Threshold method

SEM analysis using threshold method

One commonly-used method of SEM image analysis relies on establishing a threshold of brightness for the top and bottom of an image. This threshold often represents the best guess of the operator, or it can be standardized from a sample with known verticle profiles. However, even with the standardized sample, this method is inaccurate.

Let us examine the figure on the right as an example. Here, the gray wall represents a known crossection, which is used to establish the standard for the threshold analysis. Clearly, the signal level determined here works great for analyzing the standard. However, using the same threshold to analyze a different sample (green) presents considerable error due to a small change in the side wall angle.

To further complicate the situation, the offset does not depend solely on the sidewall angle. Many other factors come into play, including the material, beam size, and acceleration voltage. Therefore, not only is the signal threshold model inaccurate, but the offset is different on a case-by-case basis.

It is clear that the physics of an SEM must be taken into account in metrology. MyCD is the only commerically available, physics-based software to perform this analysis with accurate CD extraction.