Fully automated spectroscopic ellipsometry analyses: Application to MoOx thin films

Abstract

In spectroscopic ellipsometry, the optical properties of materials are obtained indirectly by generally assuming dielectric function and optical models. This ellipsometry analysis, which typically requires numerous model parameters, has essentially been performed by a trial-and-error approach, making this method as a rather time-consuming characterization technique. Here, we propose a fully automated spectroscopic ellipsometry analysis method, which can be applied to obtain dielectric functions of light absorbing materials in a full measured energy range without any prior knowledge of model parameters. The developed method consists of a multiple-step grid search and the following non-linear regression analysis. Specifically, in our approach, the analyzed spectral region is gradually expanded toward a higher energy, while incorporating an additional optical transition peak whenever the root mean square error of the fitting analysis exceeds a critical value. In particular, we have established a unique algorithm that could be employed for the ellipsometry analyses of different types of optical materials. The proposed scheme has been applied successfully for the analyses of MoOx transparent oxides and the complex dielectric function of a MoOx layer that exhibits dual optical transitions due to band-edge and deep-level absorptions has been determined. The developed method can drastically reduce a time necessary for an ellipsometry analysis, eliminating a serious drawback of a traditional spectroscopic ellipsometry analysis method.