Self-learning effect of CsFAMAPbIBr memristor achieved by electroforming process

Abstract

Triple cation halide perovskite (TCHP) memristor as an artificial synaptic device is quite important in the realization of strong artificial intelligence. Normally, electroforming is needed before the realization of synaptic functions through resistive switching. Here, the electroforming process is developed to achieve the self-learning of strong artificial intelligence. An ultra-thin isolating modified layer polymethylmethacrylate (PMMA) was used to fabricate the FTO/TCHP/PMMA/Al memristor, while the mechanism of the self-learning process is also presented. In addition, a series of important synaptic functionalities including long-term potentiation, long-term depression and spike-rate-dependent plasticity are stimulated, which demonstrates the capacity of bio-synapse functions emulation of the device. The associative learning is also achieved with our artificial synapse. This work could be beneficial to the development of future neural morphological artificial intelligence with TCHP based synaptic device.