In this work, we report a high-performance photodetector based on perovskite CH3NH3PbBr3 quantum dots (QDs) films with a configuration of Au/CH3NH3PbBr3 QDs-Al2O3-indium tin oxide/Au as a bioinspired photoelectric synapse. The uniform CH3NH3PbBr3 QDs thin film is fabricated by a electrodepositing QDs solution and exhibits favorable long-term stability at ambient. By inserting an Al2O3 film, the dark current of the QDs film photodetectors is significantly suppressed as a result of the blocking effect, accompanied by the enhanced ON/OFF ratio. Furthermore, the optimal photodetector is utilized as a photoelectric synapse. Through modulating the light pulse stimuli, some underlying synaptic functions, including paired-pulse facilitation, axon-multi-synapses network function, and the transformation from short-term plasticity to long-term plasticity, are flexibly emulated on a single photoelectric synapse. These remarkable results are promising for building hardware units with neuromorphic architecture to mimic the human brain functionalities.

No abstract

Intrigued by the artificial intelligence and brain‐like neural networks, the photoelectronic neuromorphic devices that can simulate synaptic functions have been attracting wide attention. However, equalizing excitation‐inhibition has been seldom artificially realized despite its fundamental feature in a biological system. In this work, an artificial synapse is proposed based on a sandwiched Ag/Al2O3/graphene/ITO photodetector. The device exhibits a special ambipolar photoresponse at the millivoltage as a result of the interfacial trapping effects. Under the coupled light and electrical stimulus, the synaptic excitatory, inhibitory signals and their recovery are emulated on a single device via the positive and negative photoresponse, respectively. Further, the biologically dynamic balance of excitation and inhibition is reproduced on a simple analogous system integrating multiple photoelectronic synaptic cells. These remarkable results provide a potential foundation for building hardware units with neuromorphic architecture to mimic the complex human brain functionalities.