Test method for moisture-driven failure: Effect of VLSI device self-heating and local temperature rise

Peck's model for temperature-humidity-bias test correlation has been well studied and applied to estimate field failure at the use condition. The model postulates the effect of environment temperature and humidity to time-to-failure at stress or at use condition. Since Peck's model was published in 1986, device scaling and integration led to an increase in device self-heating which reduces the local relative humidity around the device in use and during reliability stress testing. Reduced local humidity will reduce the acceleration of humidity driven failure mechanisms in accelerated Temperature Humidity Bias (THB) stress. This paper demonstrates and discusses the effect of device self-heating on the local temperature and humidity during accelerated reliability stress. The study demonstrates that self-heat reduces the moisture locally and hence changes the failure probability dramatically. Therefore test vehicles without transistors (no self-heating) are recommended to envelope accelerated THB stress risk for humidity accelerated failure modes.