Natural physical phenomena occurring at length scales of a few nm in EUV lithography give rise to variation in photoresist images: edge, width, and top roughness, feature-to-feature CD or shape variability, edge placement errors, etc. The most damaging are stochastic printing failures caused by undesirable film thickness loss, admitting etch in line regions, or film thickness gain, preventing etch in space regions. In this work, we begin from analysis of well-calibrated rigorous physical stochastic EUV lithography models to study nanoscale exposure effects affecting stochastic failures. We apply acceleration to the stochastic model and perform computational inspection and classification of hot spots on a large layout area. The agreement between predicted probabilities of occurrence and observed defect frequencies are given for both line and space hot spots. We then perform computational inspection upon a virtual process and select hot spot locations and affect repairs. The actual mask is then fabricated, real wafers are exposed, processed, inspected and measured to compare the predicted reductions in defect probabilities with actual measured defect frequencies on wafer.
|