xy Position Reconstruction in DarkSide-50

Abstract

The DarkSide-50 experiment seeks to directly detect dark matter in a liquid argon time projection chamber. In this dissertation, I present an algorithm of my design that determines the position of particle interactions with the liquid argon. This position reconstruction algorithm will be used by DarkSide-50 to reject backgrounds, particularly backgrounds from radioactive elements on the detector surface.

The position reconstruction algorithm functions by constructing light response functions (LRFs) that map locations in the detector to the expected distribution of signal in DarkSide-50's 38 photomultiplier tubes. Accurate LRFs cannot be produced by simulations of DarkSide-50's optics because such simulations are known to be flawed. Instead, this algorithm constructs LRFs using an iterative process driven by data. Initial, flawed LRFs are produced using simulated events but then used to produce new LRFs from data events. Multiple generations of LRFs are created from data with each generation driven to better satisfy a known feature of the detector: the dominant argon-39 background is uniformly distributed.

I also discuss a method of discriminating against surface background as an alternative to the common approach of fiducialization. This method considers the difference in goodness-of-fit between the best-fit reconstructed position and the best-fit position at the detector's surface.

I conclude by presenting results on the performance and validity of this algorithm, including some discussion of reconstruction errors.