Computer-aided integrated circuit reliability analysis

Date

1970

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

Integrated circuits have evolved from early transistor technology as a result of the increasing reliability requirements of the systems of the last decade. The actual realization of required reliability levels has been due in large part to the discipline of reliability physics and to the steady improvement and stabilization of manufacturing processes. Historically the reliability physics approach has been a simple three-step process of: 1. Create failures through stress testing or system operation. 2. Perform a failure analysis to determine failure defect, mode, and mechanism. 3. Develope a qualitative model describing the mechanism and initiate corrective action. This procedure has been highly effective in reducing failure rates to the levels of the order of .001% per thousand hours. However, it has not been very successful in the area of predicting the time-to-failure of devices through a development and application of quantitative models. It is the purpose of this research to extend the reliability physics approach to accomplish this latter objective. In general, this research develops and illustrates an approach that can be effective in quantifying defect models to the point of permitting defect time-to-failure predictions. Specifically, this research has taken the well accepted qualitative model for oxide charge migration under temperature and voltage stresses and derived the analytic and numerical relationships for the device parameters versus defect and stress conditions. These relationships are then implemented and the resulting behaviour of the devices under various stress conditions and defect distributions are studied. The results demonstrate the ability of the approach to determine times- to-failure for device components as well as to illustrate the power of the approach for investigating the effect of different defect-geometry-stress configurations. Neither of these results could be determined experimentally without considerable difficulty and expense. Thus, the research points to a new technique for extracting the maximum amount of information out of reliability physics efforts at a minimum cost.

Description

Keywords

Citation