Flexible, Lead-Free (C ₆ H ₅ CH ₂ NH ₃ ) ₂ CuBr ₄ Optoelectronic Memristors for Multimodal Synaptic Emulation and Neuromorphic Learning

Inspired by the highly parallel information processing of biological neural networks, neuromorphic computing based on artificial synapses offers a promising route to overcome the von Neumann bottleneck. Here, we report a flexible optoelectronic memristor employing two-dimensional (C 6 H 5 CH 2 NH 3 ) 2 CuBr 4 perovskite, which simultaneously supports electrical and optical stimuli. The device exhibits bidirectional and multilevel conductance modulation, emulating key synaptic functions including pairedpulse facilitation (PPF), spiketimingdependent plasticity (STDP), spikedurationdependent plasticity (SDDP), and long-term potentiation and depression (LTP, LTD). Ultraviolet illumination lowers SET/RESET voltages and induces photopairedpulse facilitation and cumulative conductance potentiation, demonstrating photoinduced synaptic plasticity. Besides, the memristor maintains stable switching characteristics over 1000 electrical cycles, 10 4 s retention, 500 bending cycles, and twoweek exposure at 50% relative humidity, underscoring its environmental and mechanical robustness. Finally, a 7 × 7 memristor array network is shown via simulation to perform unsupervised image learning with high fidelity. These results highlight a versatile, leadfree perovskite platform for flexible, lowpower, multimodal neuromorphic hardware.