Mechanically Gated Vertical Ion Channels for Fast Strain-Sensitive Neuromorphic Memristor.

Abstract

Integrating sensing functions into memristors is promising to realize in-sensor computing with unpreceded energy efficiency and minimized latency. Strain-sensitive memristors gradually draw attention in neuromorphic tactile sensing applications, but still face the sensitivity-response time tradeoff dilemma. Here, we demonstrate that antiphase boundaries (APBs) in thin-film structures function as mechanically gated vertical ion channels in a strain-sensitive neuromorphic memristor, which can achieve high sensitivity (strain gauge factor of 1.7 × 10⁴) and rapid response (≤3 ms) while exhibiting synaptic plasticity. Through atomic-scale scanning transmission electron microscopy (STEM), electron energy loss spectroscopy (EELS), and nanoscale conducting atomic force microscopy (C-AFM), we confirm APBs as preferential oxygen vacancy migration paths, whose conductivity is dynamically modulated by electrical and mechanical stimuli synchronous and directly. Excellent electrically (25500% on/off ratio at -7 V) and mechanically tunable conductive (enhanced 15167% with ∼3.3 µN force) behavior along APBs pillars is clearly observed. While synapse-like information processing functions are also further demonstrated with APBs pillars, and a high image classification accuracy (97.7%) within 100 learning epochs is achieved in a two-level artificial neural network via simulation. This work establishes a potential pathway for integrated sensing and computing systems for next-generation intelligent robotics and adaptive prosthetics.