Katehi Linda, Yi Su-In, Lin Yuxuan Cosmi, Banerjee Sarbajit, Xia Qiangfei, Yang J Joshua
Electrical Engineering and Materials Science and Engineering Texas A&M University, College Station, Texas 77843, United States.
Chemistry Department, Texas A&M University, College Station, Texas 77843, United States.
ACS Omega. 2025 Jun 30;10(27):28492-28498. doi: 10.1021/acsomega.5c01414. eCollection 2025 Jul 15.
Memory technologies are central to modern computing systems, performing essential functions that range from primary data storage to advanced tasks, such as in-memory computing for artificial intelligence (AI) and machine learning (ML) applications. Initially developed solely for data retention, these technologies are evolving to support new paradigms, such as in-memory computing, where processing occurs directly within the memory array. This evolution significantly enhances computational efficiency by minimizing data transfer between processors and memory, resulting in increased speed and reduced energy consumption, critical factors for AI and ML workloads. Such demanding requirements are driving innovations beyond traditional complementary metal-oxide semiconductor (CMOS) technologies. Emerging nonvolatile memories (eNVMs) represent a promising class of technologies designed to replace or augment conventional volatile memories, such as random-access memory (RAM). Unlike RAM, which loses stored information when the power is disconnected, eNVMs maintain data integrity during power interruptions and system shutdowns. This review examines a range of emerging memory materials and device architectures, including resistive random-access memories (ReRAMs), magnetic random-access memories (MRAMs), ferroelectric random-access memories (FeRAMs), and phase-change memories (PCMs). Additionally, novel eNVMs based on two-dimensional (2D) and organic materials are explored, along with a discussion of the transition from digital to synaptic computing and the potential it offers to address significant technological barriers that may impede the use of AI in accelerating discovery. The discussion encompasses a comprehensive analysis of technological advancements, current development trajectories, and the challenges that still need to be addressed.
存储技术是现代计算系统的核心，执行着从基本数据存储到高级任务等重要功能，比如用于人工智能（AI）和机器学习（ML）应用的内存计算。这些技术最初仅为数据保留而开发，如今正在不断演进以支持新的范式，如内存计算，即处理直接在内存阵列中进行。这种演进通过最小化处理器与内存之间的数据传输，显著提高了计算效率，从而实现了速度提升和能耗降低，这对于AI和ML工作负载来说是至关重要的因素。如此苛刻的要求正在推动超越传统互补金属氧化物半导体（CMOS）技术的创新。新兴的非易失性存储器（eNVM）代表了一类有前景的技术，旨在取代或增强传统的易失性存储器，如随机存取存储器（RAM）。与断电时会丢失存储信息的RAM不同，eNVM在电源中断和系统关闭期间能保持数据完整性。本文综述了一系列新兴的存储材料和器件架构，包括电阻式随机存取存储器（ReRAM）、磁性随机存取存储器（MRAM）、铁电随机存取存储器（FeRAM）和相变存储器（PCM）。此外，还探讨了基于二维（2D）和有机材料的新型eNVM，以及从数字计算向突触计算的转变及其在解决可能阻碍AI加速发现应用的重大技术障碍方面的潜力。讨论涵盖了对技术进步、当前发展轨迹以及仍需解决的挑战的全面分析。
