High-speed analysis of surface topography on semiconductor wafers by optical diffractive techniques.

Abstract

A system for high-speed analysis of optical diffraction (scatterometry) data of non-destructive critical dimension (CD) and topography measurements is reported. The approach includes two parts. In the first part (forward problem), diffraction from periodic structures on wafer surface is simulated accurately using an electromagnetic scattering model based on a surface integral equation (SEE) approach. In the second part (inverse problem), a nonlinear regression method is employed to characterize surface topography by iterations of calculating weighted square errors between simulations and measurements. This dissertation covers the development of the forward model, comparisons with other analytical techniques and SEM measurements, and simulated results of parameter extraction from periodic trapezoidal structures on the basis of the developed SEE model and nonlinear regression method.