Molecular Generation And Crystal Prediction Based On Organic Nanogridarenes Aromatics

In recent years,the research on porous materials such as molecular sieve materials,covalent organic frameworks(COFs),metal-organic frameworks(MOFs),hydrogen-bonded organic frameworks(HOFs)and some polymer materials has aroused much attention,and porous materials have developed and applied in the fields of catalysis,adsorption/separation,sensors,molecular machines,bio-drug loading and therapy,energy storage and conversion,however,these porous materials are limited due to low stability,or due to high preparation costs,poor solubility and degradability.Because of its rich molecular recognition,intrinsic cavity,simple synthesis,adaptive conformation and various functionalization sites,organic macrocyclic aromatic hydrocarbons can form multifunctional crystal materials with unique cavity,high stability,high crystallinity and easy design and assembly through crystal engineering,and have unique advantages in constructing porous materials.Among them,macrocyclic aromatic hydrocarbons,organic nanomiger hydrocarbons,are rigid macrocyclic molecules with fixed shape and clear vertices and regular corners,which have strong shape and dimension expansion and have been studied and applied in photoelectric materials and other aspects,and are ideal candidates for molecular-grade porous materials.Because organic nanomiger aromatics developed late in crystal engineering,it is important to design and construct a method for large-scale development of organic nanomiger aromiger crystals to accelerate and expand the multifunctional application of organic nanomiger aromiger molecules in the field of porous materials.Therefore,this paper designs and completes the research methods suitable for organic nanogarene generation from virtual molecules to related prediction,including: organic nanogarene molecular exhaustive generation,crystal space group and structure prediction,crystal morphology prediction.The results are as follows:(1)Because there are few synthetic molecular samples of organic nanogaromigaromigaromatics which can not support the machine learning model,the automated exhaustive generation model of organic nanogridarenes aromatics,In addition,based on BRICS algorithm,the molecular fragmentation process is completed,and the backbone library and substituent library for molecular generation are obtained.Then based on the structural characteristics of organic nanogarene molecules,fragment combination scheme and other strategies were designed to complete the large-scale exhaustion of lattice arene molecules,and the number of generated lattice arene molecules reached tens of thousands of orders of magnitude.The results showed that the generated molecules completely met the design,and the backbone and uniqueness were strong.It provides a considerable molecular sample database and an expandable virtual generation model for the study of organic nanomiger aromatics.(2)Traditional crystal structure prediction relies on blind selection and multiple selection space group parameters.Due to the high computational cost,the prediction cycle and cost are relatively huge for organic nanogarene macromolecules.In order to construct a rapid prediction of molecular crystals,the molecular space group is predicted by machine learning model,in which the establishment of molecular crystal datasets including organic macrocyclic molecules,organic nanogarene molecules,fluorenyl non-planar molecules for space group prediction is completed,including 828 molecular crystal data,and molecular crystals are mainly concentrated in the P21/c space group and P-1 space group.Then based on the open source package,we complete the characterization of database molecules,in which a total of 102 model input features suitable for random forests are extracted,and the features of information transfer neural network are constructed based on the matrix map of molecules,and then use random forest model and information transfer neural network model to train the characterized molecules for prediction space groups.After testing the accuracy of the model is 50% and 55%,respectively,the possible reason is that the molecular representativeness of the sample set is insufficient,and the complex multiple classification prediction problem requires a higher dataset.Model parameter selection is not yet optimal,and existing descriptors cannot establish strong correlations with spatial groups such as abstract mathematical symmetry structure classification from single molecules alone.However,this model can also reduce the search range of space group effectively.Finally,the space group of semiglass molecules is predicted by the model and its crystal structure is predicted by simulated annealing algorithm in the molecular force field.The results show that the crystal parameters are consistent with the crystal structure prediction experiment.It provides a basic research route to establish a rapid crystal structure prediction work from the space group prediction work.(3)Understanding and predicting the crystal morphology is the basis of modern material function control,and it is important to understand the crystal morphology growth of lattice arene molecules.In this paper,the morphology of lattice arene crystals is predicted,and the aggregation mode of cistrans-TWG windmill molecular crystals is analyzed by Density Function Theory(DFT)calculation,and the theoretical dimension is obtained as two-dimensional block.The conventional crystal morphology prediction model was used to predict cis-trans-TWG windmill molecular crystals,and it was predicted that these methods had a large difference in morphology in the prediction of organic nanomiger aromatics molecules.At the same time,three improvements were made to IE exponential model,and PBC crystal plane discrimination method was redefined,that is,strong bond chain parallel or coplanar discrimination,minimum repeat element clusters were selected,and growth surface spacing parameters were added.Based on the improved model,the crystal morphology of cis-transTWG windmill molecular crystal was predicted,and the results showed that the prediction performance was good,which was consistent with the experimental results.Based on this model,the non-lattice arene molecules are predicted and the results are close to the experiment,so this model has certain qualitative prediction accuracy in the crystal form prediction of some organic molecules and organic nanogram arene molecules.