Selection And Design Of Metal-Organic Framework Adsorbents For Alternative Gas Cryogenic Separation

Gas adsorption is closely related to various aspects of human society,such as the storage of clean energy gases,the management of greenhouse gases,the control of toxic gases,and the separation of important substances in industry.Cryogenic separation,also known as cryogenic distillation,is also used for the separation of many systems,such as large-scale air oxygen production,separation of cracked gas,and separation of mixed xylene.Unfortunately,although this method produces products with a high purity,it is quite energy intensive,and the development of increasingly sophisticated pressure swing adsorption(PSA)shows significant potential to replace it.In this work,the adsorption separation method was used to find its possibility to replace the cryogenic separation method.As a low-energy separation method,adsorption separation has also become a promising alternative technology for gas separation and purification.Therefore,the development of new adsorbent materials is key for the safe and cost-effective storage and separation of gases.Metal-organic frameworks(MOFs),as representatives of new porous materials,have displayed broad application prospects in gas adsorption separation because of their high porosities,adjustable structures,and ease of functionalization.However,due to the exponential explosion of the number of MOFs,it is difficult to meet the needs of the research and development of large-scale new MOFs by relying on traditional trial-and-error.Therefore,the material structure of this study is based on the database of computation-ready experimental MOFs(CoRE-MOFs)that have been synthesized by experiments in the Cambridge Structural Database(CSD),using high-throughtput computational screening(HTCS)technology and machine learning(ML)method,the cryogenic distillation separation methods commonly used in industry(xylene isomer separation and separation of O₂ and N₂)and the capture of energy gases(CH₄ and H₂)in the air were systematically studied.The specific studies are as follows:1.In this study,we synergize computational screening and ML to explore the selective adsorption of p-xylene over o-and m-xylene in MOFs.First,a large set(4764)of computation-ready experimental MOFs are screened by geometric analysis and molecular simulations.The relationships between MOF structural descriptors(including void fraction(?),largest cavity diameter(LCD),pore limiting diameter(PLD),density(?)and volumetric surface area(VSA))and separation performance metrics(adsorption capacity of p-xylene N_p-xylene and the selectivity of p-xylene over o-and m-xylene S_p/(m+o))are established.Then,machine learning methods(back propagation neural network(BPNN)and decision tree(DT)),as well as particle swarm optimization(PSO),are utilized to analyze and optimize N_p-xylene and S_p/(m+o).Based on the threshold values of N_p-xylene>0.5 mol/kg and S_p/(m+o)>5,seven top-performing MOFs are identified,which provides guidance for the design and synthesis of new materials.2.In this study,ML-assisted HTCS echniques were performed to screen the dynamic adsorption of O₂ and N₂ in 6013 CoRE-MOFs,including the competitive adsorption of O₂ and the diffusion of pure N₂ and O₂,to identify the best materials for O₂/N₂ separation.First,we established the relationships between the structural/energy descriptors with the performance indicators.In addition,ML results show that the metal center type of MOFs is a key factor for the separation of O₂/N₂.Finally,the proposed three types of design strategies could significantly improve the performance of MOFs.The combination of HTCS,ML,and designed strategies from bottom to top provide powerful microscopic insights for the development of MOF adsorbents for the separation of O₂ at room temperature.3.In this work,GCMC was performed to calculate the capture of clean energy gases(CH₄and H₂)from air in 6013 CoRE-MOFs.Firstly,the influence laws of adsorption and diffusion of CH₄ and H₂ in air were obtained by molecular simulation as a structure-performance relationship,and then a Python-based automated machine learning development tool(tree-based pipeline optimization tool,TPOT)and two commonly used ML algorithms(DT and random forest(RF))were used to model and evaluate the predictive ability of the model for the two systems(CH₄/O₂+N₂ and H₂/O₂+N₂).Finally,some high-performance materials for different systems were screened out,which provided theoretical guidance for researchers to further experimental synthesis.In this study,using the CoRE-MOFs database,the high-throughput screening of MOFs structure adsorption and selective separation of target molecules in a challenging mixed system was completed based on ML,and the relationship between the structural properties of MOFs with performance was deeply explored.The results show that the ML-assisted HTCS method can not only achieve rapid evaluation and screening of gas adsorption separation performance of porous materials,but also explore the deeper hidden information between material structure and properties,and then guide the design and synthesis of new MOFs.