Probing technique for energy distribution of positive charges in gate dielectrics and its application to lifetime prediction / Sharifah Fatmadiana Wan Muhamad Hatta

Sharifah Fatmadiana , Wan Muhamad Hatta (2013) Probing technique for energy distribution of positive charges in gate dielectrics and its application to lifetime prediction / Sharifah Fatmadiana Wan Muhamad Hatta. PhD thesis, University of Malaya.

Preview

PDF (Thesis PhD) Download (3534Kb) | Preview

Abstract

The continuous reduction of the dimensions of CMOS devices has increased the negative bias temperature instability (NBTI) of pMOSFETs to such a level that it is limiting their lifetime. This increase of NBTI is caused mainly by three factors: an increase of nitrogen concentration in gate dielectric, a higher operation electrical field, and a higher temperature. Despite of many years’ research work, there are questions on the correctness of the NBTI lifetime predicted through voltage acceleration and extrapolation. The conventional lifetime prediction technique measures the degradation slowly and it typically takes 10 ms or longer to record one threshold voltage shift. It has been reported that NBTI can recover substantially in this time and the degradation is underestimated. To minimize the recovery, ultra-fast technique has been developed and the measurement time has been reduced to the order of microseconds. Once the recovery is suppressed, however, the degradation no longer follows a power law and there is no industry-wide accepted method for lifetime prediction. The objective of this project is to overcome this challenge and to develop a reliable NBTI lifetime prediction technique after freezing the recovery. To achieve this objective, it is essential to have an in-depth knowledge on the defects responsible for the recovery. It has been generally accepted that the NBTI recovery is dominated by the discharge of trapped holes. For the thin dielectric (e.g. < 3 nm) used by current industry, all hole traps are within direct tunnelling distance from the substrate and their discharging is mainly controlled by their energy levels against the Fermi level at the substrate interface. As a result, it is crucial to have the energy distribution of positive charges (PC) in the gate dielectric, but there is no technique available for probing this energy profile. A major achievement of this project is to develop a new technique that can probe the energy distribution of PCs both within and beyond the silicon energy gap. After charging up the hole traps, they are allowed to discharge progressively by changing the gate bias, Vg, in the positive direction in steps. This allows the Fermi level at the interface to be swept from a level below the valence band edge to a level above the conduction band edge, giving the required energy profile. Results show that PCs can vary by one order of magnitude with energy level. The PCs in different energy regions clearly originate from different defects. The PCs below the valence band edge are as-grown hole traps which are insensitive to stress time and temperature, and substantially higher in thermal SiON. The PCs above the valence band edge are from the created defects. The PCs within bandgap saturate for either longer stress time or higher stress temperature. In contrast, the PCs above conduction band edge, namely the anti-neutralization positive charges, do not saturate and their generation is clearly thermally accelerated. This energy profile technique is applicable to both SiON and high-k/SiON stack. It is found that both of them have a high level of as-grown hole traps below the valence band edge and their main difference is that there is a clear peak in the energy density near to the conduction band edge for the High-k/SiON stack, but not for the SiON. Based on this newly developed energy profile technique and the improved understanding, a new lifetime prediction technique has been proposed. The principle used is that a defect must be chargeable at an operation voltage, if it is to be included in the lifetime prediction. At the stress voltage, some as-grown hole traps further below Ev are charged, but they are neutral under an operation bias and must be excluded in the lifetime prediction. The new technique allows quantitative determination of the correct level of as-grown hole trapping to be included in the lifetime prediction. A main advantage of the proposed technique is that the contribution of as-grown hole traps is experimentally measured, avoiding the use of trap-filling models and the associated fitting parameters. The successful separation of as-grown hole trapping from the total degradation allows the extraction of generated defects and restores the power-law kinetics. Based on this new lifetime prediction technique, it is concluded that the maximum operation voltage for a 10 years lifetime is substantially overestimated by the conventional prediction technique. This new lifetime prediction technique has been accepted for presentation at the 2013 International Electron Devices Meeting (IEDM).

Actions (For repository staff only : Login required)