Proper choice of the error function in modeling spectroellipsometric data

Publisher Website
Google Scholar
Abstract
The modeling procedure for analyzing the rotating analyzer spectroellipsometric data using the linear regression analysis technique is examined. The best set of optical parameters which should be employed in the modeling procedure is searched out with the help of variance–covariance matrices and the correlation properties of optical parameters. It was found that α and β, which are the Fourier coefficients of sinusoidal components of the intensity, constitute the best set for such an analysis for the case of single-crystal silicon and gold film evaporated on glass substrate. Also suitable expressions of α and β in terms of the ellipsometric parameters Δ and ψ have been derived to enable a linear regression analysis using α and β.