield measurements, and superconducting quantum sensors to achieve ultra-low energy thresholds and effective background control.
Robust background mitigation, statistical signal extraction, and integrated deep-learning techniques are key to extending sensitivity towards phenomena like dark matter interactions and neutrinoless double beta decay.

Rare-event search experiments are a class of precision measurements in particle and astroparticle physics targeting processes with extremely low rates, such as WIMP dark matter interactions, neutrinoless double beta decay (0νββ), coherent elastic neutrino–nucleus scattering, axion and hidden-photon absorption, and exotic nuclear decays. These experiments demand detector architectures, background suppression methodologies, and statistical analyses tailored to operate at ultra-low thresholds and background levels, frequently at or below 10^–4 counts/(keV·kg·yr) in their regions of interest. The operational and design principles in rare-event searches are distinguished by their focus on pushing sensitivity at or very near the physical and noise limits of the detection apparatus.

1. Fundamental Detection Principles and Technologies

Rare-event detectors are engineered to convert sub-keV energy deposits from weakly interacting processes into measurable signals with maximum efficiency and minimal noise. The principal transduction channels employed are:

A. Phonon-based calorimetry: Operating typically in the 10–100 mK range, phonon calorimeters utilize target crystals (e.g., Ge, Si, CaWO₄, Li₂MoO₄) in conjunction with sensor technologies such as transition-edge sensors (TES), neutron-transmutation-doped (NTD) thermistors, and metallic magnetic calorimeters (MMC). Particle interactions create athermal phonons, which are sensed as temperature increases ΔT ∼ E/C with C denoting heat capacity. Calibration exploits sources such as ^55Fe for X-ray and ^57Co for γ-ray events (Das et al., 2 Dec 2025, Collaboration et al., 2020).

B. Ionization and Scintillation Yield: Semiconducting and noble elements (Ge, Xe, Ar) permit parallel measurement of ionization (Q_ion) and scintillation (N_γ), yielding discrimination between nuclear and electron recoils through ratios like Y_i = Q_ion/E_dep and Y_s = N_γ/E_dep. Dual-phase TPCs, and bolometric architectures equipped with light sensors (PMT, Ge bolometers, TES photon counters) provide event-by-event ER/NR classification (Bandac et al., 2023, Zhang et al., 2020, Das et al., 2 Dec 2025).

C. Superconducting quantum sensors: TES, KID, and SNSPD technologies expand threshold capabilities to the single-eV domain by leveraging sharp transitions in resistance, kinetic inductance, or photon absorption, thus enabling searches for sub-GeV dark matter and even sub-eV axion-like particles (Kim et al., 2021, Manenti et al., 5 Feb 2024).

Typical performance metrics are summarized as follows:

Detector	Threshold (eV)	σ_E (eV RMS)	Discrimination	Reference
CRESST-III	30 eV_nr	∼10–40	Phonon/Light	(Das et al., 2 Dec 2025)
SuperCDMS	70 eV_ee	∼10–30	Ion/Phonon	(Das et al., 2 Dec 2025)
CPD (Si TES)	<10 eV	3.86	Photon Timing	(Collaboration et al., 2020)
Li₂MoO₄	∼3000 eV	3.8–6	α/γ Quenching	(Bandac et al., 2023)

2. Sources of Background and Mitigation Strategies

The quest for ultra-low backgrounds in rare-event searches involves extensive control of both physical and environmental noise sources:

A. Ambient radioactivity: Bulk and surface contamination from U/Th chains, ^40K, and ^222Rn are managed via material selection (U/Th <1 mBq/kg), surface cleaning (epitaxy, acid etching), radiopurity screening (GDMS, HPGe spectrometry), and Radon suppression (inline charcoal traps with k_a>30 L/g at cryogenic temperatures) (Pushkin et al., 2018, Cebrian, 26 Mar 2024). Nitric-acid-etched Saratech charcoal provides cost-effective ^222Rn filtration in Xe at ~2.9 L/g (Pushkin et al., 2018).

B. Cosmogenic activation: Exposure of detector components to cosmic rays induces long-lived radioisotopes (^68Ge, ^60Co, ^57Co in Ge and Cu, ^127Xe in Xe) that contribute irreducible backgrounds. Production rates at sea level reach O(10–100) nuclei/(kg·day), scaling with neutron and muon fluxes and mitigated by underground processing, shielded transport, and extended underground storage for decay (e.g., 2–3 years for ^68Ge and ^60Co suppression) (Wei et al., 2017, She et al., 2021, Zhang et al., 2016).

Material	Key Isotope	Production Rate (kg⁻¹d⁻¹)	Recommended Mitigation
Germanium (enr)	^68Ge	21.8 (unshielded)	Shielding, underground storage
Copper	^60Co	39.7	Electroforming, shielded transport
Xenon	^127Xe	233.3	Underground sourcing, filtered gas

C. Neutron and γ-ray backgrounds: Both radiogenic (rock-derived) and cosmogenic neutrons require extensive shielding, typically consisting of 30+ cm lead for γ attenuation and 40+ cm polyethylene or borated PE to moderate/capture fast and thermal neutrons. Measured fluxes at deep sites such as YangYang yield total neutron rates of (4.46 ± 0.66) × 10⁻⁵ cm⁻²s⁻¹ (Yoon et al., 2021, Ghosh et al., 2021).

