Light-stimulated synaptic transistor based on two-dimensional/organic heterojunction for neuromorphic computing

Neuromorphic computing has considerable potential in simulating the efficient information processing capabilities of the human brain. Implementing neuromorphic computing requires the development of artificial synaptic devices that mimic biological synapses, which are the basic for information processing, storage, and transmission in biological neural networks. Herein, we demonstrate a light-stimulated synaptic transistor (LSST) device based on graphene/hexagonal boron nitride (h-BN)/pentacene heterojunction for emulating the basic functions of the human brain. The LSST devices can detect light stimulus at a wavelength of 520 nm and exhibits a variety of typical synaptic properties, such as excitatory postsynaptic current, paired pulse facilitation, and transition from short-term memory to long-term memory. In addition, the LSST device is capable of simulating the learning and forgetting processes of the human brain. Based on the optically and electrically controlled conductivity modulation characteristics of the LSST device, we construct an artificial neural network for perform pattern recognition tasks, and recognition accuracy of handwritten digits is 88.5%. These results mean that our LSST devices have great potential for future applications in neuromorphic computing.