Theoretical Study On Adsorption Properties Of Nitrogen In Zeolites

Zeolites are crystalline aluminosilicates with unique frameworks and pores built by SiO4 and AlO4 units,which are linked by bridging oxygen atoms to form various four-connected 3D frameworks.Zeolites are widely used in petrochemical industry,medical treatment,sewage treatment and other fields.How to simulate their adsorption isotherms quickly and accurately,so as to screen out the materials with excellent adsorption performance?Density functional theory(DFT)combined with machine learning(ML)and grand canonical Monte Carlo(GCMC)simulation was used to study the interaction between zeolites and nitrogen.Langmuir adsorption model was used to simulate the adsorption isotherms,and it was verified with the BET experiment.It provides a simple method to screen out zeolites with good nitrogen adsorption performance.The results are summarized as followsThe distribution of nitrogen in MFI zeolite was calculated by GCMC,showing that nitrogen molecules preferred to adsorb in the space with suitable volume.The narrow space made no contribution to the adsorption because it was too small to hold the nitrogen.The binding energy between MFI zeolite and nitrogen was calculated by DFT.Each silicon and oxygen atom around channel can interact with both ends of the nitrogen molecule by electrostatic polarization interaction with the symmetric adsorption mode,simultaneously,displaying stronger binding energy.The binding energies between nitrogen and other zeolites were trained for machine learning.Three descriptors,namely,the number of vertices(Ⅴ)of NBU units,the maximum included sphere diameter(PLD)in channels and degree of geometric deformation(RDLs)can predict the binding energy between zeolites and nitrogen reasonably.The space size was represented by V;and PLD described the"concave and convex"characteristics of zeolite pores.RDLS was correlated with the difference between local charges of silicon and oxygen atoms,which indicated that the electrostatic interaction was the driving force of nitrogen adsorption.The results of machine learning were validated by DFT calculation,which showed the reasonable prediction.Based on the Langmuir adsorption theory,the adsorption isotherms were simulated by calculating the binding energy between nitrogen and 235 topological types zeolites,obtained by combining DFT with machine learning.It was found that the nitrogen adsorbed rapidly in the pore of zeolite in the low pressure region,reflecting the characteristics of the microporous structure of zeolite,showing the qualitative consistence with the results of BET measurement and GCMC simulation.The method above can simulate the adsorption isotherms of zeolites quickly and accurately.It is expected to simulate the adsorption isotherms of other porous materials with more complex structures,such as MOFs and COFs,and to screen out materials with excellent adsorption performance.