D. Electronic and thermal noise: Setting energy thresholds optimally involves detailed characterization of baseline fluctuations. The method of (Mancuso et al., 2017) fits the distribution of maxima in filtered baseline windows to Gaussian order statistics, providing a recipe to select the lowest trigger threshold E_th yielding acceptable noise-trigger rates (NTR). This allows maximization of sensitivity while controlling fake events down to NTR ≲ 1/(kg·day).

3. Statistical Analysis, Signal Extraction, and Confidence Limits

Event selection in rare-event experiments employs frequentist statistical methodologies adapted to the low-count regime:

Profile likelihood ratio analysis forms the backbone of discovery and exclusion, encoding signal and background PDFs, nuisance parameters, and systematic uncertainties. Extended likelihoods are constructed as

$L(\mu, \theta) = e^{-(\mu + \sum \theta_l)} \prod_{k=1}^N [\mu f_s(x_k) + \sum \theta_l f_l(x_k)] \prod_l C_l(\theta_l)$

Signal-strength upper limits and significances derive from profile likelihood ratios λ(μ), test statistics q_μ/q_0, and the CLs prescription. When N is very small, critical values are determined through toy-MC, not asymptotic distributions (Billard, 2013, Othman, 2019). For multi-channel analyses, coincidence, pulse-shape discrimination, and topology-based vetoes further reduce background acceptance, as in the Majorana and CRESST experiments (Othman, 2019).

Rare-event searches span from kg- and ton-scale TPCs (LZ/XENONnT/NEXT) to gram- and sub-gram cryogenic calorimeters. Design considerations include:

A. Large TPCs (LXe, LAr, HPXe): 3-D spatial reconstruction, dual-phase S1/S2 discrimination, and pulse-shape analysis for ER/NR separation with background indices B≲10⁻⁴ cts/(keV·kg·yr). Deep underground operation (d>3500 m.w.e.) is standard (Cebrian, 26 Mar 2024).

B. Cryogenic bolometers: Dual-mode heat/light readout (e.g., Li₂MoO₄, CaWO₄), offering <10 keV FWHM and powerful α/γ separation. Clean-room assembly and passive+active shielding are essential (Bandac et al., 2023, Zhang et al., 2020, Kim et al., 2021).

C. Superconducting photon and charge sensors: For optical dark-photon, axion, and low-mass dark matter searches, TES and KID arrays provide sub-10 eV thresholds and ultra-low dark-count rates (e.g., R_DC = 3.6 × 10⁻⁴ Hz in 0.8–3.2 eV band) (Manenti et al., 5 Feb 2024, Collaboration et al., 2020).

D. Integration of AI-based real-time analysis: Deep-learning object detectors (YOLOv8, Faster R-CNN) now perform online event classification and topology recognition in high-throughput experiments, enabling massive frame reduction (e.g., 20 million → 826 Migdal candidates in MIGDAL, 100% recall for complex two-particle decays in GADGET II) and transferability to other rare processes (Schueler et al., 11 Jun 2024, Wheeler et al., 28 Jan 2025).

5. Surface Effects, Simulation, and Systematic Modelling

Background predictions require detailed simulation of energy deposition and transport, including:

A. Surface roughness and contamination: Geant4-based libraries like SCoRe4 simulate microscopic roughness and contamination depth profiles, improving accuracy of background estimates for β, γ, and α events escaping or scattering near surfaces (Grüner, 19 Nov 2025).

B. Calibration and benchmarking: Validation of MC predictions against measured neutron, γ, muon fluxes, and activation rates informs the design of shield geometries, cuts, and cleaning protocols (Ghosh et al., 2021, Yoon et al., 2021).

C. Systematic uncertainties: All background and efficiency estimators are examined under constraint terms (C_l(θ_l)), and profile over nuisance parameters to correctly propagate errors to discovery and exclusion statistics (Billard, 2013).

6. Impact on Physics Reach and Future Perspectives

State-of-the-art rare-event searches have achieved record sensitivities in dark matter cross section (σ_n^{SI ≈ 10⁻⁴⁸ cm²,} XENONnT/LZ), 0νββ half-lives (T₁/₂^{0ν>10²⁶–10²⁷ yr,} CUORE/CUPID), and first reactor CEνNS observation in HPGe (CONUS+). Ongoing advances focus on:

Lower thresholds via quantum sensors (TES, KIDs, SNSPDs) and phonon-amplification (NTL effect).
Improved background rejection with active vetoes, surface engineering, and adaptive hardware-software architectures.
Application of real-time deep learning for event selection and background characterisation in high-throughput imagers.
Expansion to search for sub-eV and ultralight new physics candidates.

Continued evolution in sensor multiplexing, scaling of dilution refrigeration, and high-fidelity simulation will further extend reach toward the detection of ever more elusive rare phenomena (Das et al., 2 Dec 2025, Kim et al., 2021, Schueler et al., 11 Jun 2024).