Determining the bubble nucleation efficiency of low-energy nuclear recoils in superheated C$_3$F$_8$ dark matter detectors

TL;DR

This paper measures the efficiency of bubble formation by low-energy nuclear recoils in superheated C$_3$F$_8$, crucial for dark matter detection, using a flexible model and MCMC analysis to improve accuracy.

Contribution

It introduces a generalized piecewise linear model with systematic error handling for nuclear recoil efficiency in superheated liquids, applied to PICO detector calibration data.

Findings

Fluorine recoil efficiency is ≥50% at 3.3 keV (2.45 keV threshold).

Carbon recoil efficiency is ≥50% at 10.6 keV (2.45 keV threshold).

Model fit is statistically compatible with calibration data.

Abstract

The bubble nucleation efficiency of low-energy nuclear recoils in superheated liquids plays a crucial role in interpreting results from direct searches for weakly interacting massive particle (WIMP) dark matter. The PICO Collaboration presents the results of the efficiencies for bubble nucleation from carbon and fluorine recoils in superheated C$_3$F$_8$ from calibration data taken with 5 distinct neutron spectra at various thermodynamic thresholds ranging from 2.1 keV to 3.9 keV. Instead of assuming any particular functional forms for the nuclear recoil efficiency, a generalized piecewise linear model is proposed with systematic errors included as nuisance parameters to minimize model-introduced uncertainties. A Markov-Chain Monte-Carlo (MCMC) routine is applied to sample the nuclear recoil efficiency for fluorine and carbon at 2.45 keV and 3.29 keV thermodynamic thresholds simultaneously. The nucleation efficiency for fluorine was found to be $\geq 50\, \%$ for nuclear recoils of 3.3 keV (3.7 keV) at a thermodynamic Seitz threshold of 2.45 keV (3.29 keV), and for carbon the efficiency was found to be $\geq 50\, \%$ for recoils of 10.6 keV (11.1 keV) at a threshold of 2.45 keV (3.29 keV). Simulated data sets are used to calculate a p-value for the fit, confirming that the model used is compatible with the data. The fit paradigm is also assessed for potential systematic biases, which although small, are corrected for. Additional steps are performed to calculate the expected interaction rates of WIMPs in the PICO-60 detector, a requirement for calculating WIMP exclusion limits.