Exploring the Inelastic Dark Matter frontier with the CRESST experiment

The quest for the nature of dark matter (DM) is one of the most fundamental in today’s physics research. Until today the existence of DM is inferred by indirect (gravitational) effects. The best interpretation to date of these observations is the existence of new massive particles with an interaction cross section at the (sub)weak scale. This explanation sets some strict requests for the DM candidate (cold, stable and interacting only gravitationally and sub-weak) but leaves a wide range of properties completely unconstrained. The most commonly searched dark matter candidate, the WIMP (Weak Interacting Massive Particle), is only one among all the possible candidates to solve the DM puzzle. One interesting possibility, explored in many theoretical frameworks, is that dark matter predominantly scatters inelastically off nuclei, causing the DM particle to up-scatter into an excited state. In the work of Bramante et al. [1] the inelastic scattering framework is widely investigated with a particular focus on the available kinematic phase space depending on the target nucleus. This article pointed out that, for a fixed dark matter mass, the heavier the target mass the larger mass splitting will be accessible. Among all the direct dark matter searches CRESST [2] (Cryogenic Rare Event Search with Superconducting Thermometer) is the most suitable for probing the inelastic dark matter (iDM) scenario, being tungsten the heaviest target element used in such experiments at present. CRESST is one of the leading experiments for a light direct dark matter search. Located at the Laboratori Nazionali del Gran Sasso (Abruzzo, Italy), CRESST target consists of arrays CaWO4 crystals operated at a temperature of few mK. For each crystal both the phonon and light signal produced by a particle interaction are detected, allowing particle identification as well as a precise energy measurement. In this PhD thesis the analysis of CRESST data in the framework of the inelastic dark matter scenario is presented. This work is focused on the combination of the results from multiple detector modules to exploit the total exposure of the experiment. Due to kinematic reasons, the iDM is characterised by a suppressed effective DM-nuclear scattering rate and a minimum recoil energy in the detector, corresponding to the minimum required energy for inelastic DM-nuclear collisions to happen. In light of these considerations, CRESST-II [3] phase 2 data have been chosen for this analysis instead of more recent data in order to increase the total exposure available. The target mass of a CRESST-II detector is more than 10 times larger than the one of CRESST-III detector leading to a gross exposure of ∼ 160 kg d for each detector module in the data taking campaign considered in this work. In the manuscript the raw data analysis procedures, which require in this case optimizing calibration and linearisation of the individual detector response for a non-standard region of interest, are described in detail. Particular attention is devoted to the explanation of the data quality selections performed on the background data to remove all the signals due to nonphysical processes. This procedure is not trivial because each detector has an individual behaviour and specific populations of events that need to be understood and associated with the physical/non-physical process that caused them. A blind analysis is performed using a small part of the data set (∼ 20%) as training data to define all the selections. These are then applied blindly to the full data set avoiding any unwanted bias. For all detector modules a dedicated simultaneous fit of the energy spectra measured by the phonon and light detector has been performed to optimise particle identification and thus improve background discrimination. The expected iDM spectrum in the CRESST modules has been produced and compared with the measured one for each module and the exclusion limit on the iDM cross section was obtained for each module. The background discrimination performances as well as the exclusion limit of each detector have been evaluated carefully and only the most suitable modules have been selected for the final combined analysis. Finally the results obtained with the chosen detectors have been combined. With the resulting enhanced exposure obtained from CRESST-II phase 2 data the final exclusion limit on the inelastic dark matter cross section is given.