Physical model for Random Telegraph Noise amplitudes and implications

Abstract

Random Telegraph Noise (RTN) has been shown to surpass random dopant fluctuations as a cause for decananometer device variability, through the measurement of a large number of ultra-scaled devices [1]. The most worrisome aspect of RTN is the tail of the amplitude distribution - the limiting cases that are rare but nevertheless wreak havoc on circuit yield and reliability. Since one cannot realistically measure enough devices to imitate a large circuit, a physics-based quantitative model is urgently needed to replace the brute force approach. Recently we introduced a physical model for RTN [2-3] but it contains a serious error. In this paper, we developed and experimentally verified a new model that provides a physical understanding of RTN amplitude. By providing a quantitative link to device parameters, it points the way to control RTN in decananometer devices.