Applications of Artificial Neural Networks to Proton and Electron Impact Ionization of Biologically Relevant Atoms and Molecules

Abstract

We use Artificial Neural Networks (ANNs) to study proton and electron impact single ionization of biologically relevant atoms and molecules. In these processes, an incident proton or electron collides with a target atom or molecule causing a single electron to be ionized from the target. The process of ionization is biologically relevant for two reasons. Firstly, ionization results in reactive ions that are able to take part in damaging chemical reactions. Secondly, free electrons have been shown to result in DNA strand breaks, which can lead to cell death. While atomic and molecular collisions have been studied experimentally for decades, most of the available data is for noble gas atoms and small diatomic molecules. ANNs are computer programs that are trained to learn patterns in data and make predictions for cases where no data exists. They have been widely used in other fields, but this is one of the first applications of ANNs to atomic and molecular collision processes. In this work, we use the available experimental data to train the ANN, and then make predictions for biologically important target atoms and molecules where no data currently exists.