Machine Learning Virtual SEM Metrology

Abstract

E-beam metrology, both CDSEM metrology and defect scan metrology, have been playing a very critical role in assessing post lithography or post etch patterning quality. SEM images can provide rich visual information for engineers to do qualitative and quantitative analysis. However, the lowe-beam metrology tool throughput makes it impossible to obtain SEM images for very large area. Monte Carlo based SEM image simulations are slow and they also require post lithography or post etch pattern 3D structures as prerequisite. To bridge the gap, we have proposed a Virtual SEM Metrology solution using physics based feature maps and the U-net neural network. With information in aerial image space encoded properly, SEM images of both post lithography and post etch can be predicted accurately enough for practical applications using our proposed Virtual SEM Metrology models.