Reassessing the Mechanisms of Negative-Bias Temperature Instability by Repetitive Stress/Relaxation Experiments

On the permanent component profiling of the negative bias temperature instability in p-MOSFET devices

“…Schroder et al [17] have reported dispersion of NBTI E a,eff values ranging from 0.15 to 0.325 eV, depending on the chemical composition of the dielectric and on the phenomenon that dominates the reaction kinetics as well as the species involved. Published reports on macroscopic NBTI modeling [11] showed that E a,eff for hole trapping in the pre-existing traps is about $0.04 eV, which is typically expected for tunneling processes without structural relaxation, while the signature of DN it is 0.1 eV in nitrided oxides [9]. In contrast, using time-dependent defect spectroscopy (TDDS) method, Grasser et al [18] found that the physical microscopic E a,eff of the individual traps is about 0.5-1.…”

“…It has been shown elsewhere [4] that the reaction-diffusion (R-D) model and its variants cannot explain the relaxation phase as well as the duty cycle dependence. In addition, R-D model alone fails to explain the cycling behavior and the relaxation in on the fly (OTF) threshold voltage OTF-V th and ultra fast V th experiments observed by Ang et al [9]. These findings increase doubts over whether the diffusion mechanism is really driving interface state generation and recovery [4,9,10].…”

“…Discrepancies have been reported by several groups [4,9] regarding the relaxation phase [4], and n [9]. Although the improvements brought to the R-D model to match experimental data giving subsequently a steady value of 0.16 for exponent n. Ang et al [9] have shown that T dependence of n emanates from the superposition of two mechanisms with different thermal activations in cyclic NBTI.…”

“…2 area devices. They argued that the small overall macroscopic E a,eff (0.1 eV) [8,9], observed in large gate area devices, is the result of individual trap energy superposition. Thus, the macroscopic E a,eff in large area is an apparent activation energy resulting from a large number of defects [17].…”

“…The large spread in NBTI experimental features complicates all attempts to model the NBTI degradation. These disparities suggest that different defects with different structures [3,5,9,16,18,19] are involved in the NBTI phenomenon. These defects do not necessarily have the same E a .…”