Optimized test circuits for SER characterization of a manufacturingprocess

Novel test circuits for the accurate determination of soft error rate (SER) dependency on critical charges Q_CRIT have been developed. The minimum charge necessary for flipping the state of a sensor cell, denoted by Q_CRIT, is measured with 1%-2% accuracy before exposing the circuits to radiation. During the accelerated testing, circuits biased with multiple different supply voltages V_CC are simultaneously placed into a beam and any bit flips are logged. From the measured SER dependency on V_CC and previously measured Q_CRIT dependency on V_CC, the dependency of SER on Q_CRIT can be deduced by correlating V_CC's for the two measurements. Furthermore, the sensor cell utilizes a single dynamic node which can be programmed to detect strikes on either N- or P-type diffusions, but not both at the same time. The measured dependency SER(Q_CRIT), normalized by the diffusion area, can be used for predicting SER of any other circuit fabricated in the same process and aid designers in optimization for reduced SER. Predictions of a theoretical SER model, if one is available, can be compared directly with the measurements. Since the true Q_CRIT of the test circuits is known accurately, any discrepancy larger than given by the measurement uncertainty of SER(Q_CRIT) would be clearly due to limitations of the SER model. We implemented the test circuits in a 0.6-μm bulk CMOS process and verified accuracy of Q_CRIT(V_CC) calibration method