Deutsche Neutronenstreutagung

Focusing on the prominent technique of neutron diffraction for phase identification, the principal goal is to autonomously identify the presence of different crystalline phases such as: Al2O3, LiAlH4, TiO2, ZnS, etc., efficiently and without being dependent on a reference database. Although deep learning approaches require high amount of training data, they only use the learned weights and biases during inference which makes them rapid and robust solutions. Being known for their high throughput rates, they were employed in this study as a path to efficiently recognize phases of a neutron diffraction pattern without rule-based methods.