Naveen 박사와 Malkesh 박사

Transparent semiconductor logic gate and memory technology using light signals 

A research team led by Prof. Kim Jun-dong of the Department of Electrical Engineering at Incheon National University (President Park Jong-tae) recently announced a neuromorphic AI computing technique using light. 

The human brain is composed of more than 100 billion neurons and 1,000 times more synapses. The human brain has so many nerve unit systems, but it performs bio-computing, such as detecting input signals from the human body, determining response movements, and remembering results with a low power of 20 W. 

[*Example of human reaction: When a person touches something hot with his or her hands, the nerve cells sense high temperature, generate a signal, send an 'occurrence signal' to the brain through a neural network, and order to quickly remove the object through immediate judgment. Then, remember information about the event.]

In this study, a transparent artificial neural network (neuromorphic device) operating based on light signals was developed. This neuromorphic device was developed as a transparent semiconductor, and it is characterized by self-producing electrical energy from light and operating on its own without supplying external power based on this. 

The characteristics of this transparent artificial neural network-neuromorphic device are characterized by logical operation, and the ON/OFF function can be expressed using light with different wavelengths. Various logic gates (AND, OR, NOT) have been implemented using light signals in place of existing electrical signals, and the characteristics of the transparent device can be applied to various fields such as artificial eyes, skin, and communication.

The newly developed transparent neuromorphic device can simultaneously perform memory performance (memory action) for a reaction along with the function of detection and judgment (logic gate) of a signal. Existing computing methods have problems such as overpower use and space constraints/density as stage work in which input, data processing, and memory functions are separated. However, integrated performance of no-power (signal detection-processing-memory) using light solved existing problems and proposed a new neuromorphic computing plan.

Dr. Naveen Kumar of India conducted this study through the Korea Research Foundation-Brain Pool Project (NRF-2021H1D3A2A02096147). He is currently a senior researcher at the National Research Council of Italy (CNR) and is conducting continuous cooperative research with this school. 

Prof. Kim Jun-dong, the head of the research, said, "This study uses the transcendence of transparent semiconductors that supply and operate energy by themselves. We plan to announce the development of more advanced technologies in the field of characteristic semiconductors and energy utilization beyond the existing framework." 

The results of this study were recently published in the latest issue of Applied Materials under the title of '2D-SnS-Embedded Schottky Device with Neurotransmitter-Like Functionality Produced Vapor Transfer Method for Photonic Neurocomputering'. The related technology has been registered in the first patent (10-2743896, transparent memristor device operating as an optical signal, registration date: December 12, 2024) and is proceeding with a patent application related to advanced technology.