Computational And Directional Design Of 2D/3D Metal-organic Framework Materials And Membranes With High Separation Performanc

With the development of industrialization,separation of Xe and Kr is extremely important in several applications,such as spent nuclear fuel reprocessing,and with the attention paid to environmental protection,the removal of low concentration of xylene from the air has attracted much attention.The main materials of this study are based on hypothetical metal–organic frameworks（h MOFs）database and computation-ready experimental metal–organic frameworks（Co RE-MOFs）database.High-throughtput computational screening（HTCS）machine learning（ML）and Molecular fingerprint（MF）methods were used to systematically study the adsorption and separation of gases（xylene adsorption and Kr/Xe separation）.The molecular simulation of HTCS include grand canonical Monte Carlo（GCMC）and molecular dynamic（MD）.Supervised learning ML algorithms were uesd,such as Decision tree（DT）,Random Forest（RF）,Support vector Machine（SVM）,Back Propagation Neural Network（BPNN）and K-nearest neighbor（KNN）.And MACCS fingerprint is used in MF method.The specific studies are as follows:1.In this work,firstly,634 two-dimensional metal organic frameworks（2D MOFs）were screened from 6013 Co RE-MOF databases and the dynamic adsorption behavior for the separation of Kr and Xe was simulated by HTCS.The relationship between the structure of 2D MOFs membranes and the separation performance（permeability（P）and permselectivity（S））was established.It was shown that PLD,LCD,K and VSA were positively correlated with the permeability.Then,the excellent organic linker（linker）was selected as carboxyl group by MF,and the excellent metals（node）were identified as Zn,Fe and Cd.Mechanism of the existence of Kr/Xe separation by 2D MOFs was further investigated.Finally,based on the excellent linker and node,three design strategies are proposed to improve the performance of 2D MOFs membranes.2.In this work,HTCS was used to simulate the dynamic behavior of Kr/Xe separation for6013 Co RE-MOFs membranes under atmospheric conditions.First,the structure–performance relationships of the metal–organic framework（MOFs）membrane for Kr/Xe separation were analyzed by univariate analysis,indicating that the pore limiting diameter（PLD）was a key descriptor.Then,four ML algorithms（RF,DT,SVM and KNN）were employed for classification of permeability（P）and permselectivity（S）,which were further predicted by the RF regression algorithm.Results show that a higher classification accuracy could be obtained for P and the comprehensive trade-off between P and S（MPS）had a good fitting effect.Besides,the excellent bits of linkers were determined by MF,and the excellent nodes and separation mechanisms were also discussed.The linker with aromatic rings,hydroxyl groups and double bonds connected with oxygen atoms and the node with Zn were found to be beneficial for improving the Kr/Xe separation performance.Finally,three design strategies were proposed to boost the Kr/Xe separation performance of MOFs membranes.Combining HTCS,ML and MF,we provide a new direction for designing high-performance MOFs membranes for Kr/Xe separation.3.In this study,the adsorption behavior of 31399 h MOFs for xylene in air was simulated by HTCS at atmospheric pressure.First,the structure-performance relationship of h MOFs was established through univariate analysis.It was shown that there was a strong correlation among heat of adsorption(Q_st⁰)and Henry’s coefficient（K）and the three performances adsorption（N）,selectivity（S）and trade-off value（TSC）.Further,four machine learning algorithms（BPNN,DT,RF and SVM）were used to classify and predict the adsorption performance of h MOFs.Results showed that P1 had a high classification effect and TSC had a good fitting effect.Then,the relative importance of descriptors was analyzed by RF model,indicating that Q_st⁰ and K were important descriptors affecting the MOFs adsorption for xylene.Then,MACCS fingerprint was employed to analysis the high-performance area（P1）and its adsorption mechanism was discussed,the results found that MOFs adsorption depends mainly onπ-πinteraction for xylene.Finally,high-performance h MOFs were successfully screened for xylene adsorption based on excellent bits.In this study,the adsorption and separation performance of different gas molecules（xenon,krypton and xylene）in multiple databases（h MOFs and Co RE-MOFs）were investigated by a variety of analytical methods（HTCS,ML and MF）.From the aspects of MOFs components,structure-activity relationship,adsorption and separation mechanism,the target performance of MOFs materials was described,evaluated and screened.