Neural Behavior Simulation And Integrated Application Of ZnO Light-controlled Synaptic Devices

In the current era of big data,the enhancement in data processing speed is urgently needed to cope with the increasing requirements for data storage and processing.However,the classical computing architecture based on the von Neumann architecture faces a slowdown in performance improvement owing to the reason that Moore's Law gradually reaches its limit,and there is also a limit in computing speed due to the separate design of the memory and processor.Therefore,it gradually tends to be at a disadvantage in the face of emerging technologies such as pattern recognition,big data processing,and the Internet of Things which rely on massive real-time data.In this dilemma,brain-inspired computing,a new computing model that mimics the way the human brain works,has attracted the interest of many researchers due to its high data processing speed and low power consumption in data processing.At present,the research of brain-inspired computing is mainly divided into two directions: electronic-controlled synaptic devices and light-controlled synaptic devices where the electronic-controlled synaptic devices usually use a series of electrical signals to change the weight of synapses,which limits the speed of data processing owing to the limitation of bandwidth connection density.On the contrary,the light-controlled synaptic devices have higher operating speeds and lower power consumption than electronic-controlled devices due to the advantages of high bandwidth and low crosstalk of optical signals.Based on the above analysis of needs and comparison of advantages,the work of this paper mainly focuses on light-controlled devices.The specific work content mainly includes the following three parts:1)ZnO,an n-type semiconductor,has good persistent photoconductivity characteristics and can well simulate synaptic behaviors.At the same time,it has the advantages of low production cost,non-toxicity,high stability,and is compatible with traditional semiconductor manufacturing technology,which makes it suitable for industrial production.Therefore,ZnO is chosen as the light-sensitive material.In this work,the sensory memory,short-term memory,long-term memory,learning experience,neural alienation,and many other neural behaviors are successfully simulated using a highly light-sensitive planar ZnO light-controlled synaptic device.The recognition rates of 90%,92%,and 86% are successfully achieved in the file type,small number type,and large number type data sets respectively by directly mapping the numerical weight to the conductance state of the synaptic devices.2)Currently,most of the semiconductor parameter testers such as Agilent B1500 A and Keithley 4200 are equipped with only two to three probes,which are only suitable for performance research of discrete light-controlled synaptic devices,which seriously limits the research on the integration of synaptic devices.In this work,a multi-channel synaptic device current test platform based on STM32F407VET6 microprocessor is designed and fabricated,which realizes real-time acquisition of up to 64 device currents,and provides a good experimental platform for the integrated application research of ZnO light-controlled synapse devices.3)At present,the research work of light-controlled synaptic devices mainly focuses on the research and improvement of the performance of a single device,and there is little research on the integration work of light-controlled synaptic devices.Therefore,this work mainly realizes the integrated application research for synaptic device arrays.In this work,the multi-channel device current test platform based on the STM32F407VET6 microprocessor,which equipped with 8 × 8 synaptic device array modules and 8 channel analog-to-digital conversion modules(ADCs)and other peripherals is used to realize the preliminary exploration of the integration of ZnO photo-controlled synapse devices.The neural behaviors such as short-term memory,long-term memory,and learning experience are realized using the real-time 64 device currents measured from the synaptic device arrays,further providing a good reference for the integrated application of light-controlled synaptic devices.