6 Conclusion and Outlook

The impact of mixed NBTI/HC stress conditions on SiON pMOSFET characteristics was studied with a focus on the recoverable component of degradation. It was found that recovery of large-area devices after different stress conditions can clearly deviate from the behavior expected in a simple electrostatic model. This is a strong evidence that fewer defects contribute to recovery than would have been expected. In this context, the study of the degradation behavior in large-area devices gives a hint that impact ionization and secondary generated carriers play an important role at mixed NBTI/HC conditions. However, from measurements of large-area devices, the processes relevant for the recoverable component cannot be fully characterized. The study of the impact of mixed NBTI/HC stress on the behavior of single oxide defects showed that source-side defects can remain neutral after mixed NBTI/HC stress and do not contribute to recovery although they are charged after homogeneous NBTI stress. Such a behavior cannot be explained by an electrostatic model only since the electrostatic conditions at the source side are almost unaffected even at high . The founding is that defect characteristics like and distort and, as a result, if a defect captures a charge carrier it emits it immediately afterwards still at stress conditions. Consequently, the defect remains neutral after mixed NBTI/HC stress. This effect is determined by non-equilibrium processes triggered by hot carriers with sufficient energies and depends on the detailed defect configuration. With a model extension, including a thorough carrier transport treatment and under consideration of secondary majority carriers in the channel, recovery after different stress conditions can be modeled properly.

Measurements show that recovery after homogeneous NBTI stress depends strongly on the “history" of the device. Especially if stress and recovery cycles with mixed NBTI/HC stress are performed, recovery after homogeneous NBTI stress is reduced continuously and considerably with respect to the cumulative mixed NBTI/HC stress time. Measurements at a single defect level show that this reduction is due to both, changes of the defects step height and deactivation of defects. The change of the step height is most probably attributed to a distortion of the electrostatic surrounding due to the activation or deactivation of other defects in the vicinity of the observed defect. With a much greater impact on the reduction of recovery, some defects simply disappear from the measurements after numerous cycles of mixed NBTI/HC stress. As a result, recovery after homogeneous NBTI stress can be reduced even to zero. Such a deactivation of oxide defects has been assumed to be attributed to defect volatility, a repeated dis- and reappearance of oxide defects. However, a continuous reduction of recovery as well as a seemingly “permanent" deactivation of some of the defects are evidence that a permanent deactivation might play a role here. For a full characterization of such a reduction of recovery, further long-term measurements of a large number of individual defects and a thorough analysis of self-heating effects are requried.