Defect Level

Abstract

Consumers of large quantities of integrated circuits usually consider the reliability of the parts to be a major factor when selecting a manufacturer from whom they will purchase the devices they need. Specifically, if some consumer determines that the mass production of a system will require exactly 1000 copies of an IC, they obviously desire to be shipped at least 1000 functional chips. IC manufacturers generally protect themselves by shipping more parts than were ordered, knowing that since the production testing of the parts was not exhaustive, some parts may have been declared “good” when they were actually faulty. Because this issue of “quality level” is intimately tied to the level of testing to which the parts were subjected and the quality of the test itself, researchers have sought to more formally define quality level. The goals of this research are to study fault models and the quality of test sets that they provide. Thus, one possible benefit of this work is a more effective use of testing to screen out bad parts before they are shipped to the customer. It is therefore proper to review the more interesting developments in IC quality research.