Errors in ellipsometry data fitting

Ellipsometry is a powerful technique for thin film structure and optical properties determination. However the interpretation of ellipsometric data requires an additional processing involving a representative model of the sample and an inversion strategy. In the standard fitting approach, unknown materials are represented by dispersion functions, and the difference between generated and experimental data is minimized by varying some model parameters. In this work we are interested in the errors on the fitted parameters and derived optical observables due to errors on the other fixed parameters and experimental data. The formulas for error propagation are defined; a strategy for reasonable choice of input error is given; and the use of these error estimates is proposed as a tool for critical analysis of model appropriateness.