ETRI-Knowledge Sharing Plaform

ENGLISH

성과물

논문 검색
구분 SCI
연도 ~ 키워드

상세정보

학술지 Multi-Level Long-Term Memory Resembling Human Memory Based on Photosensitive Field-Effect Transistors with Stable Interfacial Deep Traps
Cited 5 time in scopus Download 8 time Share share facebook twitter linkedin kakaostory
저자
김태윤, 임정욱, 윤선진, 이성현, 정광훈
발행일
202004
출처
Advanced Electronic Materials, v.66 no.4, pp.1-7
ISSN
2199-160X
출판사
Wiley
DOI
https://dx.doi.org/10.1002/aelm.201901044
협약과제
19HB1100, 차세대 신기능 스마트디바이스 플랫폼을 위한 대면적 이차원소재 및 소자 원천기술 개발, 윤선진
초록
The development of electronic devices that emulate the learning mechanism of the human brain is crucial for replicating the capabilities of the human brain. Photoelectric devices with learning capabilities are good candidates for input-energy-effective and multi-level memory devices. Herein, a novel TiO2-metaloxide-semiconductor field-effect transistor (MOSFET)-based memory device is reported, that effectively converts learned data to long-term memory (LTM) by longer stimulus intervals and realizes multi-state memory. This device is programmed by UV light and uses deep trap sites at the interface between the TiO2 and Al2O3 layers to retain data. Moreover, the current levels of LTM are determined by varying pulse width, light power density, and pulse number. The pulse interval is confirmed to be beneficial for converting input data to LTM without additional energy consumption. In particular, the unique characteristics of the device, which delivers a higher current with an increasing number of pulse intervals, describe the characteristics of the actual human brain, which is a feat that no other device has demonstrated. Furthermore, excellent multi-level storage capabilities and remarkable retention characteristics (≈30 000 s) of the TiO2?밠OSFET device are demonstrated. Thus, an advanced memory device that will contribute to next-generation artificial intelligence systems is realized.
KSP 제안 키워드
Deep traps, Energy-effective, Field-effect transistors(FETs), Long-term memory, Memory-based, Multi-state, Next-generation, Number of pulse, Power Density, Pulse number, State memory