Visible Light Wavelength-Dependent Erasing in AOS-Based Charge Trap TFTs for Enhanced Neuromorphic Display Performance

Abstract

Amorphous oxide semiconductor (AOS)-based charge trap thin-film transistors (CTTFTs), which are well-suited for 3D vertical stacking process, have emerged as a promising solution for data-intensive neuromorphic applications. However, AOS-based CTTFTs face challenges in erasing operations due to insufficient hole density in AOS channels. To address this issue, we investigated the impact of the photoelectric effect on the erasing operation of amorphous indium tin zinc oxide (a-ITZO)/amorphous indium gallium zinc oxide (a-IGZO) CTTFTs under illumination at various wavelengths of 650 nm (red), 532 nm (green), and 405 nm (blue). We successfully demonstrated the fully erased state of a-ITZO/a-IGZO CTTFTs under blue light, which has photon energy approximately equal to the a-ITZO/a-IGZO bandgaps. Additionally, partially erased states were observed under green and red light, attributed to the involvement of subgap states and band-tail states. The synaptic properties of a-ITZO/a-IGZO CTTFTs under blue light presented excellent linearity with nonlinearity factor (alpha p/alpha d) = 0.43/-0.72 and a wide dynamic range with a conductance on/off ratio (G max/G min) = 483.9, demonstrating a high pattern recognition accuracy of 93.44% in simulations using the MNIST data set. These results highlight the significant potential of a-ITZO/a-IGZO CTTFTs for integrating low-power neuromorphic systems with display technologies by leveraging various light sources in the visible wavelength.