Synaptic a-Si:H/a-Ga2O3 phototransistor inspired by the phototaxis behavior of organisms with all-optical and all-electrical stimulation modes

Youngbin Yoon,Youngki Kim,Soowon Choi,Jungdae Kwon,Myunghun Shin
DOI: https://doi.org/10.1007/s12274-024-6754-7
IF: 9.9
2024-06-19
Nano Research
Abstract:To improve neuromorphic computing performance, neuromorphic system components should mimic the behaviors of organic systems. In this study, a synaptic a-Si:H/a-Ga 2 O 3 phototransistor featuring all-optical and -electrical emulation is fabricated in a manner advantageous for complementary metal-oxide-semiconductor process integration. The phototransistor exhibits excitatory and inhibitory synaptic behaviors under stimulation by both optical and electrical signals. It mimics several essential synaptic functions, including excitatory postsynaptic current, inhibitory postsynaptic current, short-term memory, long-term memory, paired-pulse facilitation, and spike-timing-dependent plasticity. The optical and electrical modulation mechanisms are confirmed to arise from the a-Si:H/a-Ga 2 O 3 heterojunction structure and interface effects, and the device is shown to operate at low power in both optical and electrical modes. The all-optical weight modulation function is applied to the wavelength-differential behavior response of zebrafish, successfully mimicking the color perception process of the organism. Finally, to verify the translation of the optoelectrical-derived synaptic behaviors of the phototransistor into artificial neuromorphic computation, handwritten digit image recognition of the Modified National Institute of Standards and Technology dataset is performed by a convolutional neural network, with a demonstrated average learning accuracy of 98.46%. These findings verify the applicability of the synaptic a-Si:H/a-Ga 2 O 3 phototransistor in neuromorphic computing.
materials science, multidisciplinary,nanoscience & nanotechnology,physics, applied,chemistry, physical
What problem does this paper attempt to address?