Accelerated Testing in FAMOS Devices - 8K EPROM

Abstract

The results of multi-temperature operating life and charge retention bake tests of 8K EPROM FAMOS devices indicate that there is a higher activation energy for the charge loss mechanism during a bake without bias than during a dynamic operating life test. This single outcome is sufficient to prove that charge retention bake data should not be used to predict device reliability at system conditions. In fact, the bake data should not even be used to determine the failure rate contribution due to charge loss at system conditions. The charge retention bake study shows derated failure rates two orders of magnitude lower than the derated failure rate found from the operating life tests. Furthermore, failure modes were observed on operating life tests which were not found on the charge retention bake. Even a non-derated failure rate comparison in a common temperature range shows a significant difference between the operating life and the charge retention bake. For the purpose of measuring device reliability, the operating life test is recommended. The operating life test does model system use conditions, and its derated failure rates are more on the order of what is generally observed for other MOS memories.