Hyung Wook Park, Ulsan National Institute of Science and Technology, Ulsan, Republic of Korea
This study conducted the experiments of Electron Beam (EB) drilling on thin ceramic material for semiconductor which has high brittleness and fragility. Process parameters for energy density of electron beam were adopted, and experimental conditions and phenomena obtaining the high-aspect-ratio microholes were investigated. On the other hand, due to the inherent properties of the thin ceramic material, microholes have a very high difficulty in morphological analysis. We have devised a conditional Generative Adversarial Network (c-GAN) model that can easily predict the profile of thin ceramic microholes. With a limited number of training data, c-GAN was successfully trained and tested. Furthermore, generative images from the proposed model were evaluated for the real hole dimensions. It was also optimized to predict the cross-sectional images of microholes close to the SEM image by simply providing the hole inlet image. The proposed model has high potential to be utilized as an AI model for non-destructive analysis of microstructures.