A joint research team from Hanyang University and Dankook University has successfully developed a next-generation AI semiconductor—known as a neuromorphic semiconductor—that learns using light, mimicking the human brain. Professors Yoo Ho-cheon and Oh Se-yong of Hanyang University, together with Professor Choi Joon-hwan of the Department of Chemical Engineering at Dankook University, announced on August 6 that they had implemented a “light-signal-based neuromorphic semiconductor” that functions even on paper-thin substrates.

This newly developed neuromorphic semiconductor is designed to imitate the structure of the human brain’s neural network. Unlike conventional devices that rely on electrical signals, it uses light to simultaneously process data and perform learning functions. By integrating processing units (CPUs) and memory into a single chip, it significantly reduces power consumption while rapidly handling large amounts of data, making it a promising candidate for next-generation semiconductors. The research team applied a dual-insulator structure and polymer material to the transistor device, enabling the capture and storage of charges generated in response to light. The technology also supports low-temperature fabrication using gas-phase room-temperature processes, allowing implementation on flexible substrates like paper business cards.

This achievement is expected to expand into various industries—including wearable devices, AI sensors, and security platforms—thanks to its advancement in low-power, high-efficiency signal processing and learning devices. The research findings were published online in July in npj Flexible Electronics (2024 IF = 15.5), a top 2% international journal in the field of electrical and electronic engineering.