Advanced process development and control based on a fully automated SEM with ADC

A novel system is presented which provides complete automation of the SEM review and classification process, allowing full integration of the SEM into the production line. This automation facilitates a paradigm shift in which the manual SEM and expert operator are replaced by an automatic, high-throughput SEM. This production SEM is utilized in conjunction with defect inspection tools to provide classified defect density, a key output for process development and line monitoring, in a quantitative form which can be directly transferred to the wafers-in-process (WIP) and yield management (YM) systems. To achieve full automation of SEM review, two broad categories of operator activities had to be replaced: (1) Automatic Defect Review (ADR) replaces the sequences related to the acquisition of defect images, from wafer alignment, through setting of optimal imaging parameters, to redetection of defects. (2) Automatic Defect Classification (ADC) replaces the decision process of assigning a pre-defined class to each of the defects. A system was developed which relies on a unique set of defect images, generated simultaneously, to achieve high levels of performance in both ADR and ADC. The information available from these images is directly related to the physical properties of defects, and used to derive a set of invariant classes, which we call Core Classes. The imaging features of the system as well as the architecture of the ADC subsystem, are discussed. The system was utilized in an advanced fab for process development and control, and results are presented of both ADR and ADC on multiple layers.