Metal-Core-Specific\nScreening with Machine Learning:\nAccelerating the Discovery of Metal Oxide Clusters for Enhanced EUV\nLithography Resolution

Obtaining effective extreme ultraviolet lithography\n(EUVL) materials\nfor pragmatic applications remains challenging. The experimental design\nand conventional theoretical prediction are time-consuming and costly\nand hardly affordable to accelerate the discovery of commercial EUVL\nmaterials. In this work, we employed the machine learning (ML) technique\nto predict the ionization potential of promising EUVL materials, which\nis closely related to the photoresists’ solubility switch.\nThe developed ML model presents a strong generalization ability and\ncan predict new EUVL materials containing different metals (i.e.,\nMg, Cu, Ca, Cd, Ni). Furthermore, feature analysis indicates that\nthe number of hydrogen bond donors in a compound plays a vital role\nin determining the ionization potential of EUVL materials. The work\nprovides not only an effective ML model to predict EUVL materials\nbut also crucial insights into the correlation between the structure\nand properties. Finally, the developed ML model has been integrated\ninto an online platform (https://zinc-oxo-cluster-predictor.streamlit.app/), allowing users to quickly evaluate their designed materials and\ndevelop a comprehensive scheme for discovering promising EUVL compounds\nbased on our platform.