Understanding the Variability in GeAsTe Ovonic Threshold Switching Devices

Abstract

Ovonic threshold switching (OTS) devices have recently demonstrated strong performance as selectors in high-density cross-point 1S1R emerging memory arrays to suppress the sneak current path and as selector-only-memory (SOM), with excellent lon, lon/Ioff nonlinearity, and endurance. However, a detailed variability study is still lacking, and understanding of the responsible mechanisms is limited. In this work, different cycle-to-cycle (C2C) Vth variability mechanisms in GeAsTe OTS are identified: (i) Defects that remain active at the off-state lead to the less than 0.2 V intrinsic small random variability (SRV); (ii) Defects activated at the on-state result in up to 1V large pseudo-random variability (LPV). Novel techniques such as the fast frequency-domain noise analysis and the sequential variability analysis are developed to identify the SRV and LPV mechanisms and to characterize their statistical correlations with different defects. The observed single modal, bimodal and multimodal C2C Vth variability distributions can be explained by the combination of SRV and LPV at different pulse operation conditions. This work provides critical guidance for OTS variability optimization.