Presentation Information
[25p-1BC-11]Prediction of optical transmittance for polymer-network liquid crystals from polarizing optical microscopy using pretrained machine learning models
〇Hiroshi Kakiuchida1, Kensuke Suzuki2, Takuto Kojima2 (1.AIST, 2.Nagoya Univ.)
Keywords:
polymer network liquid crystal,machine learning,feature extraction
The nonuniform distribution of refractive index between liquid crystal and polymer phases in polymer network liquid crystal (PNLC) causes light scattering. This nonuniformity can be observed using polarizing optical microscopy (POM). From many of the observed images, the degree of light scattering can be intuitively predicted, but there are also some images that are not, which is a problem when determining the fabrication conditions. In this research, we used machine learning to predict optical transmittance from the POM images and discovered a universal relationship between POM images and optical properties. Furthermore, based on this knowledge, we found that two fabrication parameters, crosslinkers and UV exposure temperature, are important to improve optical performance of the PNLCs.