Abstract
There is plenty of evidence from various astrophysical observations suggesting that most of the matter in the universe comes in the form of dark matter. Yet the underlying nature of this dark matter remains one of the important unsolved puzzles in physics today. One hypothesis is the existence of weakly interacting massive particles (WIMPs). The CRESST experiment aims at the detection of such particles. It employs cryogenic particle detectors to search for nuclear recoils induced by the elastic scattering of those dark matter particles off the nuclei composing the detectors. These detectors are able to detect very low recoil energies ( 1 keV), which makes them especially suited to search for light dark matter particles ( 5 GeV/c 2 ) not accessible to other experiments with higher energy thresholds. In this work, data taken with the CRESST experiment between July 2013 and August 2015 are analyzed, focusing in particular on light dark matter particles. This requires novel methods to reject backgrounds at low energies. One such method which identifies background events based on their pulse shape using machine learning techniques is presented. As a result, limits could be set on the spin-independent dark matter-nucleon cross section extending the parameter space covered by direct detection searches to masses below 1 GeV/c 2 . Additionally, momentum-dependent asymmetric dark matter models could be constrained. Finally, the influence of various detector parameters on the sensitivity of future upgrades to the CRESST experiment is investigated.
Author and committee
dc:creator, dc:contributor.*- Author
-
- Strandhagen, Christian
Identifiers
dc:identifier.*- Identifier
- hdl:10900/74055