Construction And Plasticity Study Of HfS₂-based Photoelectric Synaptic Devices

With the development of the information age,people have higher requirements for information processing,calculation and processing.However,Moore’s Law is breaking down,and the traditional von Neumann architecture has reached a bottleneck.To find a new way forward,the researchers turned their attention to neuromorphic computing.The human brain has thousands of neurons that can respond quickly to stimuli.Inspired by the high speed and low power computing of human brain,the use of two-dimensional layered materials to construct transistor devices to simulate artificial synapses,and further simulate synaptic plasticity and human brain function through electrical or light stimulation,which has potential significance for breaking von Neumann bottleneck.The layered HfS₂ semiconductor has high carrier mobility and is highly sensitive to photoelectric,so it can be used as a good channel material for the fabrication of artificial synaptic devices.In addition,the device stability is improved by depositing protective layer,so as to more effectively simulate the human brain related synaptic functions.The research work in this paper is based on HfS₂ semiconductor material,including the following aspects:（1）The investigation of electronic performance of HfS₂ transistors with In protective layer.A 5 nm-thick In layer is deposited as a protective layer on HfS₂ FET and its electronic properties such as output and transfer characteristic curves are studied.The results show that the transistor switch ratio reaches 10⁵,and the mobility is～3 cm²/（V·s）.The linear output curves show an Ohmic contact between the channel and the electrodes,and the transfer curves show hysteresis windows since the charge trapping effect at the HfS₂/Si O₂ interface.In addition,in order to study the stability of devices,the devices with and without the In protective layer are respectively tracked and tested.The results indicate that the stability of devices with In protective layer is improved from 3 weeks to 7 weeks.（2）Simulation of HfS₂ artificial electronic synapses and study of synaptic plasticity.Firstly,it is proved through contrast experiment that the interfacial charge trapping effect between the channel and the dielectric layer in HfS₂ FET is the main reason for the hysteresis phenomenon.Based on the charge trapping effect,the synaptic plasticities of excitatory and inhibitory synapses have been successfully simulated using electrical stimulation,including spike-amplitude dependent plasticity,spike-time dependent plasticity,and spike-number dependent plasticity.In addition,the long-term plasticity is modulated by electrical pulses,and under 100 electrical pulses with an amplitude of-20 V,it exhibits the long-term potentiation,with a corresponding synaptic weight of 124%.Under different positive electrical pulses modulation,it exhibits different states of the long-term depression,with synaptic weights reaching 83%,81%,and 77%,respectively.Finally,human experiential learning behavior is simulated,and it is found that HfS₂-based electronic synapses only need 6 electrical pulses for the second learning to reach the level achieved by the first 30 electrical pulses.These results indicate that HfS₂ synaptic devices can well simulate synaptic plasticities and human brain learning and forgetting functions under electrical stimulation.（3）Synaptic plasticities regulation and emotional memory simulation of HfS₂ artificial optical synapsesExperiments show that the response of the HfS₂-based optical synaptic device is as high as 23 A/W under illumination at a wavelength of 405 nm.Firstly,we simulate the amplitude,width,and frequency dependent synaptic plasticities of light pulses using a wavelength of 405 nm.Secondly,tunable synaptic plasticity under light stimulation is achieved with the aid of gate bias,such as the transition from paired-pulses facilitation（PPF,128%）to paired-pulses depression（PPD,89%）and from long-term potentiation（LTP,283%）to long-term depression（LTD,68%）.Finally,based on the modulated synaptic plasticity,five emotions and the corresponding learning and forgetting processes are simulated.These results indicate that HfS₂ synaptic devices have tunable synaptic plasticity under light stimulation,which is expected to expand the application of synaptic devices in the field of neuromorphic computing.In summary,we improve the stability of HfS₂ transistors with In protective layer,construct artificial synaptic devices with optical/electrical dual modes,study synaptic plasticity in different modes,and achieve the application of emotional memory.The construction and plasticity research of dual-mode and multifunctional HfS₂ artificial synapses will contribute to the development of the next generation of brain like devices with highly functional integration and capable of handing complex problems.