Theory and Practice of On-the-fly and Ultra-fast V<inf>T</inf> Measurements for NBTI Degradation: Challenges and Opportunities

Defect Generation in p-MOSFETs Under Negative-Bias Stress: An Experimental Perspective

“…V T shift during NBT stress is extracted using OTF I DLIN [33] with 1-ms initial delay, using the expression [15], where I D0 is the initial I DLIN measured within 1 ms of the application of stress, V G is stress bias, and V T 0 is the prestress V T . The obtained ∆V T is slightly different than the actual, as mobility degradation is not taken into account (as discussed in [36]). CP measurement was done on separately stressed samples for direct determination of interface traps when required, after interrupting stress at log time intervals (SMS).…”

“…6) Different measurement techniques (e.g., I DLIN and CP) suffers from different delay and scans different zone in the energy-band gap. Moreover, degradation calculated from OTF I DLIN in its simplest form [14], [15], [33] ignores mobility degradation (details about mobilitycorrection procedure has been presented elsewhere [36]), which results in inaccurate ∆V T . Such issues must be accounted for before 1 : 1 comparison can be made across different measurement techniques.…”

“…First, it is important to choose proper stress condition such that only the damage responsible for NBTI is accelerated and no new damage mode is triggered [28]- [32]. Second, the measurement techniques must be chosen carefully so that measured data reflect the intrinsic degradation and are not corrupted by any measurement-related artifacts [33]- [36]. Finally, it is also important to understand and model the physical-degradation process [37]- [46] to perform reliable extrapolation from stress to operating condition.…”

A Finite-Oxide Thickness-Based Analytical Model for Negative Bias Temperature Instability

“…The above work also shows that the R-D theory is consistent with the experimental results obtained using OTFM techniques, and the log-like recovery (equal recovery in every decade) observed in [18], [19] is consistent with a UFM-based technique. The authors in [40] also state that the log-based recovery of V th observed in [19] is due to the inappropriate usage of the quasistate relationship:…”

“…However, [29] distinguishes the gate dielectrics into two types (Type I and Type II) depending on whether they are PNO (plasma nitrided oxides) or TNO (thermal nitrided oxides), and explains the discrepancy between the bulk trapping and the R-D models, for each of these types. Recently, [40] highlights the differences between an OTFM and a UFMbased technique for analyzing the impact of NBTI. The above work also shows that the R-D theory is consistent with the experimental results obtained using OTFM techniques, and the log-like recovery (equal recovery in every decade) observed in [18], [19] is consistent with a UFM-based technique.…”