High-performance multi-level memristor based on Ag NPs-doped MXene for reliable artificial synapse

Memristors has been considered as a promising candidate for synaptic devices in next-generation neuromorphic computing. The two-dimensional MXene attracts considerable interest for use in memristor devices owing to its excellent physicochemical properties. However, MXene-based devices still suffer from unsteady conductive filaments and significant degradation. In this work, we present a highly-robust multilevel resistive switching memristor based Ti 3 C 2 T x by alkalization synthesis process and incorporate doped Ag nanoparticles into the active layer. This memristor exhibits a significant improvement in the uniformity of device parameters, including low threshold voltages, high switching ratio, high endurance, and long data retention compared to conventional memristors. The Ag nanoparticles embedded in oxidized Ti 3 C 2 T x renders a reduced variability in threshold voltages from 33 % to 8 %. Additionally, a remarkable accuracy (∼92.69 %) is achieved on the handwritten digits dataset by constructing an artificial neural network based on a memristor array. Overall, the proposed effective strategy could advance the performance of future manufactured MXene-based memristors.