Synaptic plasticity and handwritten digit recognition of a memristor based on a high-stability lead-free Cs3Bi2Br9 perovskite thin film†

Abstract

The lead-free Cs₃Bi₂Br₉ perovskite has emerged as a promising candidate for memristor and artificial synapse devices due to its high environmental stability and low toxicity compared to lead-based alternatives. In this work, we successfully prepared a high-quality Cs₃Bi₂Br₉ perovskite film via a simple spin-coating method combined with low-pressure assisted treatment. Based on the obtained Cs₃Bi₂Br₉ film, a memristor with the structure of W/Cs₃Bi₂Br₉/ITO was fabricated. The memristor demonstrated excellent resistive switching performance, including analog-switching behavior, high environmental stability (>11 months), low operating voltages (V_FORMING ∼ 0.65 V, V_SET ∼ 0.53 ± 0.08 V, and V_RESET ∼ −0.83 ± 0.11 V), fast switching speed (<1 μs), and long switching endurance (>1100 cycles). Furthermore, the synaptic plasticity aspects such as short-term plasticity, long-term plasticity, and synaptic weight potentiation and depression were successfully simulated via pulse-train measurement. Finally, a fully connected neural network built using the W/Cs₃Bi₂Br₉/ITO memristor can obtain an accuracy of about 90% in recognizing handwritten digits. The results indicate that the lead-free Cs₃Bi₂Br₉-based memristor has great potential in high-stability, cost-effective, eco-friendly, and low-power consumption nonvolatile memory and neuromorphic computing applications.