ReconfigurableNeuromorphic Computing Using Methyl-EngineeredOne-Dimensional Covalent Organic Framework Memristors

The rapid evolution of neuromorphic devices seeks to bridge biological neural networks and artificial systems, enabling energy-efficient and scalable computing for next-generation artificial intelligence. Herein, we introduce methyl-engineered one-dimensional covalent organic framework (1D COF)-based memristors as a transformative platform for reconfigurable neuromorphic computing. The incorporation of methyl groups enhances localized polarization effects within the COF framework, effectively mitigating random Ag⁺ migration/diffusion and stabilizing conductive filament morphology. This strategic modification yields devices with exceptional multilevel storage capabilities, exhibiting superior stability, linearity, and reproducibility. Moreover, the highly ordered architecture and customizable chemical environment of the methyl-functionalized 1D COF allows for precise control over resistive switching behaviors, facilitating the emulation of synaptic functions and the development of artificial neural network architectures. Demonstrating exceptional performance in neuromorphic tasks such as high-accuracy image recognition, these devices showcase significant promise as the foundation for energy-efficient, next-generation neuromorphic computing systems.