Erlangen 2026 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

T: Fachverband Teilchenphysik

T 7: Data, AI, Computing, Electronics I

T 7.8: Talk

Monday, March 16, 2026, 18:00–18:15, KH 00.024

Machine learning based Particle Identification in a Diffusion Cloud Chamber. — •Benjamin Rosendahl, Jasper von Lepel, Mario Schwarz, and Stefan Schönert — Department of Physics, TUM School of Natural Sciences, Technical University of Munich, 85748 Garching b. München, Germany

We present a work on a diffusion cloud chamber, which establishes quasi-stationary alcohol vapor diffusion through air by a temperature gradient, generating a supersaturated layer (S) in which ions seed droplet formation. By measuring the vertical temperature profile T(y) and applying heat and diffusion equations as well as nucleation theory, we characterize local supersaturation and growth kinetics during vapor condensation on ions. Together with a camera setup, a U-Net based machine learning pipeline performs segmentation of particle tracks, followed by skeletonization and vectorisation. The live time setup enables joint thermodynamic and particle physics analyses: Mapping T(y) and S(y), quantifying fluxes of charged particles by both cosmic and ambient radiation (e.g., muons, betas and alphas) calibrated by comparison to radioactive sources and identifying phenomena such as radon decay. Beyond measurement, the instrument functions as an exhibit with high educational value, translating otherwise abstract concepts into observable events. We acknowledge support from the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany’s Excellence Strategy - EXC 2094 - 390783311 and through the Sonderforschungsbereich (Collaborative Research Center) SFB 1258 ’Neutrinos and Dark Matter in Astro- and Particle Physics’.

Keywords: Diffusion Cloud Chamber; Supersaturated alcohol vapor; U-Net based machine learning; Particle Identification