Modification Of The Conductive Behavior Of PEDOT: PSS And Its Application On Neural Simulation

Since the beginning of the21th century, electrical technology based onsemiconductor materials has developed dramatically. Transistor, integrated circuit, largescale integrated circuit and very large scale integrated circuit has induced the humansociety into an information age. The mix of poly(3,4-ethylenedioxythiophene) andpoly(styrenesulfonate), denoted as PEDOT:PSS has attracted researchers for itsnumerous electrical characteristics. This work concentrated on PEDOT:PSS, optimizingthe resistive memory properties through modulation of its conductive behavior.Resistive memory is one of the very promising memory, which combines theadvantages of current memory devices, including fast-writing, nonvolatile, low-cost andeasy-fabricating. Organic resistive memory has a special potential of flexible devices.Then we compared PEDOT:PSS with traditional metal filament, illustrating itsadvantages as resistive switching channel. In addition, we applied the PEDOT:PSSresistive memory to neural network simulation, providing an element model in neuralnetwork computer.Firstly, we fabricated a series of organic resistive memories via magnetronsputtering, electron beam evaporation and spin coating. Through diluted doping of PVP,we enlarged the resistive window of Al/PEDOT:PSS/Al devices by10to100times,which could still maintain for over105s. Mechanism of the enlargement of resistivewindow was investigated by XPS, Raman spectrum and AFM, pointing out that PVPhad two functions in the system, including inducing charge traps which increase thehigh resistance and changing the conformation which decrease the low resistance, henceenlarge the resistive window. This work shows a feasible way to optimize resistivememory device.Secondly, after electrical measurements, we found PEDOT:PSS resistive switchingchannel was better than metal filaments in resistive memory based on PVP. In deviceusing mixed PEDOT:PSS and PVP, we observed a resistive memory effect with thewindow over103, and retention over105s. meanwhile we demonstrated basiccharacteristics of different resistive channel, and supplied a reference in later researches.At last, we applied Ti/PEDOT:PSS/Ti resistive junction to neural network simulation, using pulse test to mimic human behavior under external stimulus, includingpotentiation and depression. Then we analyzed the resistive variation by TEM, electricalmeasurements on devices with different thickness and electrode size. We found that theresistive switch was resulted in the redox of PEDOT and the movement of PSS-. Theredox of PEDOT enables its resistive switch within one cycle, the movement of PSS-leads to the continues increase of the mean resistance. This work makes contribution tothe realization of neural network computer.