Challenge and solution for characterizing NBTI-generated defects in nanoscale devices

Abstract

Negative bias temperature instability (NBTI) is a well known ageing process for CMOS technologies. Many early works were focused on large devices where device-to-device variations (DDV) are negligible. As device sizes downscale to nanometers, DDV becomes substantial. NBTI is a stochastic process and causes a time-dependent DDV. Characterizing the NBTI-generated defects in nanoscale devices has two main challenges. First, current fluctuates with time and this introduces uncertainties in measurements. Second, the test time is long and costly: to characterize the NBTI-induced DDV, it is essential to repeat the same test on multiple devices. This work reviews recent progresses in addressing these issues. Based on the As-grown-Generation (AG) model, it will be shown that the measurement uncertainties are dominated by As-grown hole traps and can be removed by subtracting the average value. To reduce the test time, the voltage step stress (VSS) technique is combined with the Stress-Discharge -Recharge (SDR) method. This VSS-SDR technique reduces test time to within one hour per device. The model extracted by VSS-SDR is verified by comparing its prediction with the test data obtained under conventional constant voltage stress.