Bifunctional Memristive Behavior of a Dual-Layer Structure Depending on the Configuration of Charge Transport

Abstract

This paper presents a method for integrating neuronal and synaptic functions within a thin dual-layer featuring distinct dielectric strengths. The dual-layer consists of a conductive bottom layer (e.g., MXenes or rGOs) and a top layer with a lower dielectric strength (e.g., gallium oxide). The differing dielectric strengths between the layers facilitate the modulation of breakdown, as the magnitude of the electric field applied in one layer varies with the configuration of charge transport in the other layer. In a vertical configuration, the dual-layer exhibits volatile and abrupt switching (neuronal behavior), while in a horizontal configuration, it demonstrates non-volatile and gradual changes in conductance (synaptic behavior). The experimental results indicate that the abrupt switching is attributed to filament formation, while the gradual change in conductance arises from charge transport in gallium oxide. The dual-layer shows the characteristics of integrate-and-fire depending on spiking signals with synaptic plasticity and achieves training accuracies of 91.4% and 82.3% for MNIST digit classification based on MXene and rGO, respectively.