A Pattern Recognition Tool for Automatic Etch Process Quality Check

Abstract

This paper presents an automatic tool for image analysis applied to semiconductor in-line inspections. The goal of this tool is to understand automatically which contacts in the wafer have been not completely etched during the plasma process and an etch step. For this reason we developed an algorithm able to classify the observed structures in scanning electron microscope (SEM) inspections. Such recognition provides a quick and detailed analysis of the under-etched contacts, which is based on fixed rules and thus objective. Automation controls, applied to the tool results, allow for complete monitoring of etch chambers and of the SEM images quality as well. Applying this tool should result in cost saving and yield increase.