A novel artificial synapse with dual modes using bilayer graphene as the bottom electrode. Issue 27 (28th June 2017)

• Record Type: Journal Article
• Authors: Tian, He; Mi, Wentian; Zhao, Haiming; Mohammad, Mohammad Ali; Yang, Yi; Chiu, Po-Wen; Ren, Tian-Ling
• Is Part Of: Nanoscale. Volume 9:Issue 27(2017)
• Page Start: 9275
• Page End: 9283
• Publication Date: 2017-06-28
• Subjects: Nanoscience -- Periodicals; Nanotechnology -- Periodicals; 620.505
• DOI: 10.1039/c7nr03106h
• Languages: English
• ISSNs: 2040-3364

Abstract

Abstract : We report a single device operated in two modes with enough potentiation states to mimic the synaptic learning behaviors. Abstract : Resistive Random Access Memory (RRAM) shows great potential to be used as an artificial synapse for neuromorphic applications. The resistance can be gradually reduced during reset, which can enable enough states to mimic the "forgetting" process. However, the abrupt set (Mode I) cannot generate enough states to mimic the "learning" process, which results in depression-only behavior. In this work, we introduce another mode (Mode II) in an Al/AlO x /graphene 'RRAM' stack by using oxygen vacancies as trapping centers and bottom electrode bilayer graphene as the channel material. In this way, since the pulse can gradually create the oxygen vacancies, post-synaptic current (PSC) can be gradually potentiated or depressed. By introducing Mode II, we can realize 166 potentiation states, which is higher than the previously reported conventional RRAM with insufficient potentiation states due to the abrupt set. Moreover, Mode II can help realize an inhibitory synapse. By combining modes I and II, we can realize both excitatory and inhibitory synapses in a single device. This work shows great potential to enable neuromorphic computations with greater learning and reconfigurability.