INVESTIGATION OF GOI TEST METHODS ON SILICON SURFACE DEFECT FAILURE MECHANISMS

摘要

Traditionally, both silicon wafer suppliers and IC companies have used phosphorous-doped polysilicon MOSCAP gates to monitor the quality of silicon with gate oxide integrity (GOI). However, the test method, usually a ramped voltage test to determine breakdown voltage (BV), has had limited sensitivity to silicon defects, especially for the ultra-thin gate oxides regime. Alternative GOI methods based on the non-contact, corona-based, soft BV have been developed. However, the soft BV method has a low sensitivity to surface silicon imperfections due to the limited maximum corona charge up to the Fowler-Nordheim tunneling current level. The intent of this paper is to compare the poly gate method with the corona-based method for crystallographic defects, mechanical-induced damage, and metal contamination using various oxide thickness films. Although the results show that corona-based, soft-BV measurement is optimal for monitoring metal contamination, they also show that it is relatively insensitive for monitoring crystallographic defects (COPs) and mechanical polishing silicon defects (MPSDs). The MPSDs may not be readily distinguishable from the COPs using standard inspection tools. The MPSDs behave similarly to the COPs in that the breakdown yield decreases as the gate oxide becomes thicker (greater than 75A).