Scintillation response of cryogenic CsI to few-keV and sub-keV nuclear recoils

Abstract

Monochromatic neutron emissions from photonuclear sources $^{88}$Y/Be and $^{124}$Sb/Be are employed to obtain the response of pure (undoped) cesium iodide at 80 K. The use of a low-noise, high-quantum-efficiency avalanche photodiode in combination with a novel waveshifter results in a 70 eV analysis threshold. This reach allows to observe signals from sub-keV nuclear recoils originating in neutron scattering. The extracted quenching factor drops much faster towards low energy than the extrapolation of a model developed for room-temperature CsI[Na]. We comment on the impact of our measurement on planned use of cryogenic CsI in neutrino physics and dark matter experiments.

Figures (8)

• Figure 1: Simulated nuclear recoil distributions produced by the dominant neutron emission from $^{88}$Y/Be and $^{124}$Sb/Be in the geometry of Fig. 2. Cs and I recoils are indistinguishable due to their similar nuclear mass NIMcenns. Vertical arrows indicate the maximum recoil energy from single scatters. A tail beyond originates in the modest fraction of multiple scattering by neutrons in this small crystal ($\sim$10.7 % for both sources).
• Figure 2: Elements of the $^{88}$Y irradiation geometry: 1) V-vial containing evaporated $^{88}$Y source, 2) stainless steel (SS) container, 3) beryllium oxide ceramic or aluminum metal, 4) aluminum endcap and infrared shield, 5) PTFE-wrapped cesium iodide crystal, 6) holder (sponge), 7) acrylic window coated with NOL-9 waveshifter, 8) LAAPD on alumina substrate, 9) copper cold finger and cold plate, 10) SS cryostat flange. $^{124}$Sb irradiations varied only in internal source geometry.
• Figure 3: Top: preamplifier trace from an 82 eV scintillation signal in the CsI crystal. Fitted rise time and LAAPD gain are indicated. Bottom: the same signal digitally shaped with the integration time shown. Using a light yield of 106 photons/keV at 80 K csiqf and a 70% LAAPD quantum efficiency (see text), this corresponds to the detection of $\sim$6 photons and a trigger threshold at 4 photons, similar to that in rmd2.
• Figure 4: Top: preamplifier rise time of scintillation ($\sim$1 $\mu$s) and LAAPD ($\sim$100 ns) signals during a Y/Be run. The 5.9 keV $^{55}$Fe deposition is visible for the first. The rise time of a charge-trapping noise noticeable below 1 keV (scintillation) is broadened by the effect of electronic noise (Fig. 3). A grayed region indicates events accepted for analysis (see text). Bottom: same during a $^{252}$Cf calibration run (see text).
• Figure 5: Top: Cumulative spectra from Y/Be and Y/Al runs following the rise-time acceptance cut and normalization to the same exposure. Inset: signal acceptance (SA) for pulser calibration events and scintillation signals in CsI (see text). Bottom: Y/Be-Y/Al residual, corrected for SA. The excess due to neutron scattering is visible (see text). Error bars are statistical. Inset: energy resolution measurements and their fit (see text). Most error bars are encumbered by datapoints.