忆阻器是实现神经网络和神经形态计算最有前景的硬件单元之一，掌握忆阻器电导行为的微观机理是提高其性能并成功应用于神经形态器件的基础。本课题拟在现有工作基础上，采用第一性原理计算方法研究BP/POx中离子隧穿机制，探究层状材料中离子层间迁移的动力学机理及其影响因素。利用机械剥离法和超高真空磁控溅射制备BP/POx层状结构，结合转移技术、等离子体增强脉冲激光沉积（PLD）、射频辅助高温CVD系统与半导体光刻等工艺，制备BP/POx基堆叠结构，开展BP/POx层状器件离子隧穿电导行为与忆阻效应内在关系研究，揭示其忆阻机理的微观本质。利用SPICE和MATLAB/Simulink软件，设计忆阻Crossbar架构和神经网络电路，优化Crossbar电形成过程等特性，阐明其神经形态行为机制，探索忆阻器在类脑神经形态器件中应用。该研究对开发类脑智能器件，推动认知系统研究与人工智能发展具有重要意义。

As a two-terminal synaptic device, memristor is one of the most promising candidates for the implementation of neuromorphic computing. To develop neuromorphic memristors, it is necessary to make a thorough inquiry of the memristive mechanism, and the performance of memristor needs further enhancing. In this proposal, the mechanism of memristive behavior in BP/POx based devices is investigated and the performance is optimized to implement neuromorphic hardware. Combining the first-principles calculation method and experimental technologies, the intrinsic relationships between ion transport and memristive behavior in BP/POx based devices are revealed. The ion transport mechanism in BP/POx based stacks and the influencing factors of interlayer ion migration are studied by VASP and ATK. BP/POx layered materials are prepared by mechanical delamination and ultra-high vacuum magnetron sputtering, and BP/POx memristive stacks are fabricated with layered transfer technology, plasma-enhanced pulsed laser deposition (PLD), RF assisted high temperature CVD system and semiconductor lithography. The electrical properties of BP/POx memristive stacks are characterized and further improved from the aspects of the ions, size, and electrode materials and so on. Based on the theoretical calculation results and experimental data, the mathematical models of BP/POx based memristors are built for the simulation of memristive neuromorphic applications. With this model, memristor crossbar and memristive neural network are designed to optimize the formation process of crossbar electricity, and the neuromorphic behavior of memristor are clarified to develop neuromorphic devices. These results can advance the development of memristor based neuromorphic computing system, and promote the cognitive system research and artificial intelligence technology.