Analysis of Coincidence Losses for a Monitor of Particle Contamination on Surfaces

Abstract

A widely used instrument (manufactured by Inspex, Incorporated) for detecting, sizing, and counting particles on silicon wafer surfaces determines particle size by the number of pixels on a vidicon-type screen that are activated by light scattered from particles during oblique observation of perpendicular illumination. The particle counting ability of this surface monitor is affected by the number of particles present, their sizes, the sensitivity of the tool, and the system optical and electrical noise. In this system, the overlapping of the responses of two or more particles causes them to be counted as a single particle of a larger size. We present a method for estimating the fraction of the particles that are not involved in coincidence. The full solution for the fraction of particles in any size category overlapped by those in any other category is presented. Two limits are developed: one for particles all the same size (that is, each giving the same response) and one for particles of very different sizes, the smallest being much more numerous. Practical situations are expected to fall between these two limits. It is shown that for only 5% of the pixels of the surface monitor’s screen activated, between 5% and 18% (on average) of the particles would not register as singlets, due to coincidence. For a given total number of pixels activated, coincidence errors are greatest where the particles each give the same size response. © 1987, The Electrochemical Society, Inc. All rights reserved.