With the continuing miniaturization of MOS transistors, a phenomenon called the negative bias temperature instability (NBTI) has evolved into a serious reliability concern. In the newest device technologies, its detrimental impact complicates reliable lifetime projection of the devices. Even though this fact has aroused large industrial and scientific interest, the complex mechanism behind NBTI has only been partially clarified so far. At least it has been agreed that substrate charge carriers are captured in oxide or interface defects and remain there until the end of the stress period. This is associated with a threshold voltage shift, which affects the device characteristics and considerably shortens the device lifetime. During relaxation, the same charge carriers are suspected to be emitted from the oxide defects so that the threshold voltage returns towards its initial value. The capture and emission of charge carriers during stress and relaxation are known under the name charge trapping and will be the focus of this thesis.
Charge trapping involves a transfer of charge carriers between the channel and defects. Several different models from literature describe this charge transfer at various levels of sophistication. They are re-examined in the light of NBTI and incorporated into an existing device simulator for comparison to measurements. This evaluation is based on a list of criteria, which include particular features of the device behavior under different operation temperatures, gate biases, and stress times.
In the simplest model, the charge transfer is based on elastic tunneling, which involves an electron whose energy must be preserved during a transition. This kind of a charge transfer reaction is the basis for the elastic tunneling model, studied as the first candidate for NBTI. A special focus is put on the temperature behavior of this model, requiring an investigation of the quantization effect within the channel of a MOS transistor. Furthermore, the oxide thicknesses have been downsized to a few nanometers in modern MOS technologies so that previous investigations must be extended to account for elastic tunneling to and from the gate contact.
A more sophisticated concept also accounts for the fact that the defect configuration plays a crucial role in the charge transfer process. After a defect has captured a charge carrier from the substrate, it undergoes structural relaxation, including strengthening, weakening, or even breaking of bonds. Most importantly, this relaxation results in a shift of the trap level — a fact that has remained unconsidered so far. Using first principles simulations, it is proven that several defects have a large level shift whose effect on the trapping dynamics needs to be studied in more detail. A new model is developed, which accounts for the level shift and is evaluated based on the list of experimental criteria mentioned before.
The most realistic description of the charge transfer is given by nonradiative multiphonon (NMP) theory, which has initially been developed for light absorption of molecules and later generalized to charge capture and emission in solids. This kind of process involves an activation over a thermal barrier and thus leads to a temperature dependence, which is missing in the case of the elastic tunneling model for instance. In this thesis, a simplified variant of this process is used for the so-called two-stage model, in which charge trapping is coupled to a hydrogen reaction. Even though this model can successfully reproduce the threshold voltage shift observed in NBTI experiments, it does not reflect the correct microscopic processes as shown by time-dependent defect spectroscopy (TDDS). In an extended variant of the two-stage model, this deficiency has been overcome by refining the description of the charge transfer process and incorporating a metastable state in this model. With these modifications, the improved two-stage model can also give an explanation of the noise phenomena observed in random telegraph noise and TDDS measurements.
In conclusion, hole trapping in NBTI is investigated using different explanations for the charge transfer. It is demonstrated that the refined variant of the two-stage model is consistent with the plenty of experimental features seen in NBTI and noise measurements. For this reason, this model is expected to be the best description of hole trapping from the present perspective.