<p>A rather simple compact multifilamentary circuit-level model of bipolar memristor resistive switching with controlled multilevel conductivity tuning in a metal oxide memristor is presented. The model is implemented in a SPICE code. The proposed model differs from known multifilament models by a simpler equation for total current and the presence of only one state parameter for all filaments and, accordingly, one differential equation. This leads to a decrease in the calculation and fitting time of the model. The simplifications made do not lead to a decrease in the accuracy of the model since they are compensated by using a larger number of filaments without a corresponding increase in the number of differential equations. It is shown that the model reproduces the experimental current–voltage characteristics better. This is indicated by the higher value of the determination coefficient. In addition, the model reproduces the experimental curve of spike time-dependent plasticity quite accurately. The developed model will reduce the simulation time of signal processing in large memristor arrays compared to the known compact models of multifilament memristors.</p>

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Compact multifilamentary circuit-level model for multilevel bipolar resistive switching in memristors

  • Alexander Busygin,
  • Sergey Udovichenko,
  • Oleg Maevsky,
  • Alexander Pisarev,
  • Abdulla Ebrahim

摘要

A rather simple compact multifilamentary circuit-level model of bipolar memristor resistive switching with controlled multilevel conductivity tuning in a metal oxide memristor is presented. The model is implemented in a SPICE code. The proposed model differs from known multifilament models by a simpler equation for total current and the presence of only one state parameter for all filaments and, accordingly, one differential equation. This leads to a decrease in the calculation and fitting time of the model. The simplifications made do not lead to a decrease in the accuracy of the model since they are compensated by using a larger number of filaments without a corresponding increase in the number of differential equations. It is shown that the model reproduces the experimental current–voltage characteristics better. This is indicated by the higher value of the determination coefficient. In addition, the model reproduces the experimental curve of spike time-dependent plasticity quite accurately. The developed model will reduce the simulation time of signal processing in large memristor arrays compared to the known compact models of multifilament memristors.