Preparation And Synaptic Simulation Of Microfluidic Memristors Based On Ionic Liquid

The memristors are easy to be manufactured,small in size,similar to synapses in living organisms,and support low-power parallel analog computing,which means that parallel distributed processing can be done while the information is stored and computed while being transmitted in the network.Such brain-like computing architectures and paradigms have shown excellent performance in massively parallel operations dealing with massive amounts of real-time data.Researchers are competing to develop new types of artificial synapses in order to break the power consumption and integration bottlenecks in the development of neuromorphic chips emulating brain-like intelligence.Compared with most current solid-state memristive devices,fluidic memristive devices are closer to biological synapses in structure and working mechanism,have excellent compatibility with biological systems,have better synaptic plasticity,and can introduce different chemicals,thus giving more functions to neuromorphic devices,while the challenge of fluidic memristive devices is mainly to replicate the memory mechanism in biological systems,i.e.,the dissolved ions（especially calcium）transport and accumulation in tissue fluids are used for signalization,information processing and memory building.In order to develop new fluidic memristors,the authors investigated‘‘Preparation and synaptic simulation of microfluidic memristors based on ionic liquid’’,and the results are as follows:（1）We innovatively designed and prepared a Cu/[MMIm][Cl]:H₂O/Cu microfluidic memristor using capillaries as microfluidic tubes and 1,3-dimethylimidazolium chloride（[MMIm][Cl]）as the memristor medium.The process optimization of the device preparation was explored by adjusting the electrode spacing and optimizing the scanning voltage.It is shown that the I-V curve of this fluidic memristor device is oddly symmetric,with a typical non-regressive 0 I-V curve,which is a coupled capacitive and memristor behavior.It also shows a negative differential effect,and its charge transport characteristics are consistent with the TCSCLC model.In the neurosynaptic simulation,the fluid memristor shows brain-like habituation learning behavior and also exhibits temporary memory function.（2）To investigate the influence of electrodes on the performance of this fluidic memristor device,two types of memristors with different electrode systems were designed and prepared with the structures of Cu/[MMIm][Cl]:H₂O/Ag and Ag/[MMIm][Cl]:H₂O/Ag,respectively.Compared with Cu-Cu and Ag-Ag electrode devices,the Cu-Ag electrode device has a larger switching ratio,higher cycling stability,and a diminished negative differential resistance effect.It shows habituation behavior in neurosynaptic simulations,similar to biological synaptic plasticity.It can be used for adaptation to avoid unnecessary waste and improve learning efficiency.（3）Considering the effect of Cl^-on the electrode in ionic liquids,an anion exchange method was used to introduce NO₃^-and remove Cl^-to obtain Ag/[MMIm][NO₃]:H₂O/Ag microfluidic memristor devices.The results show that the I-V curves of this series of fluidic memristors are oddly symmetric and cross the origin,with obvious negative differential effects,and their fits are also consistent with the TCSCLC model.This series of memristors is sensitive to the conditions of electrode materials（copper and silver）,electrode distance,and scanning voltage,and the nitrate ionic fluid memristors are more stable and durable at low scanning voltages compared with chloride salts.The memristor can also mimic neurosynapses and can present neurosynaptic functions such as short-range memory,habituation,and temporary memory.