Performance Improvement Of Polyelectrolyte Bilayer-based Memristor Via Molecule Structure Engineering

The tide of intelligence represented by artificial intelligence has become an irreversible trend of development.The amount of data generated around the world is surging.Therefore,data has become the lifeblood of the rapidly growing digital life.Biological brains can quickly calculate and store a large amount of data,and the synapses responsible for transmitting biological information are similar to resistive memory structures.Therefore,researchers have proposed an artificial neuromorphic device,the memristor,which may break through the bottleneck of memory separation of traditional computers "von Neumann.” And it will be well used in new technologies such as artificial intelligence,the Internet of Things,cloud computing,and 5G technology.Critical to artificial intelligence’s future is emulating biological synapses with memristors.Moreover,according to a wide variety,low cost,simple fabrication and good flexibility,organic materials provide a competitive approach in memristor and synapse emulation.In this paper,two weak polyelectrolytes of polyacrylic acid(PAA)and poly(ethylenimine)(PEI)and three strong polyelectrolytes of poly(styrene sulfonic acid)(PSS),poly(diallyl dimethylammonium chloride)(PDAC)and poly(acrylamide hydrochloride)(PAH)were used as functional layers to prepare three different memristor devices.The preparation conditions of the device were optimized to prove that the resistive properties of the device were realized by the disappearance/reappearance of the ion bilevel and the ion migration.The reasons for the differences in the resistive properties of the memristor were discussed from the perspectives of the strong/weak polyelectrolyte and its molecular chain structure.Due to the different molecular structure of polyelectrolyte,there is steric hindrance effect caused by ring structure in the PSS/PDAC molecular chain,which makes the conductance modulation of ITO/PSS/PDAC/ITO more linear,which is conducive to the realization of synaptic function.The pattern recognition accuracy of the device for handwritten data sets in artificial neural networks reaches 90%,showing great potential for application in neuromorphic computing systems.This study provides a reliable idea to improve the organic resistance by analyzing the differences in the resistance performance and synaptic function of polyelectrolyte memristors based on different molecular structures.