Sustainable All‐Inorganic Double Perovskite Memristors Enabling Synaptic Learning and Cognitive Emulation

Why It Matters

The breakthrough offers a sustainable, non‑toxic alternative for neuromorphic hardware, accelerating the shift toward energy‑efficient brain‑inspired computing systems.

Key Takeaways

• Lead-free Cs2AgBiBr6 memristors show analog switching.
• Switching driven by reversible Ag+ ion migration.
• Achieve >95% MNIST accuracy with spiking neural networks.
• Demonstrate synaptic plasticity: LTP, LTD, paired-pulse facilitation.
• Enable adaptive associative learning, mimicking Pavlovian conditioning.

Pulse Analysis

The emergence of lead‑free halide perovskites marks a pivotal shift in materials science, addressing both performance and environmental concerns. Cs2AgBiBr6, an all‑inorganic double perovskite, offers high chemical stability and low toxicity compared with traditional lead‑based counterparts. By employing a cooling‑induced crystallization process followed by thin‑film deposition, researchers achieved uniform layers with sub‑4 nm roughness, enabling reliable analog resistive switching essential for memristive applications.

Beyond conventional memory functions, these memristors demonstrate neuromorphic capabilities that mirror biological synapses. Detailed nanoscale investigations reveal that Ag⁺ ion migration creates and dissolves conductive filaments, producing controllable weight updates that replicate excitatory postsynaptic currents, paired‑pulse facilitation, and long‑term potentiation/depression. Such intrinsic plasticity reduces the need for peripheral circuitry, paving the way for compact, low‑power neuromorphic processors.

When coupled with spiking neural networks, the Cs2AgBiBr6 devices deliver over 95% classification accuracy on the MNIST dataset, surpassing many existing memristive platforms. Their ability to perform adaptive associative learning—demonstrated through Pavlovian‑style conditioning—highlights a new class of cognitive emulation hardware. This convergence of sustainability, material robustness, and advanced learning functions positions double perovskite memristors as a cornerstone for next‑generation brain‑inspired computing architectures.