Premium
Self‐Selective Organic Memristor by Engineered Conductive Nanofilament Diffusion for Realization of Practical Neuromorphic System
Author(s) -
Park HeaLim,
Kim MinHwi,
Kim Hyungjin,
Lee SinHyung
Publication year - 2021
Publication title -
advanced electronic materials
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.25
H-Index - 56
ISSN - 2199-160X
DOI - 10.1002/aelm.202100299
Subject(s) - memristor , neuromorphic engineering , materials science , realization (probability) , nanotechnology , electronics , computer science , electronic engineering , artificial neural network , electrical engineering , artificial intelligence , engineering , statistics , mathematics
Abstract Solution‐processed organic memristors are promising ingredients to realize smart wearable electronics including neural networks. In organic memristors, tunable functionality of materials allows for realizing bio‐realistic neuromorphic electronics in the view point of the mechanical and electrical characteristics. However, it is challenging to achieve high‐density crossbar arrays of organic memristors due to undesirable sneak currents arising from unselected cells. For inorganic systems, considerable effort has been made to fabricate practical arrays by employing external components to suppress sneak current. By contrast, in organic memristors, it is barely possible to achieve practical systems due to the solvent orthogonality limiting the integration of the devices. Herein, an unprecedented structure of organic memristors with high self‐selectivity is developed to realize practical crossbar arrays. In the developed memristor, the self‐selective characteristics are achieved by systematically engineering the conductive nanofilament diffusion in the polymer. The maximum size of the memristor arrays is found to be more than 1 Mbits, and the neural networks based on the developed device showed reliable recognition performance similar to ideal software systems. This novel concept of developing the organic memristor with high self‐selectivity will open a new platform for realizing next‐generation flexible memory and practical neuromorphic systems linked to artificial intelligence.