Complete event-by-event $\alpha /\gamma \left(\beta \right)$ separation in a full-size ${\mathrm{TeO}}_2$ CUORE bolometer by Neganov-Luke-magnified light detection

In the present work, we describe the results obtained with a large ($\approx 133{\mathrm{cm}}^3$) ${\mathrm{TeO}}_2$ bolometer, with a view to a search for neutrinoless double-$\beta$ decay ($0\nu \beta \beta$) of ${}^{130}\mathrm{Te}$. We demonstrate an efficient $\alpha$-particle discrimination (99.9%) with a high acceptance of the $0\nu \beta \beta$ signal (about 96%), expected at $\approx 2.5$ MeV. This unprecedented result was possible thanks to the superior performance (10-eV rms baseline noise) of a Neganov-Luke-assisted germanium bolometer used to detect a tiny (70-eV) light signal from the ${\mathrm{TeO}}_2$ detector, dominated by $\gamma$($\beta$)-induced Cherenkov radiation but exhibiting also a clear scintillation component. The obtained results represent a major breakthrough toward the ${\mathrm{TeO}}_2$-based version of the CUORE Upgrade with Particle IDentification (CUPID), a ton-scale cryogenic $0\nu \beta \beta$ experiment proposed as a followup to the Cryogenic Underground Observatory for Rare Events (CUORE) project with particle identification. The CUORE experiment recently began a search for neutrinoless double-$\beta$ decay of ${}^{130}\mathrm{Te}$ with an array of 988 125-${\mathrm{cm}}^3{\mathrm{TeO}}_2$ bolometers. The lack of $\alpha$ discrimination in CUORE makes $\alpha$ decays at the detector surface the dominant background component, at the level of $\approx 0.01$ counts/(keV kg y) in the region of interest. We show here, for the first time with a CUORE-size bolometer and using the same technology as CUORE for the readout of both heat and light signals, that surface $\alpha$ background can be fully rejected.

Figure 1

Left panel: Energy spectrum of a ${}^{232}\mathrm{Th}\gamma$ source measured by a 784-g ${\mathrm{TeO}}_2$ bolometer in the EDELWEISS setup over 148.3 h. The spectrum contains a population of $\alpha$ particles from a smeared ${}^{210}\mathrm{Po}\alpha$ source. The energies of the $\gamma$ peaks visible in the spectrum are given in keV. The 2615-keV $\gamma$ peak of ${}^{208}\mathrm{Tl}$, fitted with a Gaussian function and a flat component, is depicted in the inset. The FWHM energy resolution at the peak is $6.5\pm 0.5$ keV. Right panel: amplitude spectrum of x rays excited by a ${}^{60}\mathrm{Co}\gamma$ source irradiating the holder—including the ${\mathrm{TeO}}_2$ crystal and the Ag internal layer—and measured by the Ge light detector operated at 0-V electrode bias (6.4 h of data taking). A fit to the data by a flat distribution and three Gaussian functions is shown by the red curve. The energy scale of the light detector is given in analog-to-digit units (ADU).

Figure 2

Left panel: scatter plots of the heat vs light signals in the measurements with a 784-g ${\mathrm{TeO}}_2$ bolometer in coincidence with a Neganov-Luke Ge light detector operated respectively as a standard light detector (0 V electrode bias, top) and in a signal-amplification regime (60 V electrode bias, bottom). The ${\mathrm{TeO}}_2$ detector was irradiated by $\gamma$ quanta from a ${}^{232}\mathrm{Th}$ source and $\alpha$ particles from a partially smeared ${}^{210}\mathrm{Po}$ source. Heater events were used for the thermal response stabilization. Right panel: projections onto the $y$ axis of the events marked in red in the left panel, for 0 V (top) and 60 V (bottom) electrode bias respectively. The light signal distributions are fitted with Gaussian functions. The discrimination power (see text) between $\alpha$ and $\gamma$ events is 3.17. The energy scale of the light detector is given in ADU (analog-to-digit units).

Figure 3

Average pulses of a Neganov-Luke light detector acquired in coincidence with a 784-g ${\mathrm{TeO}}_2$ bolometer with the following event selection in the main absorber: 2.6-MeV $\gamma$ quanta (333 waveforms), 5.3-MeV $\alpha$ particles (3551 waveforms), and heater-induced cross-talk (1608 waveforms). The average light signal induced by 2.6-MeV $\alpha$'s (668 waveforms) is shown in the inset together with the one induced by 5.3-MeV $\alpha$'s. The pulses are normalized at the maximum of the 2.6-MeV $\gamma$ average pulse.

Figure 4

The rejection factor of $\alpha$ events vs the acceptance level of $\gamma$($\beta$) events based on the results of the $\alpha /\gamma$($\beta$) separation in the ROI of $0\nu 2\beta$ decay of ${}^{130}\mathrm{Te}$ achieved with the present work with a 784-g ${\mathrm{TeO}}_2$ bolometer coupled to the Neganov-Luke light detector and compared with the only previous test conducted so far with a CUORE-size ${\mathrm{TeO}}_2$ crystal [22]. The right boundary of the accepted $\gamma$($\beta$) was set at $3\times {\sigma }_{\mathrm{Light}}$, while the left boundary was varied to get a selection efficiency within 75–97%. The corresponding $\alpha$ rejection efficiency was computed with an infinite left boundary, while the level of the right boundary was set according to the left boundary of the $\gamma$($\beta$) acceptance band.