Shape-sensitive linewidth measurement passes its first test - News Briefs

Journal of Research of the National Institute of Standards and Technology, Jan, 2002

With support from International SEMATECH, NIST researchers have been developing a method for obtaining edge shape as well as line width information from images of silicon features (e.g., lines, transistor gates) in integrated circuits. The problem addressed by the new method is particularly acute when nanometer scale dimensional uncertainties are required: a scanning electron microscopy (SEM) image is a distorted representation of the feature. It is a two-dimensional intensity pattern derived from the interaction of an electron beam with a three-dimensional object. How does one deduce the shape and size of the object giving rise to the observed image?

The method works by comparing the measured image of a feature to a library of calculated images for similar features, each differing slightly in size or shape. The images in the library correspond to a range of possibilities and may be interpolated for better resolution. The closest match is deemed to be the shape of the feature that produced the measured image. This is a plausible scheme, but does it really work?

To find out, a comparison was performed this month between the shape of a line determined from a top-down SEM image using the above method with an SEM image of the same line after cross sectioning. In this first test of the shape-sensitive linewidth measurement system, the two measurements agreed to within better than 1 nm on the line's width (174.5 nm). This is probably fortuitously close, inasmuch as both measurements have uncertainties of a few nanometers. They agreed to within a few tenths of a degree on the sidewall angles. Similar agreement was achieved in a comparison on a second line. Still under investigation, there was disagreement about the roundness of the upper corners, an aspect of feature shape that shows up in the image in only a subtle way.

CONTACT: John Villarrubia, (301) 975-3958; john. villarrubia@nist.gov.