Design of an optical linear discriminant filter for classification of subwavelength concave and convex defects on dielectric substrates.

Abstract

Identification of the defect type on substrate materials is essential for enhancing their yield ratio. We propose a novel optical filter to discriminate between subwavelength-order concave and convex defects on flat surfaces. This filter performs Fisher's linear discriminant analysis using light wave diffraction. The defect type is discriminated by simply comparing the irradiance at an observation point with the threshold value. Neither the defect image nor phase data, nor a large amount of data processing by a computer, is necessary. Numerical discrimination simulations indicate a discrimination error of 0.85%, and the noise tolerance of the proposed system is also discussed.