Reliability methodology for prediction of micromachined accelerometer stiction

The accurate prediction of known failure mechanisms is required for success in the integrated circuit marketplace. This study reports an empirically-generated stiction field prediction model methodology for accelerometers prone to stiction failure. A probability distribution function was generated as a function of shock level in g's. Although the population was all very stiction prone, the failure mode is probabilistic and parts have a high survival rate.