Some Deviations Associated With Vector Perturbation Diffraction Theory

Abstract

Last year at this conference our lab presented some data which strongly supports the use of the vector perturbation relationship between light scattered from smooth surfaces and the surface power spectral density (PSD). Last year's data showed that a consistent answer was obtained for the PSD for measurements taken with S and P polarization, with incident angles up to 45', for positive and negative sweeps, and for one and two dimensional samples. If in fact the theoretical relationship is correct then this must be the case, as the PSD is determined by a combination of surface topography and the scattering situation (geometry, polarization, wavelength, etc.). It appears, however that for large scatter angles (>700) and for large angles of incidence (>60A‚°) there is some deviation in the calculated PSD. In addition, the high angle scatter region is an area where a scatterometer with a semicircular detector sweep loses the one-to-one relationship between detector position and surface spatial frequency. Or, in other words, light diffracted to high scatter angles from single spatial frequencies appears over a small band on the observation semicircle instead of a diffraction limited point. Fata supporting these two deviations is presented. Although they are not regarded as serious violations of the vector perturbation theory they do impose a limit on the useful range over which the PSD may be calculated.