随着信息技术的飞速发展，数据量亦同步激增。利用大数据技术处理海量数据，为先进存储与计算应用展现了广阔前景。然而，传统的冯·诺伊曼架构在大数据场景下面临时间延迟和高能耗等挑战。与此同时，当前广泛使用的硬件主要基于块体材料设计和制备，在进一步提高器件集成密度时存在困难。二维材料因其在减小厚度和设备微型化方面具有巨大潜力而备受关注。此外，二维材料具有丰富的电子可调特性，有利于器件功能设计。近年来，有研究发现二维电子器件可模拟神经形态计算功能，在设计基于新型神经形态架构的硬件方面展现出优势。然而，目前报道的基于二维材料的神经形态器件，其工作机制主要基于缺陷或外部添加离子的迁移，存在高能耗、大漏电流以及可调性差等问题，阻碍其进一步发展。本论文旨在研究具有本征可移动离子的离子型二维材料作为神经形态器件功能层的潜力。通过探究离子型二维材料中本征离子的不同运行机制，以设计出高性能、具备新型功能的神经形态器件。论文主要结论如下：(1) 基于离子型二维材料的本征离子迁移，构建了具有低操作电流的多态忆阻器。利用原位扫描电子显微镜/能谱仪观察了本征铜离子的可逆迁移，发现该忆阻器展现出>1000个操作态以及低操作电流（~100 pA）。基于本征离子迁移的器件具有良好的均一性，能够实现高精度、鲁棒性好的手写数字图像识别。(2) 基于离子型二维材料的本征离子累积，构建了可重构多态光伏器件。发现电学极化可以调控器件内部铜离子的分布，实现类 PN/NP 结的电学输出特性。基于两个 PN 型器件组合，实现了逻辑“或”门模拟。此外，相关器件可进行多态光伏操作编程操作（具有正/负输出），实现图像多功能处理。(3)在离子型二维材料器件中发现了调控其功能的新方法。发现离子型二维材料内部可以产生面内离子的各向异性，导致各向异性忆阻行为。基于本征离子场控作用，可产生宽回滞，基于离子-电子耦合作用，可实现各向异性调控晶体管输出。综上所述，本论文利用离子型二维材料构建二维神经形态电子器件，基于对本征离子的操控调控其器件的功能。这些电子器件在可调性、稳定性、功能性等方面有突出表现，在神经形态计算等应用领域有很好的应用潜力。

With the rapid development of information technologies, the amount of data is increased as well. Utilizing and processing the large amount of data by big data technologies are promising for advanced memory and computing applications. However, the traditional hardware architecture, Von Neumann architecture, shows drawbacks in big data scenarios such as time latency and power consumption. In addition, currently used hardware is mainly based on bulk materials, which is difficult for further increase of integration density. Two-dimensional (2D) materials are gained much attention due to their satisfactory potential in thickness decrease and device miniaturization. What is more, the 2D materials is capable of function design, as they have abundant electronic tunability. Recent works have used 2D material-based device to achieve neuromorphic computing, showing great potential in designing hardware with novel neuromorphic architecture. However, the reported 2D material-based neuromorphic devices mainly work based on the migration of defects or externally added ions, leading to the difficulties such as high-power consumption, large leakage current, and poor tunability. This dissertation studies the potential of ionic 2D materials with intrinsic moveable ions to serve as the functional layer of neuromorphic device. The different operation mechanisms of intrinsic ions in ionic 2D materials are studied to design neuromorphic devices with high performance and novel functions.The main conclusions of the dissertation are listed below:(1) Multi-state memristors with low operating current were constructed based on the intrinsic ion migration in ionic 2D materials. Using in-situ scanning electron microscopy/energy-dispersive X-ray spectroscopy, the reversible migration of intrinsic copper ions was observed. These memristors exhibited over 1000 distinct operating states and low operating current (~100 pA). The Devices leveraging intrinsic ion migration demonstrated excellent uniformity, enabling high-precision and robust handwritten digit image recognition.(2) Reconfigurable multi-state photovoltaic devices were fabricated based on the intrinsic ion accumulation in ionic 2D materials. The electrical poling modulated the distribution of copper ions within the device, achieving electrical output characteristics analogous to PN/NP junctions. Logic "OR" gate operation was simulated using a combination of two PN-type devices. Furthermore, these devices were programmed for multi-state photovoltaic operation (featuring positive/negative output), enabling versatile image processing functionalities.(3) A novel method for modulating device functionality in ionic 2D material devices was discovered. It was found that in-plane ionic anisotropy can be generated within ionic 2D materials, leading to anisotropic behavior of the memristors. Based on the intrinsic ion field-effect, a wide hysteresis could be generated. Furthermore, anisotropic modulation of transistor output was achieved based on ion-electron coupling.In summary, this work employed ionic 2D materials to construct 2D neuromorphic devices, where functionalities were modulated through the manipulation of intrinsic ions. These devices demonstrated outstanding performance in terms of tunability, stability, and functionality, exhibiting significant application potential in fields such as neuromorphic computing.