记忆电阻器
钙钛矿(结构)
材料科学
纳米技术
突触
光电子学
电气工程
化学工程
工程类
生物
神经科学
作者
Radharani Yumlembam,Himangshu Jyoti Gogoi,Roy P. Paily,Arun Tej Mallajosyula
标识
DOI:10.1088/2058-8585/adacf9
摘要
Abstract Von Neumann bottleneck necessitates the creation of dedicated processors for neuromorphic artificial intelligence tasks such as in-memory computing, where memristors are formulated as synapses. Perovskites are great candidates for memristors owing to their mixed ionic–electronic conduction and their compatibility with cost-effective processing techniques. In this work, we have fabricated a highly stable, lead-free, and flexible perovskite memristors by e-beam evaporation of hot-pressed zero-dimensional (0D) Cs3Bi2I9 pellets. These memristors exhibit reproducible bipolar resistive switching with low operating voltages of −0.18 V and 0.26 V, an excellent ON/OFF ratio (>105), and high endurance (>104 cycles). They were air-stable for more than 30d and were repeatedly tested under high humidity (>80%) atmospheric conditions without encapsulation. The resistive switching in these devices persists even under applied mechanical stress up to a bending radius of 5 mm. A 4 × 4 crossbar array of these Cs3Bi2I9 memristors has been fabricated, which gave a device yield of 81%. Furthermore, their potential for use as artificial synapses has been demonstrated by obtaining critical neuromorphic characteristics such as spike duration dependent plasticity, paired pulse facilitation, and long-term plasticity. This work also shows that 0D Cs3Bi2I9 memristors have the potential to mimic biological synaptic functions of learning and forgetting, which may be useful in realizing flexible and low-power neuromorphic circuits in the near future.
科研通智能强力驱动
Strongly Powered by AbleSci AI