Investigations On Nanofluidic Memristor Synapse Device And Optical Characterization Method

With the rapid development of information technology in the past few decades,the performance of computers has been significantly improved.Nowadays computing efficiency is regarded as a new benchmark of computing capacity,especially in big data environments.Including the Internet of Things and autonomous driving.To further increase computing efficiency,electronic devices need to keep scaling to reduce the fabrication cost,improve the speed,and reduce power consumption.Due to physical limitation and fabrication cost constraints,traditional CMOS technology nodes below 10 nm are not able to effectively reduce hardware costs and improve performance.Novel electronic devices with higher computing efficiency are needed to meet the growing application needs of the information technology industry.Memristors are considered to be a promising electronic device used as artificial synapses to implement next-generation computing systems.At present,computing systems based on traditional von Neumann architecture face enormous challenges in dealing big data tasks.The new computing architecture based on memristor has demonstrated great potential for application and may replace or supply the von Neumann computing architecture.However,there are still many technical challenges in implementing a memristor-based computing system,and many problems need to be solved.The thesis introduces a new type of interfacial memristor based on nanofluidic technology to realize artificial synaptic function,and carried out various experiments and theoretical research.In the experiment,an organic ionic liquid was introduced,and it was found that the fluid device has a memristive effect,so that it was further studied.The nanofluidic memristor is introduced in detail.Firstly,theinjection of ionic liquid and KCl solution at both ends of the nanochannel is introduced,so that the conductivity characteristics of the nanochannel become non-volatile and can simulate the behavior of biological synapses.Secondly,the method of preparing nanofluidic devices by semiconductor processes and some commonly used nano-device characterization methods are introduced in detail to characterize nanofluidic devices.Then,by using the fluorescence characterization method to investigate the movement of the interface between the ionic liquid and the KCl solution under voltage driving,the characteristics of the conductivity gradient of the nanofluidic memristor are verified.Then,using the classical mechanics and fluid equations to mathematically model and simulate the whole nanochannel system,it is concluded that the simulation results are consistent with the experimental results,thus the accuracy of the guessing mechanism of the nanofluidic memristor is further verified.Finally,the parameters of nanofluidic memristor were substituted into the artificial neural network as artificial synapses to realize the handwritten digit recognition task in simulation,and finally obtained a recognition rate of 94%.These results are very meaningful for the future implementation of nanochannel-based neuromorphic computing.