Synergistic multi-wavelength optical stimulation enhances synaptic dynamics and reservoir computing performance in ferroelectric thin-film transistors

초록

Ferroelectric thin-film transistors (FeTFTs) with light-sensitive indium gallium zinc oxide channels are promising neuromorphic devices capable of integrating sensing, memory, and processing functionalities. In this work, we present an Indium-Gallium-Zinc-Oxide (IGZO)-based FeTFT that synergistically utilizes electrical and multi-wavelength optical stimuli to emulate a broad spectrum of synaptic and neural behaviors. The device, fabricated with a Ta/IGZO/Hf<inf>0.5</inf>Zr<inf>0.5</inf>O<inf>2</inf>/TiN back-gate stack, exhibits robust ferroelectric characteristics and nonvolatile memory behavior through polarization switching. Under optical illumination at 405, 450, and 520 nm, the device demonstrates distinct short-term memory dynamics, including paired-pulse facilitation, learning–forgetting–relearning processes, and nociceptive responses such as allodynia and hyperalgesia. Moreover, the FeTFT performs light-driven logic operations and mimics classical Pavlovian conditioning using purely optical inputs. These diverse behaviors are leveraged to implement a physical reservoir computing system. Using multi-wavelength optical stimulation, the FeTFT generates well-separated reservoir states, significantly enhancing recognition accuracy in a 4-bit image classification task. This device demonstrates the highest Fashion Modified National Institute of Standards and Technology recognition rate of 85.53 % was achieved under dual-wavelength stimulation (405 nm and 450 nm), confirming the effectiveness of spectral engineering in optimizing reservoir state separability. This study highlights the potential of optoelectronic FeTFTs as multifunctional building blocks for energy-efficient, brain-inspired vision systems. © 2025 Elsevier B.V., All rights reserved.

키워드