机器学习筛选用于气体吸附分离和存储的金属有机骨架材料

金属有机框架（Metal-organic framework ，MOF）因其高孔隙率、高比表面积和结构可调性，在气体吸附分离领域广泛应用。随着MOF数量激增，传统分子模拟和实验方法验证MOF性能成本高且速度慢，因此目前MOF筛选工作已转向高通量计算辅助的机器学习（Machine-learning，ML）。机器学习作为一种高效的大数据处理方法，能够在高通量筛选（High-Throughput Computational Screening，HTCS）的基础上对数据进行拟合，从而快速而准确地筛选出气体吸附分离材料，并深入挖掘其结构与性能之间的关系。本文回顾了近年机器学习应用于MOF筛选的研究。本文重点讨论了一些运用机器学习从大量结构中筛选出可用于CH4、H2和CO2等气体吸附分离与储存的MOF材料的工作。同时，我们梳理了当前MOF材料筛选工作中的研究思路和进展，并指出了机器学习在筛选MOF材料工作中面临的一些瓶颈和挑战。最后，对该领域的未来发展前景进行了展望。

Machine Learning Screening of MOF Materials for Gas Adsorption Separation

Metal-organic framework (MOF) is widely used in the field of gas adsorption separation due to their high porosity, high specific surface area and structural adjustability. As the number of MOF surges, traditional molecular simulation and experimental methods are costly and slow to verify the performance of MOF, so the current MOF screening efforts have turned to high-throughput computation-assisted machine learning (ML). As an efficient big data processing method, ML can fit data on the basis of high-throughput computational screening(HTCS), so as to quickly and accurately screen out gas adsorption separation materials, and deeply explore the relationship between their structure and performance. This paper reviews recent studies on the application of ML to MOF screening. In this paper, we focus on some work using ML to select MOF materials from a large number of structures that can be used for adsorption, separation and storage of CH4, H2 and CO2. Moreover, we review the current research ideas and progress in MOF material screening, and point out some bottlenecks and challenges faced by ML in the screening of MOF materials. Finally, the future development prospect of this field is prospected.