Abstract

Accurate characterization of optical constants of metal films is often prevented by unwanted overlayers which may be due to oxidation, surface roughness, or other causes¹. These overlayers in general have different optical characteristics than the underlying metal film and thus can significantly alter the optical measurements (e.g. reflectance or ellipsometry) made on the sample. Ignoring these overlayers generally leads to inaccurate estimates of the actual optical constants of the metal. To avoid this problem, the overlayer must either be eliminated or included as part of the theoretical model. Unfortunately, inclusion of the overlayer in the model introduces more unknowns to be determined, and makes data analysis considerably more complicated. A more simple approach, which we employ, is to make the optical measurements through the (transparent) substrate. In this case, the radiation interacts only with the unexposed portion of the metal film (assuming the film is sufficiently thick) and hence surface oxide or roughness layers do not affect the optical measurements.

© 1988 Optical Society of America