세미나

Rise of neuromorphic computing architectures:

Accelerated Learning in wide-Band-Gap AlN artificial optoelectronic synaptic devices

 

Moonsang Lee,^1,*

 

¹Department of Materials Science and Engineering, Inha University, 100 Inha-ro, Michuhol-gu, Incheon 22212, Republic of Korea

 

Abstract

We briefly introduce the rise of neuromorphic computing architectures for deep learning applications. After the overview, we will discuss futuristic synaptic devices to overcome the limitation of the state-of-the art electronic devices. Indeed, artificial synaptic platforms are promising for next-generation semiconductor computing devices; however, state-of-the-art optoelectronic approaches remain challenging, owing to their unstable charge trap states and limited integration into modern digital architectures. We demonstrate wide bandgap (WBG) III-V materials as promising building blocks for photo-electronic neural networks. Our experimental analysis clearly shows that enhanced crystallinity of WBG synapses promotes better synaptic characteristics, such as effective multilevel states, wider dynamic range, and linearity, which lower the power consumption, improve training, and boost the recognition accuracy of artificial neural networks (ANNs). Furthermore, light frequency-dependent memory characteristics suggest that artificial optoelectronic synapses with improved crystallinity support the transition from short-term potentiation (STP) to long-term potentiation (LTP), implying a clear emulation of the psychological multi-storage model. This is attributed to suppressed shallow trap levels in WBG synapses, which encourages charge trapping in deep-level states and suppresses fast decay and non-radiative recombination in shallow traps. We believe that the fingerprints of these WBG synaptic characteristics provide an effective strategy for establishing an artificial optoelectronic synaptic architecture for innovative neuromorphic computing.