Materials Design for Phase-change Memory and Neuro-inspired Computing

Wei Zhang*

Xi’an Jiaotong University, Xi’an, 710049, China

ABSTRACT: The global demand for data storage and processing has increased exponentially in recent decades. To respond to this demand, research efforts have been devoted to the development of non-volatile memory technology that combines fast operation speed with persistent storage when power off. Chalcogenide phase-change materials (PCMs) are leading candidates for such application, and have become technologically mature with recently released competitive products, i.e. 3D Xpoint. Phase-change memory utilizes the large contrast in electrical resistance between the amorphous and crystalline phase of PCMs to encode data. Yet, the multilevel capability of PCMs enables efficient emulations of artificial neural network for neuro-inspired computing that unifies computing with storage in memory arrays. The flagship PCM is Ge₂Sb₂Te₅, but the intrinsic stochasticity of nucleation and spontaneous glass relaxation at room temperature prevents further improvement on the computing and power efficiencies of phase-change devices. In this talk, I will focus on the mechanisms of the crystallization dynamics and aging relaxation of PCMs by discussing structural and kinetic experiments, as well as ab initio atomistic modelling and calculations. Based on the knowledge gained at the atomic level, potential routes to improve the crucial parameters, such as switching speed, drift coefficient, as well as cycling endurance of phase-change devices are depicted by ab initio materials design. Two designed new materials, namely, the Sc_0.2Sb₂Te₃ alloy that breaks down the switching time to subnanosecond regime and the TiTe₂/Sb₂Te₃ superlattice that resolves the resistance drift issue, will be discussed in detail.

Fig. 1 Crystallization of PCM.

Keywords: Materials design; first-principle simulation; phase-change memory; neuro-inspired computing.