COSINE-100 NaI Experiment

• COSINE-100 NaI Experiment is a dark matter search using ultrapurified NaI(Tl) crystals and multi-layer shielding to minimize background interference.
• Its advanced data acquisition and calibration systems enable low-energy threshold detection and robust discrimination of signal events.
• The experiment aims to decisively test DAMA/LIBRA’s annual modulation claim by achieving near 5σ sensitivity for low-mass dark matter candidates.

The COSINE-100 NaI Experiment is a direct detection dark matter search utilizing an array of ultra-low-background thallium-doped sodium iodide (NaI(Tl)) crystals. Operating at the Yangyang Underground Laboratory in South Korea, COSINE-100 was initiated with the specific aim to provide a model-independent test of the annual modulation signal observed by DAMA/LIBRA using the same target material. Its design emphasizes stringent background suppression, high radiopurity, and advanced data acquisition and signal discrimination methods. The experiment has undergone successive improvements in background modeling, analysis threshold, and sensitivity, probing both canonical and non-standard models of WIMP-nucleus and WIMP-electron interactions.

1. Detector Architecture and Shielding

COSINE-100’s core comprises eight ultra–low-background NaI(Tl) crystals, individually encapsulated in oxygen-free copper tubes and optically coupled at both ends via quartz windows to Hamamatsu 3-inch PMTs. The total crystal mass is approximately 106 kg, and the detectors are arranged in a 4 × 2 matrix atop a two-layer acrylic table. Each crystal is wrapped in PTFE for enhanced reflectivity. The array is housed within a concentric, multilayer shield:

• External Veto: Plastic scintillator panels (3 cm thick) for cosmic-ray muon tagging.
• Passive Shielding: 20 cm-thick “brick castle” lead shield (with low-activity interior), followed by a 3 cm-thick oxygen-free copper box.
• Active Inner Shield: >2200 L of LAB-based liquid scintillator (LS) in an acrylic vessel, read out by eighteen 5-inch PMTs, providing a minimum of 40 cm radial separation from the crystals and maximizing light collection through specular reflective film on inner walls and crystal surfaces.

This shielding scheme achieves close to $4\pi$ background suppression and facilitates efficient muon and gamma vetoing, as well as anti-coincidence tagging of radioactive backgrounds (notably, $^{40}$K-induced events).

2. Operational and Data Acquisition Systems

The experiment utilizes a dual-level data acquisition and environmental monitoring system:

• DAQ Electronics: Fast FADC modules (500 MS/s) digitize high-gain anode and lower-gain dynode PMT signals per crystal. Event triggers require both PMTs on a crystal to exceed threshold within 200 ns, achieved via a dedicated Trigger and Clock Board that synchronizes all sub-systems. The LS and plastic scintillator panels use slower, charge-integrating ADCs with appropriate coincidence and amplitude logic.
• Active Vetos: Plastic scintillator panels provide a measured underground muon flux at $\sim$328 ± 1(stat.) ± 10 muons/m²/day. The LS system is used in time-coincidence with crystal events (200 ns window, $\sim$20 keV threshold), giving 70–80% tagging efficiency for $^{40}$K background events.
• Environmental Monitoring: Temperature, humidity, radon levels, and HV status are monitored with data logged in InfluxDB and visualized via Grafana, enabling correlation studies between environment and detector performance.

3. Performance Metrics and Calibration

• Light Yield: COSINE-100 crystals display high light yields (in photoelectrons/keV), exceeding or matching those in DAMA/LIBRA, with several crystals showing twice the light yield, thus favoring a lower energy threshold in the analysis.
• Stability: Environmental temperature is maintained at $23.5^{\circ}C$ (±0.3°C), and relative humidity at ~40%. Physics-quality live time exceeds 95%, with stable crystal trigger rates across extended data-taking periods.
• Calibration: Energy calibration uses $^{241}$Am, $^{57}$Co, $^{137}$Cs, $^{60}$Co γ-ray sources and internal spectral lines ($^{210}$Pb 46.5 keV, $^{40}$K 3 keV). Calibration of both anode (high-gain) and dynode (low-gain) channels is validated via these references.

4. Background Rejection and Modeling

• Software Cuts: Advanced pulse shape discrimination (PSD) using fast/slow charge ratio and a charge asymmetry variable,

$\text{Asymmetry} = \frac{Q_1 - Q_2}{Q_1 + Q_2}$

is employed to reject PMT noise and non-scintillation events, especially important at low energies.

• Likely Background Sources: After application of vetoes and cuts, the dominant backgrounds in the 2–20 keV region arise from internal $^{210}$Pb and $^{3}$H (tritium), with background levels at $\sim$2 counts/day/kg/keV in selected crystals.
• Monte Carlo Simulations: GEANT4-based simulations, constrained by laboratory assay of material radiopurities, are developed to model the detector response—including photon generation, photoelectron conversion, and digitization. These are crucial for quantifying tagging efficiencies and understanding residual backgrounds, especially for cosmogenic and surface contaminants.

5. Annual Modulation Search and Sensitivity

The experiment is designed for multi-year operation and is explicitly optimized for model-independent evaluation of the DAMA/LIBRA annual modulation claim:

• Projected Exposure: A minimum two-year stable run aims for a 212 kg·year exposure, probing both low- and high-mass WIMP regions (Na- and I-scatter dominated, respectively).
• Sensitivity: With a 1 keV threshold, COSINE-100 projects the ability to exclude the DAMA-preferred low-mass region at near 5σ and much of the high-mass region at $90\%$\,CL if no annual modulation is detected.
• Nuclear Recoil Calibration: Nuclear recoil measurements with monoenergetic D–D fusion neutron sources (2.42 MeV) are in progress to empirically establish quenching factors—critical for interpreting any modulation signal in terms of WIMP interactions.

6. Ongoing and Prospective Upgrades

• Noise Discrimination Refinement: Improvements in noise rejection and the development of comprehensive detector simulation frameworks (covering photon production, propagation, and DAQ response) are underway to further reduce uncertainties in the event selection near threshold.
• Background Modeling: Detailed GEANT4-based studies, with updated material assay inputs, continue to improve the background model and the reliability of Monte Carlo–data matching in both low- and high-energy regions.
• Long-Term Operation: The stability of environmental and operational parameters underpins COSINE-100’s plan for multi-year exposure.
• Preparations for COSINE-200: The experience and technologies developed in COSINE-100 lay the groundwork for the expanded 200 kg NaI(Tl) COSINE-200 experiment, which will build on the mass production of ultra-pure crystals and further background reductions.

7. Significance and Impact

COSINE-100 represents a comprehensive, multi-layered experimental system, robust against both statistical and systematic backgrounds. Early results have validated its design choices—combining high light yield, stringent radiopurity, and a hierarchy of active and passive background suppression. The experiment is positioned to provide a decisive test of DAMA/LIBRA’s long-standing annual modulation claim by leveraging identical target material and superior background control. Its performance demonstrates the viability and scalability of NaI(Tl)-based, low-threshold experiments, and the ongoing developments in calibration, modeling, and analysis methods establish a foundation for future sensitivity gains in the direct search for WIMP dark matter.