Synaptic Properties\nof an Interfacial Memristor Based\non a Ga<sub>2</sub>O<sub>3</sub>/Nb:SrTiO<sub>3</sub> Heterojunction

Memristors have been extensively studied for tremendous\npotential\nfor future neuromorphic computing hardware applications because of\ntheir ability to imitate biological synaptic processes. Herein, we\nreport an interfacial memristor based on a Ga<sub>2</sub>O<sub>3</sub>/Nb:SrTiO<sub>3</sub> heterojunction that shows stable bipolar resistive\nswitching behavior, long retention time, and high switching ratio.\nThe conductance of the Au/Ga<sub>2</sub>O<sub>3</sub>/Nb:SrTiO<sub>3</sub>/In memristor can be gradually modulated under the voltage\nsweep mode as well as positive and negative pulse voltage stimulations,\nrespectively, thus realizing the long-term potentiation/depression\ncharacteristics of the simulated biological synapse. A neural network\nbased on the prepared memristor was built to recognize the handwritten\npicture data set with a recognition accuracy of 92.78% by using the\nNeuroSimV3.0 platform. Our work indicates that the Ga<sub>2</sub>O<sub>3</sub>/Nb:SrTiO<sub>3</sub> heterojunction memristor has significant\npotential in a neuromorphic computing system.