Although Negative Bias Temperature Instability (NBTI) was first observed more than 30 years ago, and the phenomenon has been a focus topic of the microelectronics reliability community for some years now, there is still no generally agreed physical explanation. Even the interpretation of some empirical features of NBTI is a matter of ongoing debates. One of these features is the permanent part of the degradation: that part which does not recover upon removal of the stress gate voltage, and which only can be made to vanish by e.g. thermal annealing.
In accelerated stress tests, which force a constant stress on the device, the relaxing and permanent part build up simultaneously. Since the relaxing part in most cases overshadows the permanent part, in order to assess the latter, the former has to be made to vanish by suitable relaxation of the device. Given the fact that a typical relaxation lasts some four decades longer than the previous stress phase, only very short stress cycles are allowed. One must also be very careful when, for example, applying positive voltages in order to enhance relaxation. By doing so, some amount of the relaxing component is indeed swept out quickly, and for a certain time no more relaxation may occur, leading to the false impression of a leveling-off of the relaxation curve towards the permanent component. However, if the relaxation is continued, it can be anticipated that it will proceed as a regular relaxation without positive bias pulse later on.
In contrast to the constant stress tests used for assessment, a transistor incorporated in a circuit will hardly ever experience constant stress conditions. For such a transistor, the relaxing part of the degradation will stay within certain bounds given by the duty cycle of the transistor. In contrast, the permanent part increases steadily over time, eventually determining the lifetime of the device. For this reason, investigation of the permanent component of NBTI is of great importance.