Biomimetic Research Of Neural Synapses Based On Biological Memristors

Neuromorphic computing can solve the problem of expensive data transfer between processor and memory,and is a strong candidate to replace the traditional von Neumann computer system.With the increasing attention of artificial synapses in neuromorphic computing,it is of great significance to develop pollution-free and cyclically regenerated artificial synapses.Natural biomaterials are biocompatible,biodegradable,and environmentally friendly,and are ideal materials for building nextgeneration synaptic devices.In this paper,biomaterials such as chitosan and egg albumen were used as the main material of the dielectric layer,and four kinds of biomemristors were fabricated by spin coating and vacuum evaporation.The fabrication methods,dielectric material characterization and electrical properties of four biomemristors were studied.Through the analysis of the conduction mechanism of the biological memristor,the action process of the neurotransmitter in the biological synapse is simulated,and the synapse is constructed.Neural behaviors such as excitatory postsynaptic currents,short-term potentiation,short-term inhibition,and pulse-time-dependent plasticity were achieved,respectively.Using the characteristic that the switching current ratio of chitosan-based biomemristor decreases with the increase of the number of cycles,neural behaviors such as Pavlov associative memory and the transition from short-term memory to long-term memory are realized.Doping of graphene quantum dots enables egg albumen-based biomemristors with UV-sensitive properties.Therefore,the food signal was simulated by ultraviolet irradiation,and the ring signal was simulated by electrical pulse,and the Pavlov associative memory based on photoelectric stimulation was realized.