Multisensory synapses based on Fe3O4/graphene transistors for neuromorphic computing

Multisensory artificial synapses stimulated by multi-physical signals are highly pursued for their promising application prospects in biomimicry and artificial intelligence. An electric field and an optical field are the most effective stimuli for functional solid materials, due to their extraordinary potential in manipulating the electron band structure, spin ordering or even ion migration. Among them, the electric field induced reversible conductance regulations and ultraviolet light-induced ultrafast conductance changes are caused by proton migration under the ionic liquid gating effect and electron transition from the valence band to the conductance band, respectively. Here, we have realized synaptic behavior and logical operation in Fe3O4/graphene artificial synapses by using electrical and optical dual field regulation, which is of great significance for future artificial intelligence devices to recognize and make choices in complex systems. A three-layer artificial neural network consisting of Fe3O4/graphene synapses is built up to simulate handwritten digit recognition with high accuracy. Our work provides an intriguing opportunity for the development of multisensory artificial synapses with low energy consumption.