Quenching factors for Na recoils as a function of Tl dopant concentrations in NaI(Tl) crystals

TL;DR

This work measures the sodium quenching factor QF_Na in five NaI(Tl) crystals with varied Tl dopant levels over $E_{nr}$ from $5$ to $26$ keV$_{\textrm{nr}}$ using quasi-monoenergetic neutrons at TUNL. A Geant4-based simulation framework, coupled with Bayesian BAT analysis, relates quenched spectra to simulated $E_{nr}$ distributions while modeling background and detector resolution; two calibration schemes, $^{133}$Ba at $6.6$ keV and $^{241}$Am at $59.54$ keV, reveal an energy-dependent QF_Na and a Tl-dopant influence that depends on calibration method. A geometric systematic, manifesting as a sawtooth pattern, is identified and mitigated to estimate an associated uncertainty. The results inform the interpretation of NaI(Tl) DM experiments and motivate broader dopant-range studies and non-linear light-yield modelling for improved energy calibration and DM sensitivity.

Abstract

Thallium-doped sodium iodide (NaI(Tl)) scintillation detectors play an important role in the field of direct dark matter (DM) searches. The DAMA/LIBRA experiment stands out for its reported observation of an annually modulating DM-like signal, which is in direct contrast with other results. To accurately calibrate the energies of nuclear recoil signals with electron recoils, precise measurements of the quenching factor of the NaI(Tl) crystals are essential, as the two processes have different scintillation light yield. In this article, we present results of a systematic study carried out by the COSINUS collaboration and Duke University to measure the quenching factor of sodium (Na) recoils as a function of nuclear recoil energy and for differing Thallium (Tl) dopant concentrations in the bulk crystal. Five ultrapure NaI(Tl) crystals, manufactured by the Shanghai Institute for Ceramics, were irradiated with a quasi-monoenergetic neutron beam at the Triangle Universities Nuclear Laboratory, North Carolina, USA. The quenching factor for low nuclear recoil energies of 5-26keV$_{nr}$ was extracted for all 5 crystals. A Tl-dependence could be deduced with a proportional response calibration schema using a $^{241}$Am source. However, this effect was not observed when using a low-energy calibration line from $^{133}$Ba.

Figures (18)

• Figure 1: Top: The experimental setup at TUNL; Bottom: A close-up of the NaI(Tl) crystal coupled to the PMT.
• Figure 2: Measured pulse shapes from the PMT attached to the NaI(Tl) crystal and BD 1 for a nuclear recoil event of $\sim\unit[60]{keV_{\textrm{ee}}}$. The BPM provides a shaped signal as timing reference for identifying beam-induced neutron events.
• Figure 3: Observed and simulated spectrum in []keV from the $^{133}$Ba calibration of Crystal 1. The measured spectrum was calibrated using the [6.6]keV proportional response. The simulated spectrum is smeared by Gaussian convolution, where the width is given by the resolution function $\sigma(E) = a\sqrt{E}$. The measured peaks are fitted with a Gaussian on top of a linear background. The [6.6]keV line is close to the noise peak, then in order of increasing energy the peaks are identified as the unresolved sum of [30.62]keV and [30.97]keV, and finally the [81.0]keV line. The vertical lines denote the fitted Gaussian mean converted to energy using the [6.6]keV proportional response. Deviations of the measured spectra from simulation are expected since the simulation does not account for the nonlinearity of the scintillation response in NaI(Tl) and the measured spectrum is calibrated using a proportional response.
• Figure 4: PSD cut (top) and Time to BPM cut (bottom) applied to triggered events of BD 0 for Crystal 1.
• Figure 5: Measured PMT spectra for Crystal 1, triggered by BD0. The impact of each cut parameter is shown, where the values are illustrated in Fig. \ref{['cuts']}.