David, Anna; (2024) Background Mitigation in the LUX-ZEPLIN Experiment Using Statistical Analysis of Waveforms. Doctoral thesis (Ph.D), UCL (University College London).

Text AnnaDavidThesisUnsigned.pdf - Other Access restricted to UCL open access staff until 1 March 2025. Download (21MB)

Abstract

The LUX-ZEPLIN (LZ) dark matter direct detection experiment searches for recoils of target xenon nuclei in a time projection chamber following an interaction with a Weakly Interacting Massive Particle (WIMP). LZ produced world-leading limits for spin-independent (SI) WIMP-nucleon interactions using data from its initial 60-liveday science run. During the expected 1000-liveday total runtime, LZ will explore new regions of electroweak parameter space, leading to a possible first discovery of WIMPs. As LZ acquires more data, accidental-coincidence backgrounds, resulting from random pairings of lone primary (S1) and secondary (S2) scintillation signals, become increasingly dominant. This thesis presents a statistical approach to target these events, utilising the full time dependence provided by S1 pulse waveforms. The detection times of scintillation photons can be used to identify events with either a non-physical time separation between the S1 and S2, or a spurious pulse shape resulting from the pile-up of single photoelectrons misclassified as an S1. Data-driven templates of S1 waveforms are used to perform likelihood ratio tests targeting these background pathologies, as well as a background-agnostic S1 goodness of fit test. These form the basis of three test statistics used as input features in a boosted decision tree, which classifies events as signal or background by taking all three criteria into consideration. This machine learning approach improves the projected sensitivity of LZ to 40 GeV-mass WIMPs by 8%. The discovery potential for a 40 GeV WIMP with an SI WIMP-nucleon cross-section of 3.99⨯10⁻⁴⁸ cm² is improved by 11%. The development of waveform-based cuts benefits from the use of accurate optical simulations, so this thesis additionally presents the optimisation of several material parameters, in order to better reproduce the light collection efficiency observed in data. Also described in this work is the development of data quality monitoring tools, which are essential to achieve a robust science result for an extremely rare signal.

Download activity - last month

Export as XMLCSVJSON

Download activity - last 12 months

Export as XMLCSVJSON

Downloads by country - last 12 months

Export as XMLJSONCSV

Archive Staff Only

View Item