Simulation And Design Of Metal-organic Frameworks Based On Formaldehyde And Non-methane Total Hydrocarbon Removal

The main materials of this study are metal–organic frameworks(MOFs),which come from hypothetical MOFs that may be synthesized by computer.In this study,high-throughput computational screening(HTCS),machine learning(ML),and molecular fingerprint(MF)were used to capture formaldehyde and non-methane hydrocarbons(NMHCs)in the air.For the molecular simulation method of HTCS,grand canonical Monte Carlo(GCMC)was used for calculation.The ML algorithms mainly adopted supervised learning algorithms such as support vector machines and random forest(RF),and MACCS fingerprint was mainly used for MF methods for conversion.This study was divided into three chapters.In the first part,the descriptor range of excellent MOFs for capturing formaldehyde in the air was calculated;In the second part,excellent MOFs for capturing NMHCs(C₃–C₆)in the air were selected from the existing database;The last part puted forward some opinions on the design of excellent MOFs that capture C₃–C₆ in the air.The specific studies are as follows:1.As one of the most familiar carcinogenic and teratogenic air pollutants,formaldehyde is flooded in human daily life,and even a trace amount of formaldehyde will pose a threat to human health.Firstly,31399 hydrophobic MOFs(hMOFs)were selected from 137953hypothetical MOFs through the limit of the Henry's coefficient of water,and their adsorption performance to adsorb formaldehyde in ternary air mixture was calculated.We divided three data sets to explore the performance of different materials to capture formaldehyde from the air.ML algorithms were used to explore the influence of different algorithms,performance indicators,and data sets on the performance prediction of formaldehyde system,and the relative importance of material structure/energy descriptor in ML model was compared.Then,the probability of obtaining target MOFs under the specific path of each descriptor was calculated,quantitatively defining the path to obtain excellent MOFs according to each descriptor.2.Straight chain alkanes(C₃–C₆)have different degrees of explosive effect and neurotoxicity,so it is very necessary to obtain excellent adsorption materials for removing C₃–C₆ from the air.Firstly,the adsorption effects of 31399 hMOFs on C₃–C₆ in the air were obtained by HTCS,respectively.The distribution range and trend of the descriptors of MOFs for the efficient adsorption of C₃–C₆ were obtained by univariate analysis,and found that the existence of“second peak”provides more choices for the pore size of excellent MOFs for adsorption of C₃–C₆.ML classification algorithm predicted the different performance intervals of C₃–C₆,and it was found that RF algorithm has the highest classification accuracy for high-performance MOFs set.Finally,in the existing database,we screened three candidate MOFs for C₃–C₆,respectively.3.NMHCs(C₃–C₆)are important precursors of ozone generated by complex reaction mechanism with nitrogen oxides under ultraviolet conditions.It is always a challenge to explore the promotion mechanism of excellent adsorption materials for removing NMHCs from the air and design new materials.The ML regression algorithms found that RF algorithm has excellent prediction ability for the performance values in the medium-performance MOF set.The conclusion that the sum of the relative importance of energy descriptors exceeds 60%confirmed the important influence of energy descriptors in the system.By mining the distribution law of fingerprint tag numbers of MOFs in each interval,the excellent fingerprint number that plays a decisive role in the adsorption performance of each NMHC was gradually defined.Finally,we found that the fingerprint information carried by MOFs may be the key reason for promoting the adsorption of C₃–C₆,and designed the substructures of excellent MOFs adsorbing C₃–C₆respectively.In the system of capturing formaldehyde in the air,we used the way of path definition to obtain the probability of capturing excellent MOFs of formaldehyde from the specific path of each descriptor;In the system of screening MOFs to capture C₃–C₆ in the air,we screened some MOFs with excellent performance in the existing database;In the system of designing MOFs to capture C₃–C₆ in the air,ML and MF gave excellent fingerprint tag numbers that play a key role in the adsorption of NMHCs,so that we could independently design its functional groups according to the characteristics of different NMHCs,and it was expected to establish a new database.