Synaptic Properties of an Interfacial Memristor Based on a Ga2O3/Nb:SrTiO3 Heterojunction

Memristors have been extensively studied for tremendous potential for future neuromorphic computing hardware applications because of their ability to imitate biological synaptic processes. Herein, we report an interfacial memristor based on a Ga₂O₃/Nb:SrTiO₃ heterojunction that shows stable bipolar resistive switching behavior, long retention time, and high switching ratio. The conductance of the Au/Ga₂O₃/Nb:SrTiO₃/In memristor can be gradually modulated under the voltage sweep mode as well as positive and negative pulse voltage stimulations, respectively, thus realizing the long-term potentiation/depression characteristics of the simulated biological synapse. A neural network based on the prepared memristor was built to recognize the handwritten picture data set with a recognition accuracy of 92.78% by using the NeuroSimV3.0 platform. Our work indicates that the Ga₂O₃/Nb:SrTiO₃ heterojunction memristor has significant potential in a neuromorphic computing system.