Investigation and Modeling of Interface and Bulk Trap Generation During Negative Bias Temperature Instability of p-MOSFETs

Abstract

Negative bias temperature instability is studied in thick and thin gate oxide p-MOSFETs. The relative contributions of interface-and bulk-trap generation to this device degradation mode are analyzed for a wide range of stress bias and stress temperature. The effects of gate voltage and oxide field, as well as those of inversion layer holes, impact ionized hot holes, and hot electrons on interface-and bulk-trap generation, are identified. The bulk-trap generation process is interpreted within the modified anode-hole injection model and the mechanism of interface-trap generation is modeled within the framework of the classical reaction-diffusion theory. The diffusion species for interface-trap generation is unambiguously identified. Moreover, a high-temperature, diffusion-triggered, enhanced interface-trap generation mechanism is discussed for thin gate oxide p-MOSFETs. Finally, a novel scaling methodology is proposed for interface-trap generation that helps in obtaining a simple, analytical model useful for reliability projection.

A Comparative NBTI Study of $\hbox{HfO}_{2}$, $\hbox{HfSiO}_{x}$, and SiON p-MOSFETs Using UF-OTF $I_{\rm DLIN}$ Technique

“…2 shows extracted time exponent (n) for t-stress > 10 s at different stress E OX for HfO 2 and HfSiO x devices having different high-k thicknesses and for PNO (with different interfacial N density) and RTNO devices. Note that, although n is different for different devices, it is independent of stress V G (E OX ), suggesting absence of trap generation in high-k or SiON bulk during NBT stress [14]. Although not explicitly shown, n was found to be independent of stress T as well for all the devices used in this letter.…”

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

“…4 (top) shows the T activation of NBTI obtained at a fixed level of degradation by timeaxis scaling of T -dependent ∆V T (t) data for various PNO (varying N dose and EOT) and control samples. The activation energy so obtained is defined as activation of diffusion, E A (D), for reasons explained in Section II-D [8]. Identical E A (D) is obtained for all splits, which also equals to that obtained from similar time-axis scaling of T -dependent ∆N IT (t) data (from CP) for thick PNO samples [17], [20].…”

“…At high V G , ∆V T shows identical (as low V G ) degradation rate for short stress time but breaks off from the initial trend and drastically increases at long stress time. The break-off time decreases and postbreak slope increases as stress V G is increased [8], [28]. Such behavior has also been observed in thinner oxynitride devices [30] and needs careful attention.…”

“…It is generally believed that N IT generation would show power-law time dependence and strong T activation [8], [14], [15], [17], [19], while hole trapping in preexisting traps would show log time dependence and weak T activation [23]- [27]. Note that power-law time dependence (Fig.…”

“…A number of attempts have been made to model NBTI [8]- [28], [37]- [46], but given the complexity/diversity of the experimental dataset as discussed earlier, none of the theoretical models have been able to explain NBTI degradation within a common framework. We suggest that this is primarily due to large scatter in published data, as several groups have used different stress conditions, measurement methods, and gateoxide processes (unspecified chemical composition and thermal treatment).…”