Characterization, Understanding, and Modeling of Lorentzian Noise Phenomenon in Advanced Dynamically Depleted SOI MOSFET

In this work, the current noise power spectral density (PSD) of advanced dynamically depleted (DD) silicon on insulator (SOI) MOSFETs have been characterized. By comparing the PSD in body contact (BC) mode with that in floating body (FB) mode, the excessive Lorentzian-like noise in BC mode is investigated. Through analyzing the physical mechanisms of various Lorentzian noises and using kernel density estimation (KDE) method, this distinctive noise phenomenon is mainly attributing of the physical locations of traps in the silicon film, as well as the filling levels of the energy states under various bias voltages in different device operating modes. To model this observed phenomenon, a physics-based analytical body potential model is developed, and then a new trap-based Lorentzian-like noise model is built based on the body potential. The model agrees well with multiple sets of measured data for various device geometries, validating the proposed mechanisms. Furthermore, the new noise model is integrated into the standard compact model (e.g., BSIM-SOI-100) for circuit-level analysis to provide accurate noise characteristics. This work paves the way for accurate noise simulations in future integrated circuit designs on the SOI technology platform.