Simulation of ionizing radiation in cell phone camera image sensors

Abstract

The Distributed Electronic Cosmic-ray Observatory (DECO) is a cell phone app that uses a cell phone camera image sensor to detect cosmic-ray particles and particles from radioactive decay. Images recorded by DECO are classified by a convolutional neural network (CNN) according to their morphology. In this project, we develop a GEANT4-derived simulation of particle interactions inside the CMOS sensor using the Allpix$^2$ modular framework. We simulate muons, electrons, and photons with energy range 10 keV to 100 GeV, and their deposited energy agrees well with expectations. Simulated events are recorded and processed in a similar way as data images taken by DECO, and the result shows both similar image morphology with data events and good quantitative data-Monte Carlo agreement.