Phosphorylation Enables Nano‐Graphene for Tunable Artificial Synapses

Abstract Flexible and robust memristors with controllable resistance‐switching characteristics are important to neuromorphic computing. However, the nanomaterials‐based, solution‐processed resistance switching layer usually has poor reliability and tunability due to uneven morphology and invariable surface properties. Herein, phosphorylated graphene nanoflakes (phos‐GPs) are synthesized for high‐performance solution‐processed flexible memristors. In situ conductive atomic force microscopy reveals that the tightly stacked uniform nanoflakes and modified phosphorate groups jointly reduce the formation barrier of the conductive filaments. Furthermore, phosphorylation gives rise to surface silver ion coordination leading to enhanced radial growth of the conductive filaments. The memristor shows volatile characteristics in the Ag/phos‐GPs/ITO architecture and exhibits non‐volatile properties in the Ag/Ag + ‐(phos‐GPs)/ITO structure. Both types of memristors display consistent I‐ ‐ V curves during long‐term cycling and under repetitive mechanical bending, in addition to excellent synaptic plasticity. Moreover, ultrasmall nonlinearity is observed from non‐volatile long‐term synaptic potentiation and depression. By utilizing the tunable artificial synapses, the processes of memory‐forgetting and re‐recognition are simulated, and the image recognition tasks are accomplished by the artificial neural networks.