Increasing the IDDQ test resolution using current prediction

The Variations in IDDQ due to process disturbances for sub-micron ICs is comparable to magnitude of defect induced currents. This is making traditional IDDQ testing ineffective in detecting defects in sub-micron ICs. This paper presents a methodology called current prediction for enhancing the effectiveness of IDDQ testing. In the proposed methodology, a set of IDDQ measurements are performed on the device and the value of each IDDQ current are predicted using regression models. Absolute value of the difference between measured and predicted IDDQ (residuals of current prediction) are used for identifying defective devices. The residuals of current prediction are very sensitive to the defect currents and is insensitive to process variations thus increasing the IDDQ test resolution. This technique is compared with traditional and delta IDDQ testing techniques using the production test data from two different ICs. Results show that considerable improvement in the IDDQ test quality can be achieved with the proposed technique.