Abstract

Whatever industrial environment, failure rate prediction is important at different levels. Firstly, this prediction will be provided to customers when several failures with a same signature are observed in field: customers’ request will fit with the number of future failures expected in field, from the number of already observed ones. A field modeling method is now typical for this case. But, before field step, failure rate prediction is performed during qualification for a new product: all the accelerated stress tests implemented in qualification aim field failure rate prediction: at this qualification step, we speak about reliability tests to guarantee product working during the full mission profile, or robustness tests to study product working limits until part breakage. In this paper, an innovative risk assessment method is presented for a gate oxide rupture defect, in automotive semiconductor industry. For the first time, a field risk assessment method uses results from a specific manufacturing test: features of this manufacturing test are not those of a reliability test nor a robustness one, and the most important input data for this assessment is defect density measured at manufacturing. A study case on a valve driver allows to precisely describe this innovative method, to identify its limits and to study possible implementation to other defects or products.