忆阻器由于记忆特性以及纳米尺度，更精确的模拟神经突触功能，将给人工智能带来理论和应用研究的突破性进展。介电层是忆阻器的重要组成部分，影响着导电通路的形成。介电层合成一般通过物理沉积方法，其微观结构为颗粒密实堆积，纳米孔隙无序分布。因此导电通路在介电层的形成是无序，且介电层功能单一，限制其在神经元突触更复杂记忆模拟中的应用。本项目拟设计孔道尺寸梯度分布的SiO2膜和功能化介孔TiO2和WO3膜为介电层构筑忆阻器,并用于神经元突触功能的模拟。首先，通过对孔道尺寸、膜厚度和孔道连接方式的调控，探索导电细丝在SiO2膜中生长机理，实现神经元突触复杂功能模拟。其次，探索功能化有序介孔TiO2和WO3膜的合成，模拟神经元在不同气氛下（CO、NH3等）的记忆行为。最后，探索上转换纳米颗粒与介孔TiO2膜的复合方式，模拟在近红外光和热敏条件下神经元突触的记忆行为，为构建人工智能计算的微处理器提供新的思路。

Memristors with characteristics of memory and nano-scale can be applied to accurately simulate neuromorphic synapse functions, which can bring about the breakthrough in the artificial intelligence. Dielectric film is an important component in the memristors that greatly impacts the formation of conductive path between the electrodes. The fabrication of dielectric film is generally conducted by physical deposition methods, resulting to the formation of dense dielectric film with dense pack of nanoparticles and random distribution of nanospace in it. Therefore, the conductive pathway forms randomly in the dielectric film. Furthermore, the function of corresponded dielectric film is not versatile. These disadvantages significantly restrict the application of memristors in simulating complex functions of synapse. This project is proposed to design the silica dielectric film with pore size gradient distribution and functional ordered mesoporous titania and tungsten oxides film to construct memristors. Based on these kinds of memristors, the simulation of synapse are intended to explore. Firstly, the pore size, thickness and the linkage between gradient pores are planned to be controlled to study the filament formation mechanism in the silica dielectric film. The memristors with pore size gradient distribution are proposed to implement the simulation of complex synaptic functions. Secondly, the functional ordered mesoporous titania and tungsten oxides film are fabricated and applied to construct memristors. The synaptic memristive behaviors are intended to simulate in the condition of different environment (CO, NH3, etc.). Finally, the combination of up-conversion nanoparticles and mesoporous TiO2 film is planned to explore and applied to the construction of memristor. The synapse behaviors are intended to simulate in the condition of near-infrared light and different temperature. All above memristors can provide new ideas for constructing new microprocessor of artificial intelligence.