Ultrasensitive Pa-Level Persistent Mechanoluminescent Material Toward All-Optical Neural Synapses for Tactile-Visual Information Recognition and Memory.

Abstract

Mechanoluminescence (ML), a self-recovering and passive luminescent modality, offers a promising path toward tactile-visual all-optical neuromorphic computing, potentially overcoming the inefficiency of von Neumann architecture. However, existing ML materials are hindered by high response thresholds and single-mode luminescence, preventing sub-kPa perception and multilevel neural transmission. Here, we employ Li⁺/Dy³⁺ co-doping in Sr₂SiO₄:Eu²⁺ (LSSO) to implement a defect engineering strategy that synergistically optimizes oxygen vacancies and suppresses strontium vacancies, achieving dual breakthroughs in sensitivity and signal clarity. This approach yields a record-low ML threshold of 72 Pa-the only Pa-level system achieved without external electricity or elastomeric structural modifications. This material also responds to sunlight, force, and heat, emulating diverse synaptic functions like tactile/optic nerve perception, short-term potentiation, and memory. It exhibits a 7-s persistent ML with a signal-to-noise ratio of 20.57 which is 15.6 times higher than commercial SrAl₂O₄:Eu²⁺,Dy³⁺, a micron-scale imaging resolution (≈200 µm), and a 36-hour memory capacity. These properties enable thermal-activated information awakening and visual imaging over 1000 cycles, with a memory accuracy 209% superior to the Ebbinghaus curve. This work not only advances the design of all-optical synapses but also forges a pivotal connection between ML and neuromorphic engineering, propelling energy-efficient, light-driven artificial intelligence.