Fabrication And Performance Research Of P3HT/C₆₀ Organic Heterojunction Memristor

The flaws of the Von Neumann architecture’s pipelined instructions are gradually enlarged with the improvement of chip performance.Exploring neural computing architectures based on memristor storage and computation integration provides new opportunities for efficient artificial neural network systems.The resistance of the memristor will change with the stimulation of the external voltage and maintain for a period of time,which is similar to the plasticity of biological synapses.The two-terminal electrodes and the active layer of memristor can be viewed as the presynaptic membrane,the postsynaptic membrane and the synaptic gap,the memristor resistance state corresponds to synaptic weight.The number and level of change of which affect the accuracy of synaptic simulation and largely determine the response accuracy of neural computing to stimulus signals.The number and linearity of synaptic weight changes affect the accuracy of synaptic simulation,largely determining the accuracy of the neuromorphic computation in response to stimulation signals.Currently,resistance mutation memristor based on the conductive filament mechanism have high randomness and fewer resistance states in resistance change behavior under voltage stimulation signals.Most analogue memristors with continuous resistance states exhibit uneven resistance changes and insufficient linearity.Resulting in poor computing accuracy when constructing artificial neural networks.The research in this paper uses the organic polymer P3HT as the active layer and the organic small molecule C₆₀ as the oxygen storage layer to prepare an analog organic heterojunction memristor with multilevel stable resistance states and continuous,uniform,and high linearity resistance changes.The specific research contents are as follows:1.Based on the energy level difference between P3HT and C₆₀,the preparation order of the C₆₀layer and P3HT layer was adjusted to construct two heterojunction structures of memristors.The memristive properties and film characteristics of both were characterized and analyzed.Both devices of the two structures exhibite the pinched hysteresis loops under the scan of the bipolar periodic voltage,which is the typical feature of an analogue memristor.Between them,the memristor of the Al/P3HT/C₆₀/ITO structure shows more uniform changes in conductivity under periodic voltage scanning.Moreover,under the stimulation of pulse voltage,there are multilevel stable resistance state value（quantity>100）and higher linearity of resistance state change.2.Comparative experiments of inert metal Au substitution for Al and ITO electrodes,Microscopic real-time characterization of in-situ Raman testing and fitting of SCLC and FN tunneling charge transfer mechanism model reveal the working principle of the analogue memristor.Specifically,applying a voltage at the ITO electrode causes reversible migration of oxygen ions in the Al₂O₃ layer,and the injection of oxygen ions reduces the energy barrier of the P3HT level,resulting in FN tunneling of charges and thus producing continuous changes in the device conductivity.In addition,based on the multilevel resistance state continuous variation characteristic of the memristor,a series of neural synaptic plasticity behaviors are simulated.3.Aiming at the structural characteristics of the cross-array of devices,a circuit system was built to test the Parallel operation characteristics of the memristor.By connecting shunt resistors in parallel,the stability and linearity of the conductance change of the device during pulse voltage stimulation are further improved.And demonstrates the process of linear optimization of synaptic plasticity after memristor artificial synapse training.