On the "switching defects" in the SiON and high-k gate dielectrics subjected to bias-temperature stressing
Boo, Ann Ann
Date of Issue2014
School of Electrical and Electronic Engineering
Since 1960s, the most commonly cited model to explain NBTI (negative-biastemperature- instability) mechanism was the Reaction-Diffusion (R-D) model which described the evolution of Si/ Si02 interface states (t..Ni1) contributing to NBTI based on a hydrogen-transport mechanism. However, there were still many debates regarding the commonly observed instantaneous threshold voltage ( V1h) shift during stress and relaxation phase. Some believed that this instantaneous Vth shift could be solely explained by the generation and passivation of Nit at the Si/ Si02 interface based on the framework of R-D model. On the other hand, there were reports which indicated that NBTI has two mechanisms involved: 1) hole-trapping (Not) and 2) interface states generation (Nit) which were used to explain the fast and slow V111 shift observed using fast measurements. Recent reports have revealed a major intrinsic limitation of the R-D model through dynamic stress experiments. It is found that no evidence of self-limiting recovery, one of the key features of the transport-based R-D model, after repeating the stress and relaxation cycles alternately for many times. It is also observed that the jt-.Vd recovery is cyclic in nature and its amount of recovery per cycle is shown to remain constant, independent of the number of stress/ recovery cycles. This behaviour observed is inconsistent with the R-D model, which stipulated that interface states relaxation should decrease progressively with the stress/ recovery cycle. Several groups have ascribed this cyclic nature of DNBTI to the charging/ discharging of oxide traps based on thermal activation result. However, till date, the underlying nature of this fast component is still yet to be identified. It is now generally accepted that there are two main components which contribute to NBTI degradation: 1) a relatively permanent component (P) which does not recover spontaneously upon the release of stress 2) a transient, "switching" hole-trapping component (R) which is able to recover spontaneously upon the removal of gate bias. The constituent of P is believed to be the Ph centers (interface states) while the constituent of R is proposed to be the oxygen vacancy defects. We discovered that the link between NBT instability and bulk trap generation is actually found in this transient "switching" hole-trapping component (R) . In this work, evidence shows that substantial interface degradation under NBT stressing does not result in any apparent bulk trap generation. It is noticed that oxide trap generation only occurs when a correlated decrease of R is observed. Analysis suggests that the generated oxide traps are due to a portion of the transient trapped holes being transformed into a more permanent form of defect. Besides, experimental evidences showed that this NBT stress induced transient hole-trapping evolution could be thermally-activated and it correlates with the generation of stress induced leakage current. A similar observation applies to the Hf02 gate pMOSFET, implying that the observed hole-trapping transformation is a common mechanism for bulk trap generation across different gate oxide technologies. The activation energy of this observed hole-trapping evolution, which has non-Arrhenius temperature dependence, has been examined for both SiON and HfSiON p-MOSFETs. Most importantly, this transformation is observed to be frequency independent. These results further imply that preexisting oxide defects, usually deemed as irrelevant to NBTI, have a definite role on long term device parametric drifts. And, evidences have been provided on the possible correlation among NBTI, TDDB and ESD. The later part of this report is focused on BTl of MOSFETs with HfSiON gate dielectric. Previously, Du et a!. has revealed an abnormal slow recovery characteristic of La-doped HfSiON n-MOSFETs subjected to positive-bias elevated temperature stress. This La-induced new degradation mechanism has been further affirmed and supported by our new experimental results based on our understanding of transient trapped holes transformation. The La impact has been distinguished from the conventional PBTI mechanism. Thus, the transformation ofNBT stress induced transient hole-trapping at pre-existing oxide defects observed in this work is crucial as it marks the start of a new series of investigations and understanding of the devices ' reliability, with the aim to have a more accurate lifetime prediction (which is very much affected by scaling). Lastly, a plausible explanation for the nature of the transient "switching" hole-trapping evolution has been proposed and discussed in light of the new experimental findings.