6 Short-Term NBTI

Chapter 6
Short-Term NBTI

During long-term stress, most measurements indicate that $ΔVTH$ follows a power-law as $Atn$ [7, 49, 102, 103]. However, log-like behavior, in particular at short times, has also been reported [84, 11, 40]. Both cases are depicted in Fig. 6.1 and Fig. 6.2.

Figure 6.1: A pMOS with $tox = 1.4nm$ $SiON$ yields a different stress behavior when comparing the short-term to the long-term behavior and the question arises whether there are two mechanisms contributing to NBTI or one. Unscaled stress and restress phases of the extended MSM sequence provided by Kaczer et al. [17, 18].

Figure 6.2: A pMOS with $tox = 1.4nm$ $SiON$ yields a different stress behavior when comparing the short-term to the long-term behavior and the question arises whether there are two mechanisms contributing to NBTI or one. Scaling the data shown left gives a universal curve which more clearly reveals the log(t) versus the power-law dependence.

The conventional explanation of the resulting degradation uses elastic hole trapping due to tunneling carrier exchange with the substrate (initial degradation) [84, 11] and the creation of interface states (long-term degradation) [94, 104]. While [94, 104] claim that processes in the short-time scale show a negligible temperature dependence, the latest results support a thermally activated tunneling mechanism [98, 78, 105] (cf. Chapter 5).

Previous short-time measurements using conventional parameter analyzers with a time resolution in the millisecond regime indicate that at least for up to medium stresses a logarithmic time dependence is observed during the first three decades ( $1ms$ up to $1s$ ) [106]. This logarithmic short-term degradation shows a strong temperature activation and a quadratic stress field dependence ( $∼E2 ox$ ) up to medium stress ( $≈ 5MV ∕cm$ ). For longer stress times and higher stress fields ( $≈ 7MV ∕cm$ ), degradation starts to deviate from the logarithmic behavior [106, 98].

To better understand the underlying mechanisms of short-term NBTI degradation, an extensive study of the short stress time behavior far below the range of milliseconds to seconds needs to be performed. Unfortunately, accurate measurements in these time scales are difficult to access due to noise [107, 42]. In particular, the noise in the $μs$ regime makes it difficult to extract information on the smallest time-constants contributing to the degradation. The currently used measurement methods for fast NBTI evaluation [12] are briefly summarized, based on Chapter 2:

(i) The fast- $VTH$ method [11, 15] (Chapter 2.1.2) interrupts the stress ( $μs$ delay) to quickly record $VTH$ during recovery.

(ii) The fast- $ID$ method [18, 17, 106, 30, 20] (Chapter 2.1.1) monitors the drain current $ID$ near $VTH$ , which is then converted to $ΔVTH$ [106] using an initial $I (V ) D G$ curve. This characteristic is only recorded around $V TH$ so as not to prestress the device.

(iii) The on-the-fly (OTF) method [36, 28, 6] records the degradation during stress and hence does not introduce unwanted recovery, but suffers from mobility degradation, which leads to a spurious $ΔVTH$ [108, 41] (Chapter 2.3).

While OTF suffers from the problem of the initial reference measurement, which already stresses the device, the fast- $V TH$ and the fast- $I D$ methods can record an unstressed reference value but suffer from the delay during measurement [18, 12]. Due to its non-stop recording nature, methods (i) and (ii) [18, 12] can continuously monitor recovery and, thus, allow an extrapolation back to shorter measuring delays.

Based on this experience fast rectangular gate pulses are used for short-term NBTI degradation in the range of $1μs$ to $1s$ here. Recalling that previously published results [106] only feature a minimum time of $1ms$ means that the number of decades in time for short-term degradation is doubled from three to six. This method is called improved fast pulse method (Chapter 2.2.2) and will be explained thoroughly in the next two sections and is finally compared against the fast- $VTH$ method of [15].

Chapter 6Short-Term NBTI

Chapter 6
Short-Term NBTI