MemriSim: A theoretical framework for simulating electron transport in oxide memristors

We have developed a theoretical framework MemriSim for simulating the resistive switching behaviors of oxide memristors. MemriSim comprises two major parts, i) structural evolution of oxygen vacancies during conductive filament formation/rupture by kinetic Monte Carlo (kMC) algorithm, and ii) transport calculations based on the scenario of electron tunneling and thermionic emission with the kMC derived structures. As prototype probes, we have computed the current-voltage (I-V) curves of HfO2 and TaOx based memristors and compared the results with experimental measurements, which show perfect agreement. By tuning the physical parameters, MemriSim can describe resistive switching devices with different oxide layers and metal electrodes. In addition, the pulse transient current can also be simulated by considering the transient response of RLC circuit. The developed framework not only provides a general approach for understanding the fundamental mechanism of resistive switching in oxides, but also opens up new opportunities for designing and optimizing memristor-based architectures for nonvolatile memory, logic-in-memory and neuromorphic computing.

本研究开发了一款用于模拟氧化物忆阻器（oxide memristors）阻变行为的理论框架MemriSim。该框架包含两大核心模块：其一为基于动力学蒙特卡洛（kinetic Monte Carlo, kMC）算法，模拟导电丝形成与断裂过程中氧空位的结构演化；其二为基于kMC模拟得到的结构，结合电子隧穿与热电子发射机制开展输运计算。作为原型验证工具，本研究计算了基于HfO₂与TaOx的忆阻器的电流-电压（current-voltage, I-V）曲线，并将模拟结果与实验测量值进行比对，二者吻合度极佳。通过调整物理参数，MemriSim可适配具有不同氧化物层与金属电极的阻变器件。此外，通过引入RLC电路的瞬态响应模型，该框架还可模拟脉冲瞬态电流。本研究所开发的MemriSim框架，不仅为理解氧化物阻变的核心机制提供了通用研究路径，更为面向非易失性存储器、存算一体以及神经形态计算的忆阻基架构的设计与优化开辟了全新机遇。