德拉姆
可扩展性
计算机科学
布线(电子设计自动化)
集成电路
晶体管
过程(计算)
电子工程
动态随机存取存储器
电气工程
缩放比例
工程类
过程集成
节点(物理)
电容器
纵向一体化
极限(数学)
频道(广播)
电路设计
平版印刷术
逻辑门
在制品
等级制度
内存层次结构
材料科学
嵌入式系统
重构代码
相变存储器
电容感应
双极结晶体管
工作(物理)
存储单元
CMOS芯片
蚀刻(微加工)
集成电路设计
随机存取
电子线路
作者
K.J. Lee,Sungwon Cho,Po-Kai Hsu,Janak Sharda,Suman Datta,Shimeng Yu
标识
DOI:10.1109/ted.2026.3659106
摘要
This article provides a framework for understanding the state-of-the-art challenges and prospects of monolithically stackable one-transistor–one-capacitor (1T1C) 3-D dynamic random access memory (DRAM). It begins by outlining the limits of 2-D DRAM scaling and the need for monolithic vertical integration, and then examines cell architectures comparing vertical bitline (VBL) and horizontal bitline (HBL) layouts while analyzing charge-loss mechanisms viaoff-state leakage, floating-body-induced parasitic bipolar junction transistor (BJT) activation, and word-line (WL) capacitive coupling. Subsequently, this article reviews access-transistor designs and channel material options from silicon to amorphous semiconductors, along with compatible capacitor geometries and process integration issues. Then, this article extends to array and periphery architectural choices, analyzing WL/BL routing, selector (SL)-assisted sensing, and comparing integration schemes such as CMOS-near-array (CNA), CMOS-under-array (CUA), and CMOS-bonded-array (CBA). The analysis clarifies how routing parasitics, bonding pitch, and thermal budget jointly define the scalability limit of 3-D DRAM. By linking device physics, process integration, and architectural hierarchy into a cohesive framework, this work defines the technological pathways for practical monolithic 3-D DRAM technology for the next decade.
科研通智能强力驱动
Strongly Powered by AbleSci AI