Detection of Discrete Surface and Subsurface Defects

Abstract

Smooth surface topography (finish) has been related, almost exactly, to the corresponding reflected scatter pattern under the condition that the sample is a smooth, clean, front surface reflector [1-3]. However, another extremely useful application of light scatter metrology is the detection and mapping of component defects that do not meet the smooth, clean, reflective conditions of many mirror surfaces. Examples of such defects are surface contaminants, deep scratches and digs, coating globs and residues, and subsurface defects. When detecting the presence of these defects by scatter measurement, the surface topography scatter is a limiting source of noise. Although smooth non-topographic defects often scatter more light than the surrounding surface topography, they may sometimes scatter considerably less light because they have a small cross-sectional area or because they are buried just beneath a reflective surface. In such cases, a low signal to noise ratio results. If it can be established that non-topographic defects scatter light differently than surface topography, then these differences can be exploited to improve signal to noise and map the defects using various raster techniques described in the literature. This paper discusses polarization differences in topographic and defect scatter and outlines techniques that have been used to enhance defect detection.