Optimizing low-mass WIMP searches in the LUX-ZEPLIN experiment

Thesis

The nature of dark matter is an ongoing mystery in the fields of both astrophysics and particle physics. Evidence over a wide range of length scales indicates that dark matter accounts for 84% of the matter density of the universe. However, detection has remained elusive.

The LUX-ZEPLIN (LZ) experiment is a dual-phase LXe time projection chamber designed for the direct detection of dark matter. In its first science run, LZ achieved sensitivity to spin-independent WIMP-nucleon cross sections of 9.2 × 10⁻⁴⁸ cm² for a WIMP of mass 36 GeV/c² at the 90% confidence level. While LZ is designed for the detection of high-mass WIMPs, there is also ample motivation to maximise LZ’s sensitivity to dark matter of lower masses.

The contributions of this thesis to dark matter searches improve the sensitivity of LZ to low-mass dark matter candidates: both in standard analysis, where sensitivity is limited by low-energy backgrounds and signal efficiency, as well as through the use of electroluminescence-only (S2-Only) data, where this signal efficiency is maximised.

This thesis introduces a novel method of rejecting accidental coincidence events, the lowest-energy background present in LZ data. It is possible in LZ to reject 95% of these events while preserving >99% of signal, allowing for sensitivity to both low and high-mass WIMPs.

The first measurement of S2 pulse mis-classification in LZ is also presented. This work has been used to achieve maximum signal detection efficiency at low energies. The probability of mis-classification of any electroluminescence pulse, down to detection threshold of 2 e, is below 5%. These measurements are then incorporated into statistical analysis, in order to minimise uncertainty on measurements of signal sensitivity at lower masses.

The thesis concludes with a demonstration of how the capability of LZ to reject low-energy backgrounds and accept low-energy signals can be leveraged in an analysis of electroluminescence-only (S2-Only) data in LZ, a dataset sensitive to dark matter of masses as low as 80 MeV. A Boosted Decision Tree (BDT) trained to reject grid radiogenic events in S2-Only data is presented. It is shown to be possible to reject 80% of grid background events with a signal efficiency of 80%, a significant improvement on similar results from LUX which improves the sensitivity of LZ to low-mass WIMPs by up to a factor of ten.

Authors: Hunt, D

• DOI: 10.5287/ora-br5x55pxn
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: dark matter; xenon; WIMPs; Time Projection Chambers
• Subjects: Dark matter (Astronomy)