2019年第80回応用物理学会秋季学術講演会

講演情報

一般セッション(口頭講演)

フォーカストセッション「AIエレクトロニクス」 » 31.1 フォーカストセッション「AIエレクトロニクス」

[20a-F211-1~10] 31.1 フォーカストセッション「AIエレクトロニクス」

2019年9月20日(金) 09:30 〜 12:15 F211 (レクチャーホール)

赤井 恵(阪大)

10:30 〜 10:45

[20a-F211-4] Emergent dynamics on Neuromorphic Nanowire Networks

〇(P)Adrian DiazAlvarez1、Rintaro Higuchi1、Zdenka Kuncic2、Tomonobu Nakayama1,3 (1.NIMS、2.Sydney Univ.、3.Tsukuba Univ.)

キーワード:Neuromorphic Materials, Memristors, Ag Nanowires

Neuromorphic materials are those in which there is a complex interconnectivity between the different parts of the material with functional units having dynamics similar to that of neurons on living species. These materials can be created with different aggregates of nanomaterials, as atomic switch or nanoparticle networks. In this work, we have created a neuromorphic network composed of polymer coated silver nanowires synthesized under solution, which can be dispersed onto a glass substrate producing a layout of randomly connected nanojunctions. We observe that, performing IV spectroscopy as well as acquiring time series under different biasing schemes, the system behaves as an electrical network composed of individual switches whose dynamics is similar to that of memristors. Using a computer modelization of such system, it will be shown how the interplay between the memristive properties of individual junctions and the topology of the electrical network formed by nanowires can account for its emergent properties, such as critical activation, short-term retention memory or adaptability to internal changes in the network.<br