Dual-input optoelectronic synaptic transistor based on amorphous ZnAlSnO for multi-target neuromorphic simulation

Abstract

Optoelectronic synapses can perceive both optical and electrical signals, which are critical for the realization of neuromorphic computing. We have rationally designed an optoelectronic synaptic transistor based on amorphous ZnAlSnO for multi-target neuromorphic simulation and recognition. The dual-input models are well operated by applying light pulses on the channel and electric pulses on the gate, and the transformation from short-term potentiation (STP) to long-turn potentiation (LTP) is identified for tunable synaptic plasticity. In the electrical operation mode, a single-layer artificial neural network was established to recognize handwritten digits by LTP/LTD (long-turn depression) modulation, with a recognition accuracy of 89.2 % for the actual device. In the optical operation mode, the processes of repetitive learning, image recognition, and biased/correlated random-walk learning are simulated on the basis of frequency, quantity, and power of light, with an energy consumption per event as low as 4.3 pJ. This work will facilitate the development of future artificial synapses and highlights the potential of amorphous oxide semiconductors for next-generation computer hardware applications.