Statistical simulations to inspect and predict data retention and program disturbs in flash memories

Abstract

A new statistical model of stress-induced leakage current (SILC) is implemented and used to predict data retention and program disturbs of state-of-the-art flash memories, and to correlate oxide characterization outputs (density, cross section, energy level of defects) to flash memory reliability. Physical mechanisms inducing the largest threshold voltage (V/sub T/) degradation are explained, and tunnel oxide scaling effects on flash reliability are predicted.