A Robust Approach to Reliability Hotspot Detection

Abstract:
In addition to electromigration, stress migration is developing into a critical reliability risk in modern semiconductor copper metallization. In order to guarantee reliability of critical products mainly for automotive and industrial use we are currently investigating possibilities to identify critical configurations (hotspots) in layout that are sensitive to excessive stress- and electromigration. Single vias are usually the weakest elements of a circuit, and their sensitivity depends strongly on the shape of the connected metal lines. For electromigration a robust rule-based methodology has been developed to detect the most critical vias. For stress migration, however, we assume that a simple rule-based approach is not sufficient to safely detect all critical configurations with a near zero rate of "escapes" and a practically acceptable low rate of "false alarms". Instead, we are searching for a model-based approach that estimates lifetime or fail rate with respect to stress migration for each potentially critical spot (single via) on the circuit in an arbitrary surrounding metal configuration. In this paper, we will present an estimation of relative via criticality based on the characteristics of a vacancy diffusion process as a first step into this direction. Strong focus was put on user-friendly execution and presentation of results in order to enhance user acceptance and support fast, smooth, and wide-spread application.