Our recent research effort on developing transparent amorphous IGZO thin film transistors using ultrathin S/D electrodes has become published in Materials Science in Semiconductor Processing (IF: 4.644, top 26.993% journal in ENGINEERING, ELECTRICAL & ELECTRONIC) on October 23rd, 2022. This work was led by Yujin Hwang and Jungha Lee. Congratulations!
Title: a-IGZO thin-film transistors with transparent ultrathin Al/Ag bilayer source and drain for active neural interfaces
Abstract: Electro-optical neural interface technologies based on transparent electronics have been developed to take advantage of both high spatial resolution of optical cellular imaging by microscopes and high temporal resolution of electrical neural recordings. Recently, active-matrix systems based on amorphous indium gallium zinc oxide (a-IGZO) thin-film transistors (TFTs) have been suggested for high-density electrical neural recordings owing to the high mobility and optical transparency of the IGZO. However, conventional metallic source and drain electrodes forming ohmic contacts with the a-IGZO are the limiting factors of the overall system transparency because the electrodes have substantial portions in the TFT structures. In this work, we propose an ultrathin metal bilayer as the transparent source and drain electrodes of the a-IGZO TFTs for fully transparent TFT-based electro-optical neural interfaces. The ultrathin metal bilayer electrodes consist of 2.5–4.5 nm-thick aluminum as a seed layer and 8 nm-thick silver. The ultrathin Al/Ag bilayer suppressed localized surface plasmonic resonance, showing 40–70% of transmittance in the range of 350–700 nm and thus allowing us to observe cultured neurons through the transparent electrodes. We also confirmed that the ultrathin Al/Ag electrodes form decent ohmic contacts to the IGZO, showing width-normalized contact resistance of 2.9 kΩ-cm which is similar to the previously reported other transparent S/D electrodes. The a-IGZO TFTs with the ultrathin Al/Ag showed the highest linear mobility of 3.9 cm2/Vs of in long channels and 0.2–0.64 cm2/Vs in short channels. The gate-bias normalized cut-off frequency was 8–26 kHz/V at Lch = 20 µm, which would be sufficient for recording a wide range of neural signals.