ScholarWorks@동국대학교

초록

The rapid growth of data-driven technologies highlights the inefficiency of von Neumann architecture for large-scale parallel processing. Neuromorphic computing offers a promising alternative using synaptic memory elements capable of analog modulation. Here, we report a surface-engineered CsPbBr3 perovskite nanowire (PNW) platform for high-performance memristors. By removing insulating ligands and introducing surface-selective iodide passivation, we suppress bromide-related defects and promote stable filament formation. The devices exhibit reliable bipolar switching, multilevel conductance, and robust potentiation/depression. These memristors further demonstrate brain-inspired functions including time-dependent plasticity, Pavlovian learning, and 91.6% recognition accuracy on the EMNIST data set. Additionally, the devices enable low-latency 4-bit edge computing, validating their potential for energy-efficient intelligent hardware. This work establishes colloidal CsPbBr3 PNWs as a scalable platform for next-generation neuromorphic applications.

키워드