可重构性
冯·诺依曼建筑
计算机科学
晶体管
计算机体系结构
逻辑门
神经形态工程学
可编程逻辑器件
可重组计算
还原(数学)
德拉姆
电子工程
逻辑族
嵌入式系统
非易失性存储器
材料科学
非常规计算
记忆电阻器
隐藏物
现场可编程门阵列
灵活性(工程)
和大门
逻辑综合
加法器
计算机硬件
晶体管计数
内存处理
通流晶体管逻辑
编码(内存)
编码(社会科学)
微电子
微处理器
宏单元阵列
电容
电气工程
可逆计算
作者
Haoyue Lu,Yan Wang,Hao Sun,Jing Liu
标识
DOI:10.1002/adfm.202525471
摘要
ABSTRACT The rapid expansion of artificial intelligence (AI) and other data‐driven applications demands energy‐efficient and compact computing architectures that overcome the limitations of conventional von Neumann systems. 2D semiconductors offer promise for reconfigurable logic, but existing designs face a trilemma between integration density, functionality, and operational stability. Here, we present a dual‐gate field‐effect transistor based on a graphene/h‐BN/MoTe 2 /h‐BN heterostructure that unifies four reconfigurable logic operations (AND, OR, NAND, NOR) and nonvolatile n‐/p‐type memory within a single device. All logic operations are achieved under an identical back‐gate voltage, enabled by synergistic optoelectronic modulation of MoTe 2 polarity, yielding a 90 % reduction in transistor count compared to conventional CMOS. The device demonstrates robust reconfigurability through more than 24 full cycles of direct, arbitrary switching among four distinct polarities without intermediate states. Dynamic switching between logic and memory modes is controlled by the graphene top gate, where memory operation delivers a 10 4 on/off ratio and >1000 s retention. Finally, we demonstrate logic‐in‐memory computing and image processing tasks, underscoring the device's potential for next‐generation energy‐efficient computing architectures.
科研通智能强力驱动
Strongly Powered by AbleSci AI