CONTRIBUTION OF THE HIERARCHICAL GENETIC ALGORITHM TO ELLIPSOMETRIC SPECTROSCOPIC CHARACTERIZATION

In spectroscopic ellipsometric characterization, different dispersion models are generally used to accurately determine the opto-geometric parameters of a Sample. However, all current methods use a single dispersion law. The appropriate dispersion law is chosen on the experience basis. This makes conventional methods less flexible and potentially leads to erroneous results. In this paper, we develop an algorithm capable of characterizing the sample without a priori knowledge of a dispersion law describing the refractive indexes Hierarchical Genetic Algorithm (HGA) automatically selects the best law from a set of models on the basis of the Root Mean Square Error (RMSE) between the simulation with the chosen law and the measurement. The algorithm was applied on 2 silicon samples with a layer of silicon dioxide and a layer of resist. The HGA results are more robust than the conventional optimization method.