Study On NBTI Lifetime Prediction Method Of Deeply Scaled PMOSFET Devices

As the downscaling of CMOS dimensions continues, power consumption and device reliability problems become increasingly severe, and gradually become a bottleneck of the development of the integrated circuit. Among many reliability issues in nanometer CMOS devices, NBTI and random telegraph noise (RTN) become the major reliability issues below 65nm node in CMOS devices. This paper focuses on serious RTN noise in NBTI effect of deeply scaled pMOSFET devices. Under NBT stress conditions, the impact of RTN on threshold voltage shift and lifetime prediction is launched on. Pointing out that device parameter average method and TVF method exist shortcomings to solve the noise in NBTI degradation data, and we propose a new first-order differential to analyse RTN information, and further establish the NBTI lifetime prediction model which contains RTN noise. The main results are as follows:1) Based on 3D TCAD simulations, we study the impact of oxide trap on device performance at very close to the Si/SiO2. The simulation contents include the impact of gate voltage, device size, oxide thick on the current shift caused by positively charged trap. And besides, we also study that the impact of the position and number of trap on threshold voltage Vth shift.2) Measuring the time traces of threshold voltage degradation of deeply scaled pMOSFET under NBT acceleration stress condition. Using different methods to extract the signal noise, we separate the ΔVth to the Vth quasi-static component and nosie component. It is found that the conventional device parameter average method seriously underestimates NBTI degradation, while the pressure and temperature characteristics in the quasi-static component and noise component obtained by TVF method are different from the physical mechanism; This paper presents a first-order difference method for processing data. This method which separates to Vth quasi-static component and noise component is more reasonable and accurate, and the noise component is consistent with the physical mechanism of RTN noise, showing good voltage stress characteristics without temperature dependence. The first-order difference method provides accurate data processing method for noise analysis in deeply scaled devices. Under accelerated NBT stress condition, we study Vth quasi-static component and noise component using first-order difference method, and further establish NBTI lifetime prediction model and successfully achieve lifetime prediction.