To be able to characterize the reliability phenonmenon of the negative and positive
bias temperature instability (NBTI/PBTI), the experimental access to the
degrading and as well observable transistor parameters has to be explained first.
State-of-the-art measurements reveal that degradation starts earlier than
[11] and continues to proceed even beyond weeks [12]. As such, both the onset
and the saturation of degradation are outside the experimental window, which
today spans about 12 decades in time. The minimum times in this window are
due to the limited resolution of the measurement equipment, while the maximum
times are restricted by the time a reliability engineer has to perform these kind of
measurements1 .
Now, a fundamental prerequisite for the description of NBTI lies in an accurate
determination of its impacts on the device. But precise measurements of the
electric parameters as proper measures of the “real” degradation (e.g. interface
state density) are not trivial. This is on one hand due to the immediate
relaxation of the degradation once the stress is interrupted, i.e.
is set to weak inversion or even accumulation. In 1977 Jeppson et al.
already described that traps created during negative bias temperature
stress can be removed by thermal annealing. The higher the temperature
during the annealing process, the quicker the degradation process recovers
and the damage is annealed [13]. Nevertheless the NBTI community
appeared not interested in the fact that degradation may be reversible
under certain conditions for many years. Hence, there was no apparent
need to quickly measure the degradation, which of course had a serious
impact on the initial modeling attempts. Rangan et al. was one of the
first to revive the discussion on the recovery of NBTI [14]. A few years
later Reisinger et al. described the influence of very fast to very slow
components contributing to degradation and recovery due to NBTI and
contrasted their results to existing physical models in [15], which will
be thoroughly discussed in Chapter 3. Today the scientific community
has accepted that fast measurements are necessary, but unfortunately
there is always a trade-off between a fast and simultaneously accurate
method.
This chapter will give a brief overview of the various measurement methods, their delay times, their effect on the device itself, and their other limitations. Moreover, their output signal post-processing complexity is discussed using approximate formulae.