Analysis of Oxide Traps in Nanoscale MOSFETs using Random Telegraph Noise

This chapter describes the use of random telegraph noise (RTN) to obtain information about traps in highly scaled MOSFETs. A robust hidden Markov model (HMM) algorithm is presented to enable the accurate extraction of trap parameters from both single and multiple-trap signals. The results of a large number of measurements show that even in the absence of bias stress, RTN-generating traps can cause serious variation for high-k/metal gate (HKMG) FETs and that undoped channels do not reduce the problem. Trap time constants are shown to have wide ranging dependence on bias and temperature, leading to hysteretic behavior with time constants much longer than the circuit timescale. The impact of RTN on the stability of memory cells is also presented, along with experimental observations of these effects in SRAM arrays.