基于超薄BiFeO3隧道结忆阻器的制备与类脑行为及其物理机制研究

Memristor, which possesses the functions of calculating and storing, is considered to the key component of breaking the limitation of Von Neumannn computer architecture. Taking into account that the conductance modulation has a certain variation effect in current memristor, more stable and controllable BiFeO3 tunneling junctions will be studied in this project based on our previous work. In order to obtain memristor with better properties, we will employ the co-sputtering equipment of pulse laser deposition (PLD) and magnetron sputtering developed by ourselves to prepare ultra-thin films and deposit electrode films in suit. We will modify the stability of ferroelectric polarization by means of interface engineering and stress engineering, and study the effect of interface and stress on the brain-like performance of memristor. The effects of transition metal and noble metals as electrodes with different work function on the stability of ferroelectric properties will be studied in detail, and the relaxation of ferroelectric polarization and the role of oxygen vacancies in the failure mechanism are also explored systemically. Three memory modes and the learning rules of biological neural synaptic plasticity such as spike-timing dependent plasticity (STDP)，will be investigated based in BiFeO3 tunneling junctions. We will also study the transition dynamics process between short-term and long-term memory, reveal the influence law of the control of microstructure, ferroelectric polarization and brain-like performance, and establish effective physical model. The successful accomplishment of this project will provide a direct experimental support and theoretical basis for the development of the memristor device with stable performance applied in neuron system, which has important research value and application prospects.

忆阻器同时具有计算和存储的功能，被认为是能够打破计算机冯诺依曼架构的关键器件。考虑到目前的忆阻器电导调制有一定的波动性，本项目在已有工作基础上，拟采用更加稳定可控的铁酸铋隧道结作为忆阻器。采用自主研发的磁控与脉冲激光共溅射系统，制备超薄BiFeO3薄膜，并原位生长上电极，制备性能优异的隧道结忆阻器。通过界面工程和应力工程，调控稳定的铁电极化特性，进而研究界面和应力对忆阻器类脑性能的影响。研究不同功函数的过渡金属电极与贵金属对稳定铁电性的影响，探索失效机制中铁电极化的衰减和氧空位扮演的角色。研究BiFeO3隧道结忆阻器类脑三种记忆形式和生物神经突触可塑性学习规律STDP等行为表现及短时程与长时程记忆之间的过渡动力学过程，揭示微结构调控-铁电极化-类脑性能之间的影响规律，建立有效的物理模型。为发展具有稳定性能的忆阻器应用于神经系统提供直接的实验支撑和理论依据，具有重要的研究价值和应用前景。

本项目中采用脉冲激光共溅射系统制备了基于超薄BiFeO3薄膜的性能优异的隧道结忆阻器。通过溶解SAO牺牲层来转移超薄单晶BiFeO3功能层薄膜而获得的柔性人工突触，不仅可以实现稳定的忆阻行为，而且可以实现类脑三种记忆形式和生物神经突触可塑性学习规律STDP、短时程与长时程记忆之间的过渡动力学过程等行为表现及脑神经形态的计算。该忆阻器独立式隧道结在平坦和弯曲状态下均表现出出色的可塑性学习特性，因此可以潜在地用作未来的神经形态计算的人工柔性电子突触,而且受大脑启发的神经形态计算系统具有克服低功耗和智能数据分析中传统计算机的冯•诺依曼架构的局限性的能力，为下一代存储器和AI应用提供了巨大潜力。同时，由于其稳定的可控性，柔性人工突触是柔性神经形态系统的有希望的候选者。在此基础上我们对界面和应力对忆阻器类脑性能的影响进行了深入研究，观察到BiFeO3超薄膜显示出强大的铁电特性，从而导致隧道结具有稳定的忆阻性能，此外通过简单弯曲柔性器件可以实现独立BiFeO3薄膜的大尺度连续应变梯度。可控的应变梯度通过挠曲电效应产生的附加电场与BiFeO3极化相关的内电场的耦合，实现了独立BiFeO3的光电导可调。因此，可以机械地写出多电平的光电电压/光电流，而读出则可以通过在光照明下测量独立式BiFeO3的光伏响应得到。而且，通过应变梯度对光电导的调节可以很好地重复，保证了这种机械调节方法的功能性。所证明的多导状态可能在提高光伏器件的效率或作为应变传感器方面有潜在的应用。我们的研究结果拓宽了柔性电子器件中挠曲电效应的研究视野，并可能激发更多利用应变梯度来调节其物理性质的研究。此研究项目为发展具有稳定性能的忆阻器应用于神经系统提供了直接的实验支撑和理论依据，具有重要的研究价值和应用前景。