Steady, Flexible Memristor with Lead-Free Perovskite for Bionic Perception

The advancement of artificial intelligence and cloud computing aspires multifunctional components to face complex problems and bionic perception. As one of the emerging neural synapse devices, the perovskite memristor has great potential in information storage and brain-like learning due to its merits, such as low energy consumption and quick processing. Herein, the PET/Graphene/Cs₂AgBiBr₆/Ag flexible memristor has been successfully fabricated via low-temperature processing (∼100 °C), exhibiting pronounced resistive switching characteristics (ON/OFF ratio: ≈10²), long-term environmental stability (≥8 months), and excellent mechanical bending performance (600 bending cycles). The synaptic plasticity in the device is also verified, including paired pulse facilitation (PPF) and spike-timing-dependent plasticity (STDP). By applying resistance memory characteristics, the memristor achieves fear emulation, enabling the realization of key characteristics including "self-extinction", "generalization", and "avoidance". Finally, sound perception is simulated via imposing different signals controlled by different voltage parameters (frequency and amplitude) and light illuminance. Sound characteristics including "pitch", "loudness" and "timbre" are successfully determined. Our work proposes a new application avenue for lead-free flexible perovskite memristors in multimodal intelligent perception and biological simulation.