Ternary two-dimensional (2D) material-based memristors have garnered significant attention in the fields of machine learning, neuromorphic computing due to their low power consumption, rapid learning, and synaptic-like behavior. Although such memristors often exhibit high ON/OFF ratios and exceptional pulse response characteristics, they have also to face some challenges concerning reusability and switching cycles, which arise from the filament instability issues.. Here we propose a modulation strategy to improve performance of 2D-material memristors with synaptic and flexible features. By laser-modulating few-layer FePS3, we induced the formation of conductive filaments, realized a major improvement in performance of the FePS3 memristors, achieving an on/off ratio of nearly 104, low power consumption at approximately 10-7 W of single switching operation, and maintaining stability even after over 500 cycles. The performance promotion has been ascribed to enhancement of conductive filament induced by laser-modulation. Furthermore, we have identified the effectiveness of our laser modulation under strain by building the high-performance flexible FePS3 memristor. Meanwhile, we discovered a novel strain-dominant erasure method for the flexible memristors. Our work confirms that laser modulation is a viable method for enhancing the performance of 2D material-based memristive devices.
.
Read full abstract