[1I08-12] Neutron Beam

[1I10] A spatial distribution analysis of crystalline microstructure of solidified Lead-Bismuth eutectic by wavelength-resolved neutron imaging with machine learning

*Ayaka Ohashi¹, Hirotaka Sato¹, Takashi Kamiyama¹, Naoya Odaira², Daisuke Ito², Yasushi Saito² (1. Hokkaido Univ., 2. Kyoto Univ.)

Keywords:Wavelength-resolved neutron imaging, Lead-Bismuth eutectic, Machine learning, Bragg edge, Crystalline microstructure

A wavelength-resolved neutron imaging enables us to analyze spatially-dependent crystalline microstructure in a bulk material over a large field of view. With this method, the spatial distribution of solidified lead-bismuth eutectic (LBE) was analyzed. However, it was difficult for the conventional methods to apply due to large statistical errors and crystalline microstructure changing for each position. In this study, we tried to identify spectral features, reflecting microstructures, at each pixel, and classified each feature with unsupervised machine learning. As a result, we could classify the spectra of solidified LBE into three types. The visualization of their distributions and analysis of the classified spectra clarified the relationship between these results and solidification speed.