用于通用存储和神经形态计算的相变存储器的研究进展

Research Progress on Phase Change Memory for General Storage and Neuromorphic Computing

存算一体技术目前被认为是一种可以消除冯·诺依曼计算架构瓶颈的可行性技术。在众多的存算一体器件中,相变存储器(PCM)因其具有非易失性、可微缩性、高开关速度、低操作电压、循环寿命长以及与现有半导体工艺相兼容等优点,被认为是未来通用存储和神经形态计算器件中最具竞争力的候选者之一。首先介绍了PCM的工作原理和器件材料结构,并详细讨论了PCM在通用存储和神经形态计算领域的应用。PCM具有高集成度和低功耗的共性需求,但这两个应用领域对材料性能有不同的侧重点。详细分析了PCM目前存在的优缺点,如高编程电流导致的功耗问题,以及商业化应用面临的主要挑战。最后,针对PCM的研究现状提出了一系列改进措施,包括材料选择、器件结构设计、预操作、热损耗降低、3D架构,以及解决阻态漂移等问题,以推动其进一步发展和应用。

In-memory computing technology is considered as a feasible solution technology that can eliminate the bottleneck of Von Neumann computing architecture.Among many in-memory computing devices,phase change memory(PCM)is considered as one of the most competitive candidates for future general storage and neuromorphic computing devices because of its advantages such as non-volatili-ty,miniaturization,high switching speed,low operating voltage,long cycle life,and compatibility with existing semiconductor processes.Firstly,the working principles and device material structures of PCM are introduced,and its applications in general storage and neuromorphic computing fields are discussed in detail.PCM has the common requirements of high integration and low power consumption in these two application fields,although they emphasize different aspects of material performances.Then,the advantages and disadvantages of PCM are analyzed in detail,such as the increased power consumption caused by high programming current,and the primary challenges in its commercial applications.Finally,a series of improvement measures are proposed based on the current research status of PCM,including material selection,device structure design,pre-operation,reduction of thermal loss,3D architecture and addressing issues of resistive drift,in order to promote its further development and application.