[20p-231B-1~14] 12.2 Characterization and Materials Physics

△ [20p-231B-9] Electrode dependence of conductive polymer wire as a resistance change element

〇Masaru Okada¹, YAsumasa Sugito¹, Tetsuya Asai², Yuji Kuwahara¹, Megumi Akai^1,3 (1.Osaka University, 2.Hokkaido University, 3.JST PRESTO)

Keywords:neuromorphic, neural network, conductive polymer

A brain-inspired technology Artificial Neural Network(ANN) enables us to realize functions of -Artificial Intelligence(AI), e.g, image recognition, auto-driving and other various fields. ANN is implemented on software but the calculation efficiency is not good because of framework in von Neumann type computer. So dedicated ANN hardware is now expected to realize a low-power-consumption AI. Here, we offer to use conductive polymer PEDOT:PSS [poly (3,4-ethylenedioxythiophene) doped with poly (styrene sulfonate) ] as a resistance change element. monomer EDOT and electrolyte PSS solution, PEDOT:PSS wire grows between Au electrodes through electrolytic polymerization. PEDOT:PSS wires growth is selective about multitude of electrodes, and conductance values between electrodes increase when the PEDOT:PSS wire multiply bridges. Property of the wire growth seriously depends on frequency of growth voltage and figure of electrode’s edge. In this study, we tried to find the best curvature radius at front edge of electrodes to perform a desirable resistance change elements. The results suggest that PEDOT:PSS wire can play an important role for construction of non-von Neumann ANN hardware.