Photoelectric\nDual-Mode and Nonvolatile Modulation\nBased on Ferroelectric Memristors for Reservoir Computing

Drawing on the function of biological synapses, the development\nof electronic synaptic devices aims to achieve a neuromorphic computing\nsystem with high performance and low energy consumption. In this work,\nthe Ruddlesen–Popper phase Ba<sub>2</sub>TiO<sub>4</sub>/TiO<sub>2</sub> heterojunction memristor was prepared by the sol–gel\nmethod. Hysteresis loops and piezoelectric force microscopy demonstrate\nexcellent ferroelectric properties of Ba<sub>2</sub>TiO<sub>4</sub>, which allows the Ba<sub>2</sub>TiO<sub>4</sub>-based memristor\nto exhibit a multiresistive state under applied voltage. Synaptic\ncharacteristics such as short/long-term plasticity and paired-pulse\nfacilitation/depression are achieved through electrical pulses. In\naddition, the memristor achieves nonvolatile modulation under the\nstimulation of optical pulses, mainly attributed to the Ba<sub>2</sub>TiO<sub>4</sub>/TiO<sub>2</sub> heterojunction promoting the separation\nof photogenerated carriers. The “learning experience”\nprocess is successfully achieved, and the minimum energy consumption\nof a single synaptic event is only 0.051 pJ. A reservoir neural network\nis constructed to evaluate the reliability of Ba<sub>2</sub>TiO<sub>4</sub>/TiO<sub>2</sub> heterojunction memristors. The recognition\naccuracy of the clothing data sets reaches 91.2%. Our research provides\nan idea for explaining the nonvolatile modulation and reservoir computing\nof photoelectric memristors.