Image Analysis: Converting from Manual to Automated Metrology

Abstract

Many of the repetitive measurements required in the analysis of images in the semiconductor field can be replaced by partially or fully automated measurements. Automated measurements provide a number of benefits, including speed, repeatability, and removal of operator bias. The question of accuracy and precision in metrology applications is important to understand.Recently we implemented a tool for automated analysis of resist bar images that was designed for use in acceptance testing-of lithographic stepper machines. The steppers are required to meet a defined set of criteria over the entire area of the wafer and through the focus range of the lens. Prior to the development of this package, the measurements were all done with a ruler and protractor on film.The software identifies the boundary lines of the resist bar by locating a series of points along each edge and then fitting a line to those points. Edge identification is complicated by coating artifacts and by the variety in the images being analyzed. After identifying edge points along the base, side, and top of the resist bar, the software automatically generates information about the width, height, side wall angles, and various other measurements of the resist bar and stores the data in spread sheet format.