Research On Neuromorphic Devices Based On Organic Polymer Memristor

Computer platforms based on von Neumann computing architecture have to switch between computing units and storage units when processing data.The separation of memory and computation results in a series of problems such as increased power consumption and slow processing of data,which are called von Neumann bottlenecks.Compared with von Neumann computing architecture,the human brain has no clear boundary between computing and storage during information processing.Memory and computations can be performed simultaneously.Therefore,human brain can achieve low power consumption and high efficiency when computing and storaging data.In recent years,people have turned to study new micro-nano devices,hoping to simulate the properties of synapses.Among these nanodevices,memristors have attracted much attention for their similarities to synapses and neurons.The synapses of the human brain can change the weight according to the activities of the neuron and the neural synapses,and then realize the computational memory.The memristor can also change the value of the conductance through the regulation of the external voltage.Among them,the metal-based memristor is regarded as one of the most promising devices in the future non-volatile memory technology thanks to its high operation speed,low power consumption,high durability and excellent retention characteristics.Its mechanism is that the metal ion redox and reaction under the alternating electric field,the conductive filaments are formed and dissolved and then the conductance will change.The migration of metal ions during the operation of the device makes the change of conductance state extremely similar to the migration of calcium ions and the transmission of neurotransmitter in the human brain,which makes it possible to simulate brain-like neuromorphic computing by using artificial neural synapses.So far,many human brain synapses and neuron functions have been well simulated using memristors.The insulating layer in the middle of the memristor can be made of inorganic materials(such as Ti OX,Ta Ox,Hf Ox)or organic materials(such as PVP,PEO,PVA).Among them,organic materials have the advantages of easy preparation and processing,easy modification,biocompatibility,CMOS compatibility,and low turn-on voltage.However,in the metal ion-based organic memristors reported in the past,although most organic memristors can realize the resistance state transformation between high resistance and low resistance,it is difficult to control the multiple conductance states of the device,and it is difficult for the device to operate in different resistance states.Maintaining the corresponding state leads to high volatility,so it is impossible to simulate the corresponding performance of the synapse of the human brain,and it is difficult to perform subsequent brain-like neuromorphic computing.In this paper,the research is mainly to solve the problems of high volatility of memory resistance device,and difficulty in performing the device regulation of multiple conductance states.Our solution is controlling the migration of metal ions and redox process to improve the performance of the device by polymer modification,expecting to simulate human brain synaptic plasticity and apply to brain-like neuromorphic computing,it provides an effective idea for the research and improvement of ion-based memristive devices.(1)Regulation of ion-based memristors devices to simulate synaptic plasticity: we used poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate)(PEDOT:PSS)to dope into the carboxylated chitosan to improve the ion conductivity of amorphous polymer electrolyte and changed the morphology and formation position of metal conductive filaments,thus realizing the growth change of filaments from anode to cathode in the memristor dielectric layer.The uniform distribution of conductive filaments within the dielectric layer greatly weakened the accumulation of joule heat in the device operation.At the same time,the device can show great retention and non-volatility,adjustable multiple conductance states,and it can stably realize the short-term and long-term plasticity of biological synapses,and can simulate the synaptic plasticity of the human brain to realize brain-like neuromorphic computing.(2)Improving ion-based memristors device nonvolatile and achieving a large range of conductance change: this experiment by using water-soluble carboxylated chitosan mixed with polyvinyl alcohol(PVA)and inorganic salt to modify the polymer film,improve the polymer ion conductivity,reduce the device operating switching voltage,and the addition of polyvinyl alcohol also makes polymer molecules inhibit the migration of metal ions.Therefore,the metal can be bound to improve the non-volatility of the device.The device can achieve large conductance state regulation,and can excellently simulate the synaptic plasticity in the brain-like neuromorphic computing.