We describe the development of machine learning methods for the prediction of optical properties of halide scintillators. Scintillating materials are of interest in medical imaging, national security, and space detection schemes where they are used in detecting radiation. The materials studied here include A2BX6 halide scintillators, where A is a monovalent ion such as Cs, K, Li, Rb, and, Na; B is a tetravalent ion such as Ti, Zr, Hf, Sn, Se, and Te; and X is a halide ion such as Cl, Br, and I. A2BX6 materials such as Cs2HfCl6 are the subjects of a concerted effort in the search for new materials that can be incorporated into efficient, bright, cost-effective radiation detection devices. These are investigated based on their dielectric, ionic, and x-ray attenuation properties. Several machine learning models are used to utilize elemental data from the literature and are trained to learn the composition-property relationships of 550+ known scintillators compiled in the Lawrence Berkeley Scintillator Database. The trained models then use compositional correlations to predict light yield of various scintillators. The results of the model show that machine learning models can produce accurate prediction of the properties of known scintillators and can be used to predict the optical response of novel scintillators for use in radiation detection applications.

Prediction of scintillator light yield from composition using machine learning methods

255th National Meeting (2018)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.