神经形态工程学
材料科学
横杆开关
突触后电流
兴奋性突触后电位
光电子学
计算机科学
神经科学
人工神经网络
抑制性突触后电位
人工智能
电信
生物
作者
Yishu Zhang,Lin Wang,Hao Chen,Teng Ma,Xiufang Lu,Kian Ping Loh
标识
DOI:10.1002/aelm.202100609
摘要
Abstract The emergence of 2D ferroelectrics offers a promising path to implement next‐generation information technology, including digital memory and analog computing. Here, it is demonstrated that digital or analog memory can be achieved in a single crossbar type two‐terminal (Au‐Ti) α‐In 2 Se 3 device depending on the driving voltage. The analog operation is enabled by ferroelectric polarization‐modulated Schottky barrier, while resistive filament switching drives the digital operation. By tuning ferroelectric properties, multiple analog conductance states can be obtained for mimicking various synaptic behaviors like excitatory postsynaptic current, inhibitory postsynaptic current, potentiation/depression and spiking‐timing‐dependent plasticity. A simulated neural network built from these synaptic devices shows good on‐line learning accuracy (≈93.2%) for digital recognition of handwriting. Based on filament formation/rupture mechanism, α‐In 2 Se 3 devices also exhibits excellent digital memory performance with large on/off ratio (>10 3 ), high on‐current density (10 5 A cm −2 ) and fast switching speed (≈10 ns). The combination of analog and digital memory modes in two‐terminal α‐In 2 Se 3 devices is useful in highly dense and complex electronics.
科研通智能强力驱动
Strongly Powered by AbleSci AI