Electroelasticity-Regulated Vacancy Diffusion Enables Frequency-Adaptive Neuromorphic Perception and Control

Abstract Emulating frequency-dependent synaptic plasticity requires artificial devices capable of integrating temporal inputs and adapting their responses. Here, we demonstrate electroelasticity-regulated vacancy diffusion in MoSe2 memristors as a physical mechanism for frequency-adaptive synaptic dynamics. An applied electric field induces lattice deformation that lowers the vacancy diffusion barrier, whereas poststimulus electroelastic relaxation progressively restores the barrier, creating a dynamically evolving diffusion landscape. This coupled electroelastic–ionic process gives rise to intrinsic second-order memristive dynamics and enables frequency- and history-dependent plasticity, including pair- and triplet-spike-timing-dependent plasticity and Bienenstock–Cooper–Munro-type learning with a sliding modification threshold. Leveraging these dynamics, memristor-based spiking neural networks realize adaptive neuromorphic functions spanning composite color–direction perception and autonomous drone control. These results establish electroelastic regulation of vacancy diffusion as a materials-level mechanism for translating temporal input statistics into adaptive synaptic responses, providing a pathway toward neuromorphic systems with frequency-adaptive perception and control.