Scanning-electron-microscope image processing for accurate analysis of line-edge and line-width roughness

The control of line-edge or line-width roughness (LER/LWR) is a challenge especially for future devices that are fabricated using extreme-ultraviolet lithography. Accurate analysis of the LER/LWR plays an essential role in this challenge and requires the noise involved in scanning-electron-microscope (SEM) images to be reduced by appropriate image processing prior to analyses. In order to achieve this, the authors simulated SEM images using the Monte-Carlo method and detected line edges in experimental and these theoretical images after noise filtering using new image-analysis software. The validity of these simulation and software was confirmed by a good agreement between the experimental and theoretical results. In the case when the image pixels aligned perpendicular (crosswise) to line edges were averaged, the variance var(φ) that was additionally induced by the image noise decreased with the number N _PIX,X of averaged pixels but turned to increase for relatively large N _PIX,X's. Real LER/LWR, however, remained unaffected. On the other hand, averaging image pixels aligned parallel (longitudinal) to line edges not only reduced var(φ) but smoothed the real LER/LWR. As a result, the nominal variance of real LWR, obtained using simple arithmetic, monotonically decreased with the number N _PIX, L of averaged pixels. Artifactual oscillations were additionally observed in power spectral densities. var(φ) in this case decreased in an inverse proportion to the square root of N _PIX,L according to the statistical mechanism clarified here. In this way, image processing has a marked effect on the LER/LWR analysis and needs to be much more cared and appropriately applied. All the aforementioned results not only constitute a solid basis of but improve previous empirical instructions for accurate analyses. The most important instruction is to avoid the longitudinal averaging and to crosswise average an optimized number of image pixels consulting the equation derived in this study.