Visible-light modulated ferroelectric optoelectronic field-effect transistor for superior neuromorphic computing

High-performance neuromorphic computing and bionic vision hardware urgently require device simplicity and low power consumption. This paper adopts an all-solution method to prepare a visible-light modulated ferroelectric optoelectronic field-effect transistor (Fe-OFET) based on a TiO 2 /IGZO/Al 2 O 3 /PZT structure, used to achieve visible-light-driven neuromorphic synapse simulation and multilevel storage. By introducing a TiO2 layer, the light current of the IGZO-based device is extended to the red light band and the absorption efficiency in the entire visible-light wavelength is substantially improved, overcoming the bottleneck that traditional oxide semiconductors only respond to ultraviolet light and are difficult to utilize visible light. The device exhibits stable ferroelectric non-volatility and typical optoelectronic performance, successfully simulating core synaptic plasticity such as excitatory postsynaptic current (EPSC) and paired-pulse facilitation (PPF). A convolutional neural network (CNN) constructed based on this device achieves 95.5% accuracy in a target motion trajectory recognition task; a hybrid artificial neural network (ANN) integrating multi-head attention (MHA) achieves a maximum accuracy of 90.0% in the MNIST handwritten digit recognition task. This study provides a process-simple, visible-light-compatible ferroelectric optoelectronic field-effect transistor solution, providing important support for visual neuromorphic hardware and neuromorphic computing integration.