Strategic Issues on Yield, Reliability, and Their Relationship in Semiconductor Manufacturing

点击翻译按钮获取中文摘要

This grant provides funding for studying the reliability and yield relationship and its association with burn-in at the wafer level. In order to guarantee the required reliability of integrated circuits (ICs), burn-in has been conducted as an effective method for removing infant mortalities. Semiconductor yield is a process-dependent quality parameter; burn-in a time dependent control parameter; and reliability both a time- and a process-dependent parameter. This study models yield and reliability by inter-relating these three parameters. This study will investigate the following key issues: (1) to apply the physics of failure mechanisms to find a relationship between yield and reliability, (2) to develop a yield-reliability relation model and to theoretically justify this model for semiconductor manufacturing, (3) to determine burn-in effect on yield, based on the relation model, and (4) to find optimal burn-in strategies for semiconductors which include yield gain and yield loss.If successful, the relation model can be used to predict the reliability of new products. In addition, the projected reliability of gate oxide and other defects may be found. By combining burn-in effect with the yield-reliability relation model, it is possible to find the minimum effort for meeting the reliability requirement at a given operation time. To arrive at a rational decision regarding yield, this research describes models which give optimal burn-in times and levels. It can be used to find a cost-effective burn-in solution at the wafer level and the die level. The burn-in coverage strategy will provide an optimal approach for implementing the burn-in system. The research study will be the first to provide solutions to the yield-reliability problems in an analytically justified way.