講演情報
[8p-E217-13]Physical Modeling of Filament Formation and Dissolution in Silver Nanoparticle Memristors
〇(D)ALBERT NTUMBA1, Kouki Kimizuka2, Yuki Usami1,2, Hirofumi Tanaka1,2 (1.LSSE, Kyushu Institute of Technology, 2.Neumorph Center, Kyushu Institute of Technology)
キーワード：
Memristors、Silver Nanoparticles、Reservoir Computing
Silver nanoparticle resistive memories operate as Electrochemical Metallization Memory (ECM) devices in which voltage driven Ag filament formation and dissolution generate characteristic current voltage hysteresis. This work presents a physics based compact model inspired by the TEAM and VTEAM frameworks to describe filament dynamics in Ag/Ag2S nanoparticle networks. In the single junction approach, the state variable represents the normalized filament volume fraction and evolves according to a current dependent kinetic equation, while conductance follows a nonlinear exponential relationship. The model is extended to multiple junctions, where each nanoparticle gap possesses an independent state and resistance, enabling self consistent coupling between filament growth and electrical transport. Model parameters are identified using Differential Evolution and Nelder Mead optimization. The single junction model reproduces the overall switching behavior with an RMSE of 0.700 uA, whereas the four junction model improves the fit to 0.385 uA and accurately captures SET, ON state, and RESET transitions. The results demonstrate that distributed junction dynamics significantly enhance simulation accuracy and provide a reliable framework for modeling AgNP based memristive devices.