Towards next-generation cryogenic dark matter searches with superconducting thermometers

Abstract

Today we observe overwhelming gravitational evidence for the existence of dark matter in the universe and its non-null abundance along the solarcircle. Experiments using cryogenic calorimeters spearhead the effort to measure the scattering of sub-GeV/c2 dark matter particles with nuclei of a detector target. Among them, the CRESST experiment achieves the strongest sensitivity for spin-independent, elastic scattering scenarios. Itsspin-off, the COSINUS experiment, is set up to validate the long-standing dark matter discovery claim of the DAMA experiment, exploiting low energy thresholds and particle identification. These searches rely on careful detector design, measurement setup, and data analysis to provide insightsin to the nature of dark matter. We review the motivations for the searches,the operation principles, and study detector design choices in detail by constructing a dedicated detector response simulation. The analysis techniques are summarized and implemented in a modern software toolbox,allowing the execution of all established methods and the incorporation of machine learning classifiers. Building upon these methods, we summarize the characterization of two detector modules from the latest CRESST-III measurement campaign, which lead to the currently most stringent limitson spin-dependent sub-GeV/c2 dark matter-nucleus elastic scattering.Furthermore, we study detector prototypes for COSINUS and estimate achievable energy thresholds of detector designs for the first physics runs.We describe the technological challenges for a large-scale measurement setup and analysis process and study methods based on deep and reinforcement learning with CRESST-III data to automate the required manual interventions. These methods equip the next generation of cryogenic darkmatter searches with improved sensitivities and higher collected exposure.