Recently, fully optical-driving neuromorphic devices have attracted growing interest. However, the current device faces great challenges due to the lack of negative photoconductivity materials, which seriously hinders the further development of the visual perceptual applications using photosensitive device. Here, we propose an ionotronic neuromorphic InGaZnO <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">4</sub> phototransistor to establish a fully optical-driving artificial neural network. Optical neuronal paired-pulse facilitation can be switched to optical depression characteristics by tuning the gate bias more negatively through the ion-coupling bioelectrolyte. Moreover, the biological high-pass, low-pass, and band-stop photofilter behaviors can be successfully mimicked in such an all-in-one phototransistor. Finally, a fully optical-driving artificial neural network is constructed to perform artificial visual perception with an accuracy of ~90%. This device may open new avenues for the fascinating applications, such as artificial visual system, intelligent bionic robots, and smart photoelectric devices.
Read full abstract