Computation Chemistry Research About Functionalization On Metal-organic Frameworks And High-Throughput Screening

Adsorption and separation technique has brought people’s attention according to its low cost,easy operation and energy saving.The performance of adsorption and separation technique is based on that of adsorbent.Metal-organic frameworks（MOFs）,as a new class of porous adsorbents,have several advantages including high surface area,large pore volume,tunable structure and function and etc.,which give them more excellent performance about gas adsorption and separation compared with traditional materials.Nowadays the number of experimental MOFs reaches about 100,000,which make it difficult to effectively and systematically explore the relationship between structure and performance.Relatively speaking,not only can computation chemistry effectively mimic the behavior of gas adsorption and separation in MOFs and study the adsorption and mechanism,but also it can predict the performance of MOFs before synthesized and provide theoretical basis for experiment.Herein,this thesis combined several computation methods,such as Grand Canonical Monte Carlo,Density Functional Theory,Ideal Adsorbed Solution Theory,Molecular Dynamics,dynamics in fixed bed and etc.,to analysis the gas adsorption and separation properties of MOFs.Also,high-throughput screening technology combined with artificial neural network were performed aiming to probe into the relationship between structure and performance for MOFs and screen for top candidates..Designing functionalization methods for effective separation of noble gas from the air is significant to meet people’s necessities and develop the industry.This work illuminated the effect of three functionalized group,including-F,-OH and-NH₂,on the performance about adsorption and separation of mixtures of noble gas and mixtures of noble gas with nitrogen in CuBTC.The result suggests that-NH₂ can obviously enhance the polarizability of organic linker（CuBTC:70.55 bohr³,3NH₂-CuBTC:103.49 bohr³）and increase the binding energy of Xe over organic linker from-0.035 kJ/mol（CuBTC）to-8.319 kJ/mol（3NH₂-CuBTC）.Also,-NH₂ can contribute to higher selectivities of mixtures of noble gas and mixtures of noble gas with nitrogen.For example,selectivity of Xe/N₂ at 10 kPa were increased by about a hundredfold.To predict separation performance of Xe/Kr mixtures in MOFs directly by textural properties of the organic linker,which can reduce the cost of research,do afford theoretical basis for exploring new MOFs with excellent performance.This work combined high-throughput screening technique and neural network model to explore the correlation between the textural properties of organic linkers from MOF-5 analogues database and the gas adsorption and separation performance on gas mixtures of Xe/Kr.The result indicates that the effect of reduce of pore size resulted from ligand functionalization on adsorbent-adsorbate interaction for Xe is larger than that for Kr.The way to increase the API（Adsorption Performance Indicator）by ligand functionalization is to effectively enhance the polarizability of organic linker on condition of less effect on the size of organic linker.Also,it’s significant to insert a group with low weight and high force field parameter for center atom.Finally,a precise neural network model（R>0.80）was used to predict selectivities of Xe/Kr mixtures based on the size and polarizability of the organic linker.Developing MOFs which prefer to adsorb He can enhance efficiency and purity for He recovery from natural gas.This work performed high-throughput screening technique to generate 1558 MOFs from CoRE MOF database which have PLD value between kinetic dimeter of He and CH₄ in order to only allow helium flow inside the crystal.The effects of various parameters set on the accuracy of neural model were tested to predict working capacity of He in MOFs based on textural properties（Pore size,void fraction and surface area）.Then same model with highest accuracy was used to predict self-diffusivity of helium in MOFs based on textural properties.The RMSE（Root-mean square error）,MAE（Mean absolute error）,MBE（Mean bias error）between target and output for predicting working capacity are lower than 0.1 mmol/g,lower than 0.01 mmol/g,lower than 0.1 mmol/g and higher than 0.97,respectively.Also,the ANN2 model have comparatively precise activity for predicting self-diffusivity of helium in MOFs.The correlation coefficient R value for ANN2model on the diffusivity is about 0.78.Improving adsorption performance of methanol in MOFs contributes to the green economy and protect people from the harm from high concentration of methanol in environment.This work combined N-doping and Li-doping methods to explore the effect of these two functionalization methods on the methanol adsorption performance in Cu-BTC.The result indicates that doping Li atoms onto the N site from N-doped BTC organic linker of Cu-BTC can maintain the original T1 cage.Also,doped Li atoms stay as extra adsorption sites in L3 cage to increase the polarity of L3 cage and methanol uptake from low to high pressure（CuBTC:495.0 cm³（STP）/cm³,3N-CuBTC:526.0 cm³（STP）/cm³）.N-doping and alkaline earth metal doping were used to make up the shortage of low polarity and weak adsorption affinity for methanol at low pressure resulted by too large pore size.The result implies that almost no effect from N-doping on methanol adsorption at low pressur.However,metal doping can increase polarity of the pore by introducing extra adsorption sites and decrease the pore size to dramatically pull up the adsorption uptake of methanol in IRMOF-10（IRMOF-10:2.66 mmol/g,Be-2N-IRMOF-10:49.7mmol/g）.In order to solve the problem that the classical model is hard to precisely describe adsorption behaviors of ethylene and ethane in MOFs.This work performed test on accuracy of various models to describe the adsorption and separation behavior of ethylene and ethane in MOFs from the perspective of adsorption isotherm,heat of adsorption and selectivity.RMSE was used to describe the accuracy of models.The result suggests that the RMSE value would be effectively reduced when polarization effect was considered on the classical TraPPE model.Based on TraPPE₀.20 model for ethylene and TraPPE₀.10 model for ethane,this work focuses on the change of distribution of ethylene and ethane in two MOFs with open metal sites before and after considering polarization effect.The change implies that the preferential adsorption site would be changed from organic linker to open metal sites after consideration of polarization effect.Because the polarization effect can make up forπ-complexation interaction and induction effect.Finally,same TraPPE model with polarization effect was confirmed to be precise to describe gas adsorption behaviors of propylene and propane in MOFs.In the part of adsorption isotherm,RMSE value between simulation and experiment for propylene and propane are similar as that for ethylene and ethane.It indicates that the TraPPE model with polarization effect is a model with general applicability.