LAUSR

Memristors, characterized by in-memory computing and low-power consumption, are considered an ideal paradigm for building artificial neural networks and overcoming the von Neumann bottleneck. The two-terminal Li+-based memristor features simple structure and controllable weight update. However, existing works normally focus on the exclusive resistive switching layer, which is commonly the Li-source layer, and ignore the effect of another variable layer. In this study, a synchronous conductance modulation approach is developed by coupling the synchronously modulated layers of TT-Nb2O5 and LiCoO2 in the device. The linearity of the device was measured at 0.29, leading to a high recognition accuracy, with an average image recognition rate of 95.8% and a low standard deviation of 1.7%. This work offers an alternative option for developing two-terminal memristors.