Tuning the memory and synaptic functions through rational halide-alloying in Cs3Bi2I9 dimer-type perovskite-based memristor

Abstract

In the dynamic landscape of next-generation memory and neuromorphic systems, memristor bridges the gap between conventional electronics and brain-like functionalities. Accordingly, memristors employing metal halide perovskites have garnered considerable attention for the compatible design of resistive memory architectures and energy-efficient neuromorphic synapses. The presence of mixed ionic-electronic conduction aids low voltage switching and tunable current ON/OFF ratio. However, the topic related to lead (Pb) toxicity and structural stability restricts potential applications. In this study, we have successfully deposited thin films of Cs3Bi2I9 perovskites and its halide-mixed counterparts via one-step solution process incorporating bromide and chloride in a specific ratio. All the fabricated perovskite-based devices demonstrated decent bipolar resistive switching performance. However, the chloride-alloyed device demonstrated highest current ON/OFF ratio (>102) and lowest SET voltage (0.32 V), which is attributed to the synergies of increase in Schottky barrier height at the electrode/perovskite interface and induction of chloride vacancies having least activation energy. In addition, the pulse-dependent measurements could produce core synaptic functionalities such as short and long-term potentiation/depression and spike parameter dependent plasticity with enhanced excitatory postsynaptic current than the pristine Cs3Bi2I9-based memristor device. Furthermore, an artificial neural network is accomplished by training with the potentiation/depression data of the device, which revealed 97% accuracy for the MNIST handwritten digit. This study offers insights into the optimization of perovskite materials and highlights the influences of halide-alloying in memristive performance, thereby demonstrating the device's capability to emulate biological synapses.