In recent years,artificial intelligence technology has developed rapidly,and traditional computer architectures have not been able to meet the requirements of high efficiency and low power consumption.Research and development of brain-like neuron-based computing devices based on new non-volatile memory technology to build artificial brain artificial neural network computing The system is the key to improving computing performance;phase change memory cells have both multi-value storage and energy accumulation characteristics,which is one of the ideal choices for the development of brain neuron-like computing devices.Therefore,it is important to study the calculation functions of phase change memory cells.Based on the MATLAB simulation platform and the working principle of the phase change memory cell,a three-dimensional finite element simulation system of phase change memory cell is established in this paper.First,the feasibility of the addition calculation function of the phase change memory cell is simulated by controlling the pulse parameters and the number of pulses.Secondly,on the basis of the above,the effects of cell structure parameters and pulse waveforms on the addition calculation performance of phase change memory cells were studied.Simulation results show that:(1)Reducing the SET operation pulse voltage or pulse width can partially crystallize the high-resistance phase-change memory cell.Applying different numbers of pulses can obtain a corresponding number of intermediate resistance states.Based on the number of pulses and intermediate The relationship of the resistance state can realize the addition calculation function of the phase change memory cell.(2)When the cross-sectional area of the heater decreases,the phase-change memory cell addition calculation performance increases first and then decreases,and the energy consumption required to achieve the addition calculation operation gradually decreases;the change in heater thickness does not affect the phase-change memory cell addition calculation performance.Large;reducing the thickness of the phase change layer GST can reduce the energy consumption required to implement the phase change memory cell addition calculation operation,and the addition calculation performance increases first and then decreases,and there are more suitable sizes for implementing the phase change memory cell addition calculation function.(3)Adjusting the single-pulse parameter has little effect on the addition calculation performance of the phase-change memory cell,while controlling the double-pulse parameter can make the distribution of the intermediate resistance state of the phase-change memory cell more uniform,and the addition calculation performance is relatively better. |