Sensitivity of the CUPID experiment to $0νββ$ decay of $^{100}$Mo

CUPID is a next-generation bolometric experiment to search for neutrinoless double-beta decay ($0νββ$) of $^{100}$Mo using Li$2$MoO$4$ scintillating crystals. It will operate 1596 crystals at $\sim$10 mK in the CUORE cryostat at the Laboratori Nazionali del Gran Sasso in Italy. Each crystal will be facing two Ge-based bolometric light detectors for $α$ rejection. We compute the discovery and the exclusion sensitivity of CUPID to $0νββ$ in a Frequentist and a Bayesian framework. This computation is done numerically based on pseudo-experiments. For the CUPID baseline scenario, with a background and an energy resolution of $1.0 \times 10^{-4}$ counts/keV/kg/yr and 5 keV FWHM at the Q-value, respectively, this results in a Bayesian exclusion sensitivity (90% c.i.) of $\hat{T}{1/2} > 1.6 \times 10^{27} \ \mathrm{yr}$, corresponding to the effective Majorana neutrino mass of $\hat{m}{ββ} < \ 9.6$ – $28 \ \mathrm{meV}$. The Frequentist discovery sensitivity (3$σ$) is $\hat{T}{1/2}= 1.0 \times 10^{27} \ \mathrm{yr}$, corresponding to $\hat{m}{ββ}= \ 12$ – $36 \ \mathrm{meV}$.

💡 Research Summary

The paper presents a comprehensive sensitivity study for the next‑generation bolometric experiment CUPID, which aims to search for neutrinoless double‑beta decay (0νββ) of ^100Mo using Li₂MoO₄ scintillating crystals. The authors describe the experimental configuration in detail: 1596 cubic crystals (45 mm per side, ~280 g each) enriched to ≥95 % in ^100Mo will be installed in the existing CUORE cryostat at the Laboratori Nazionali del Gran Sasso. Each crystal is coupled to two germanium‑based light detectors equipped with Neganov‑Trofimov‑Luke (NTL) amplification to achieve α‑particle rejection above 99.9 %. The projected detector mass is 450 kg (240 kg of ^100Mo), with a containment efficiency of 79 % and a selection efficiency of 90 %. The baseline design assumes an energy resolution of 5 keV FWHM at the Q‑value (3034 keV) and a background index of 1 × 10⁻⁴ counts · keV⁻¹ · kg⁻¹ · yr⁻¹ in a ±3σ region of interest.

The statistical methodology combines both Frequentist and Bayesian approaches. An extended unbinned likelihood is constructed for the observed energy spectrum, with the signal rate expressed as the inverse half‑life Γ = 1/T₁/₂ and nuisance parameters describing the background. Pseudo‑experiments are generated thousands of times to capture statistical fluctuations. In the Frequentist framework, a 3σ discovery sensitivity (≈99.7 % confidence) is defined as the half‑life for which the median experiment would yield a test statistic exceeding the 3σ threshold. In the Bayesian framework, a uniform prior on Γ is adopted, and the 90 % credible interval is used to set an exclusion limit.

The main results for the baseline scenario are:

• Bayesian 90 % credible exclusion limit: T₁/₂ > 1.6 × 10²⁷ yr, corresponding to an effective Majorana neutrino mass mββ < 9.6–28 meV (range reflects nuclear matrix element uncertainties).
• Frequentist 3σ discovery sensitivity: T₁/₂ ≈ 1.0 × 10²⁷ yr, corresponding to mββ ≈ 12–36 meV.

These sensitivities would allow CUPID to fully explore the inverted‑ordering (IO) region of the neutrino mass hierarchy and to begin probing the normal‑ordering (NO) region for mββ ≈ 10 meV. The authors also study the impact of varying background and energy resolution. Reducing the background index to 0.6 × 10⁻⁴ counts · keV⁻¹ · kg⁻¹ · yr⁻¹ improves the exclusion limit by roughly 30 %, while degrading the resolution to 10 keV worsens the sensitivity by about 20 %. Conversely, improving the resolution to 7.5 keV yields a modest gain.

The paper discusses the sources of background in detail: environmental radiation (muons, neutrons), intrinsic radioactivity (214Bi, 208Tl), and pile‑up from 2νββ decays. Simulations based on GEANT4, informed by measurements from CUORE and the CUPID‑Mo demonstrator, predict a dominant contribution from nearby detector components (≈4 × 10⁻⁵ counts · keV⁻¹ · kg⁻¹ · yr⁻¹) and a sub‑dominant pile‑up component that can be suppressed below 5 × 10⁻⁵ counts · keV⁻¹ · kg⁻¹ · yr⁻¹ using the NTL‑enhanced light detectors. Additional shielding (plastic scintillator veto, water Cherenkov detector, polyethylene layers) is expected to bring the muon‑induced background down to 1 × 10⁻⁶ counts · keV⁻¹ · kg⁻¹ · yr⁻¹.

The authors emphasize that the projected sensitivities are robust against reasonable variations of the experimental parameters, but they also highlight the importance of achieving the targeted background reduction and energy resolution. The study provides a quantitative framework for optimizing detector design, guiding R&D priorities (e.g., crystal surface cleaning, sensor optimization, NTL amplification), and informing the physics reach of CUPID within the broader landscape of 0νββ experiments.

In conclusion, the analysis demonstrates that CUPID, building on the operational experience of CUORE and leveraging scintillating bolometer technology, is poised to achieve a half‑life sensitivity of order 10²⁷ yr. This represents a significant step forward in the quest to observe neutrinoless double‑beta decay, potentially confirming the Majorana nature of neutrinos and providing critical insight into the absolute neutrino mass scale.