Engineering neuromorphic phase-change memory: Carbon-doped GeSbTe with high thermal stability and low resistance drift

Abstract

Phase-change memory (PCM) has emerged as a promising non-volatile memory technology, offering significant potential for next-generation artificial intelligence and neuromorphic computing systems. However, conventional Ge1Sb4Te7 (GST), a prototypical stoichiometric phase-change chalcogenide, suffers from intrinsic limitations such as inadequate thermal stability and pronounced resistance drift, hindering its practical applications in high-performance devices and chips. In this study, we demonstrate that carbon (C) doping in GST markedly enhances its thermal robustness and data retention, while elucidating the underlying microstructure property relationships. Carbon doping significantly increases the crystallization temperature of GST, shifting it to and beyond 200 °C with increasing carbon content. Higher carbon incorporation also yields up to a fourfold improvement in data retention, achieving 10-year stability at 100 °C. Moreover, GST-C-based PCM devices exhibit excellent electrical stability, featuring ultralow resistance drift (ν = 0.03) and highly reproducible multilevel resistance states. Through ab initio simulations, we uncover the atomic-scale mechanisms governing these enhancements: carbon incorporation induces the formation of robust, shortened bonds with Ge/Sb/Te, promoting tetrahedral C clusters that impede crystallization by elevating the activation energy barrier. This work identifies GST-C as a promising candidate for reliable, high-density PCM and highlights its potential for neuromorphic computing applications.