【作者】 刘小华；

【导师】 王欣；

【作者基本信息】 四川大学 ， 化学， 2023， 硕士

【摘要】 近年来,大气中细颗粒物(PM2.5)的存在造成了严重的环境污染。PM2.5的主要成分是次生有机气溶胶(SOAs),含量占20-80%。挥发性有机化合物(VOCs)中的甲苯等芳香族化合物也被认为是SOAs的主要前体。如何有效减少挥发性有机物中的甲苯等的含量,营造一个更健康、更安全的大气环境,是当前亟待解决的环境问题。挥发性有机物的处理技术繁多,通过多孔材料的气体吸附是一类有效的方法。多孔材料中的金属有机框架材料(MOFs)具有高的比表面积、孔隙率以及以易功能化等特点,已广泛应用于气体吸附等领域。本论文以典型VOCs化合物中甲苯蒸气的吸附为研究对象,结合高通量计算和机器学习实现数据驱动的MOFs材料筛选和设计,获得相应的结构功能关系,并探究甲苯在MOFs上的吸附机理,由此提出具有高性能表现的MOFs材料的设计和筛选策略。本论文首先通过巨正则蒙特卡洛(GCMC)方法计算了包含11937个实验合成MOFs的Co RE MOF数据库中的所有MOF材料的甲苯吸附量。其中有802个MOFs的甲苯吸附量均优于已报导的HKUST-1。其中,CMOF-3b具有最高的甲苯蒸气吸附量,在1900 Pa和298 K下达到了25.57 mmol/g,有意思的是该分子最初是用于非均相催化的,这表明基于GCMC的高通量计算筛选可以探索已合成物质的新用途。结构功能分析(QSPR)结果表明金属开放位点(OMS)和具有芳香环配体的MOFs可能具有较高的甲苯蒸气吸附量,并且吸附量与重量比表面积(GSA)和孔隙率(Vf)关系密切。其次,基于分子动力学和密度泛函理论计算研究表明,甲苯分子与金属开放位点(OMS)的结合能力强于有机配体。而以HKUST-1为模板的中心金属替换研究表明某些金属有可能会提高MOFs对甲苯的吸附量。最后,基于改进的MOF-CGCNN算法我们发展了用于甲苯吸附机器学习预测模型,并应用于含137953个虚拟MOFs数据库的虚拟筛选,从中筛选出4个具有高性能的MOFs作为候选结构。本论文结合高通量GCMC计算和机器学习实现了MOFs吸附甲苯蒸气的快速筛选,获得了结构性能关系,提出了MOFs材料设计策略,本文的研究有望为具有高甲苯吸附表现的新型MOFs材料的设计和筛选提供理论参考。更多还原

【Abstract】 In recent years,the presence of fine particulate matter(PM2.5)in the air has caused serious environmental pollution.The main component of PM2.5 is secondary organic aerosols(SOAs),which account for 20-80% of the total.Aromatic compounds such as toluene in volatile organic compounds(VOCs)are considered to be the main precursors of SOAs.Effectively reducing the content of toluene and other VOCs and creating a healthier and safer atmospheric environment is an urgent environmental issue that needs to be addressed.There are various treatment technologies for volatile organic compounds,and gas adsorption by porous materials is an effective method.Metal-organic framework materials(MOFs)in porous materials have high specific surface area,porosity,and ease of functionalization and have been widely used in gas adsorption and other fields.This study focuses on the adsorption of toluene vapor,a typical VOCs compound,in MOFs.By combining high-throughput calculation and machine learning,data-driven screening and design of MOFs materials were carried out,and the corresponding structure-function relationship and the adsorption mechanism of toluene on MOFs were explored.Thus,a high-performance MOFs material design and screening strategy was proposed.Firstly,the Grand Canonical Monte Carlo(GCMC)method was used to calculate the toluene adsorption capacity of all MOFs materials in the Co RE MOF database,which includes 11,937 experimentally synthesized MOFs.Among them,802 MOFs have better toluene adsorption capacity than the reported HKUST-1,with CMOF-3b having the highest toluene vapor adsorption capacity of 25.57 mmol/g at 1900 Pa and298 K.Interestingly,this molecule was originally used for heterogeneous catalysis,indicating that GCMC-based high-throughput computational screening can explore new uses for synthesized materials.Quantitative structure-property relationships(QSPR)showed that MOFs with metal open sites(OMS)and aromatic ligands may have higher toluene vapor adsorption capacity,and the adsorption capacity is closely related to the gravimetric surface area(GSA)and void fraction(Vf).Secondly,molecular dynamics and density functional theory calculations showed that toluene molecules have a stronger binding ability with metal open sites(OMS)than with organic ligands.Substitution of the central metal in HKUST-1 showed that certain metals may enhance the toluene adsorption capacity of MOFs.Finally,with the improved MOF-CGCNN algorithm,a machine learning prediction model for toluene adsorption was developed and applied to virtual screening of 137953 MOFs,and four MOFs with high performance were selected as candidate structures.This study combines high-throughput GCMC calculation and machine learning to screen MOFs for toluene vapor adsorption,obtaining QSPR,and proposing MOFs material design strategies.The research is expected to provide theoretical references for the design and screening of new high-performance MOFs materials for toluene adsorption.更多还原