A new technique for automated wafer inspection and classification of particles and crystalline defects

Abstract

The ability to detect and quantify contaminants and crystalline defects on the surfaces of silicon wafers is extremely important for IC device yield enhancement. Although automated particle inspection sensitivities exceed 0.1 /spl mu/m, visual inspection continues to be used in silicon wafer manufacturing facilities because conventional particle scanners are not capable of identifying and quantifying a variety of material defects such as Epi spikes, ESF, and pits. This work investigates a new methodology to image and classify these material defects by combining information from two independent phenomena, light scattering and reflecting. The optical system to study this unique method consists of a conventional particle scanner to detect and quantify light scattering events from contaminants on the wafer surface and a Reconvergent Specular Detection (RSD) apparatus. RSD, more commonly known as light channel detection (LC), is capable of imaging material defects by measuring attenuation of the light beam intensity reflected from the wafer surface due to diffraction, absorption, and distortion. Epi spikes, mounds, voids, dislocations, slurry burns and some other common defect features and contamination on silicon wafers are studied using this equipment. The results are compared with and confirmed by that of microscope and AFM. This work presents the results showing that a conventional wafer scanner coupled with a RSD light channel apparatus provides the ability to successfully identify and quantify crystalline material defects and distinguish them from large particulate. It provides the solution to the wafer manufacturing industry for fully automated wafer inspection and defect classification and sorting.