Exploiting machine learning clustering methods to optimize OPC model

The first and most crucial step in generating an effective term-based OPC model is defining an appropriate model structure. However, due to the modern OPC model's complexity and layout pattern diversity, optimizing a model inevitably involves a laborious iterative manual term selection. Therefore, we propose a new machine learning (ML)-based feature selection approach to facilitate the term-selection process more systematically and intelligently. Our approach utilizes the high-performance supervised clustering algorithms implemented in Calibre SONR to effectively reduce dimensionality by removing irrelevant and duplicate kernels. From the SEM measurement data, labeled data is obtained for SONR’s supervised clustering. From the generic kernel library containing hundreds for preconfigured kernels, only the terms with meaningful contribution are selected by calculating the term contributions projected on points of interest extracted from critical dimension (CD) gauges and edge placement error (EPE) gauges. The effectiveness of our approach will be presented through benchmarking against our best-known method (BKM) model generated by a manual term selection.