Toward Understanding the Wide Distribution of Time Scales in Negative Bias Temperature Instability

Abstract

Difficulties with evaluating Negative Bias Temperature Instability (NBTI) are linked to fast effects occurring at microsecond or possibly faster time scales. The wide distribution of time scales involved in NBTI relaxation suggests participation of some sort of dispersion in the underlying NBTI mechanism. A universal behavior of the relaxation is observed and used to benchmark several models incorporating dispersion. The impact of the boundary condition on the model based on dispersive transport is also briefly discussed. Furthermore, on-chip circuits are designed and fabricated to measure the effect of AC NBTI up to 2 GHz on individual devices. The results on both single pFET's and inverters indicate that AC NBTI is independent of the frequency in the entire 1 Hz - 2 GHz range. This suggests that any characteristic time scale of the NBTI mechanism must be below 1 ns.

Critical Modeling Issues in Negative Bias Temperature Instability

“…This has often been interpreted as being due to the delay: charge-pumping measurements are inherently slow (seconds range) and may therefore miss a considerable fraction of interface-state recovery (18). Secondly, it has been pointed out that CP is highly invasive due to the application of positive bias and may thus distort the measurement results (56).…”

Recovery of hot-carrier degraded nMOSFETs

“…where k b is the Boltzmann constant and ∆E b is the tunneling barrier. This results in a recovery behavior that seems to be linear on log(t)-scale of interest and can be described by an empirically found, universal relaxation equation (Equation 6.3) [136,147,156,225]:…”