First dark matter search results from a 4-kg CF$_3$I bubble chamber operated in a deep underground site

TL;DR

This paper reports the first dark matter search results from a 4.0 kg CF3I bubble chamber operated at SNOLAB, employing acoustic discrimination to suppress alpha backgrounds and three Seitz-based bubble-nucleation thresholds. The dataset yields 20 single-recoil candidates and 3 multi-bubble events in 553 kg-days, with a background expectation of a few events, leading to constraints that set world-leading limits on WIMP-proton spin-dependent scattering for WIMP masses above 20 GeV and competitive spin-independent sensitivity. A significant portion of the analysis focuses on understanding bubble nucleation efficiencies for light recoils and modeling internal neutron backgrounds, which leads to a limit band rather than a single curve. The results demonstrate the viability of bubble-chamber detectors for dark matter searches and highlight the need for reduced internal neutron backgrounds and increased exposure in future runs.

Abstract

New data are reported from the operation of a 4.0 kg CF$_{3}$I bubble chamber in the 6800-foot-deep SNOLAB underground laboratory. The effectiveness of ultrasound analysis in discriminating alpha-decay background events from single nuclear recoils has been confirmed, with a lower bound of $>$99.3% rejection of alpha-decay events. Twenty single nuclear recoil event candidates and three multiple bubble events were observed during a total exposure of 553 kg-days distributed over three different bubble nucleation thresholds. The effective exposure for single bubble recoil-like events was 437.4 kg-days. A neutron background internal to the apparatus, of known origin, is estimated to account for five single nuclear recoil events and is consistent with the observed rate of multiple bubble events. This observation provides world best direct detection constraints on WIMP-proton spin-dependent scattering for WIMP masses $>$20 GeV/c$^{2}$ and demonstrates significant sensitivity for spin-independent interactions.

Figures (7)

• Figure 1: (Color Online) Pressure history from a sample event at 33.5$^\circ$C. Time scale is linear within each region. The event is divided into four regions: (a) Chamber expands to the superheated state, (b) Pressure regulation turns on at elapsed time of 5 sec and the chamber stabilizes by elapsed time 30 sec, (c) Chamber is live (accumulating dark matter data) from 30 sec until a trigger or timeout at elapsed time of 500 sec, (d) Chamber compresses and sits compressed for 30 sec between events, or 300 sec every tenth event. The mean expansion times at 39.0$^\circ$C, 36.2$^\circ$C, and 33.5$^\circ$C are 326, 396, and 417 sec, respectively. The shorter mean times at higher temperatures are due to an increased trigger rate during the expansion and stabilization periods. The majority of events at all temperatures end with a timeout.
• Figure 2: (Color online) Data from a $553$ kg-day WIMP search, shown as a distribution in ln($AP$) in red. Twenty single nuclear recoil events candidates and 2474 alpha events were observed. The blue histogram shows the identical analysis for data taken in the presence of an AmBe neutron source. We define an acoustic cut of $0.7 < AP < 1.3$ to select nuclear recoils with an acceptance of $95.8\%$ as determined by the AmBe calibration.
• Figure 3: (Color online) Distribution of time differences between consecutive alpha-decay events. The solid curve is a fit to a simulated time difference distribution, including all live time effects and acceptance cuts, based on a component arising from decay of $^{222}$Rn and daughters and a second component arising from random alpha decays with no parent-daughter time correlations. The best fit is for a radon fraction of $0.95\pm0.05$. For comparison, the dashed gray curve shows the expected time difference distribution for uncorrelated alpha decays. The dip in rate around a $\Delta$t of 9 min is caused operationally by the forced compression of the chamber after a maximum expansion time of 500 sec.
• Figure 4: (Color online) The upper graph is the alpha-decay plateau curve for single bubble events, showing rate as a function of Seitz model bubble nucleation threshold obtained by varying the expansion pressure. The superimposed green curve shows the anticipated onset of sensitivity for $^{214}$Po, $^{218}$Po, and $^{222}$Rn recoils. The lower graph shows the comparable plateau curve for pairs of alpha-decay events separated in time by less than 500 sec. The superimposed green curve illustrates the sharper onset of sensitivity expected from the $101$ keV $^{222}$Rn recoils selected by the timing cut. In addition to the low statistics pressure scan data, the high exposure WIMP search data are also included on the plots (the points at low threshold with small error bars).
• Figure 5: (Color online) The observed count rates at the three thresholds are for one, two, three, and four bubble events induced by an AmBe neutron source. The superimposed curves represent the MCNP predictions for the bare Seitz model (black) compared to the best fit flat and exponential bubble nucleation efficiency models, with $\eta_{C,F} = 0.49$ and $\alpha_{C,F} = 0.15$ respectively. The bare Seitz model clearly over-predicts the number of observed counts, especially at high multiplicities, and these data do not distinguish between the flat and exponential efficiency models.