The Application Of Molecular Dynamics In The Study Of Intermolecular Interactions

In this thesis, we focus on the developments of TEAM force field and the applicationsof computational chemistry in the study of interactions between gas molecules andorganic aromatic molecules, and predictions of thermodynamic properties forpolymers with the quantum mechanics (QM) and force field (FF) as cores. Theinvolved materials include carbon dioxide (CO2); aromatic organism benzene (C6H6),pyridine (C5H5N), pyrrole (C4H5N); fluoroalkanes, chloroalkanes; and polymerspolyethylene, polypropylene, polystyrene, etc. The employed computational methodsinclude ab initio QM theory (HF, DFT, MP2and CCSD(T)) and classic moleculardynamics (MD). The research topics involve the study of π-π interactions betweenCO2and a series of aromatic molecules C6H6, C5H5N, C4H5N; the development ofTEAM force field for fluoroalkanes and chloroalkanes, and the accurate prediction ofsolubility parameters (SP) for polymers. These works include the following resultsand conclusions.1. Firstly, we studied the π-π interactions between CO2and aromatic moleculesC6H6, C5H5N, C4H5N by using quantum mechanics methods. π-π interactionsexisted in the top-on configurations, where CO2is positioned on top of thearomatic ring plane. Four top-on stable configurations without imaginaryfrequencies were found by using QM optimization methods. We employed aseries of methods to calculate the binding energies (BEs) of the fourconfigurations. The CCSD(T)/CBS was used to calibrate HF, DFT(B3LYP、PBE�'�TPSS), and MP2with resolution of the identity approximation calculations.Results at the MP2/def2-QZVPP level showed the smallest deviations (only about1kJ/mol) compared with those at the CCSD(T)/CBS level of theory. Although theCCSD(T)/CBS level of theory can provide the best results, it is much expensive than others. Thus, MP2/def2-QZVPP can be found as an alternative method forcalculations in related work. In our study, the aromaticities of three aromaticmolecules were calculated by using NICS method. Through the comparisons oftheir aromaticities and BEs, we found a link between the two. In general, thestrengths of the π-π interactions became stronger as the aromaticities of themolecules increased. In addition, compared with hydrogen-bond or electrondonor–acceptor interactions observed during BEs calculations, π-π interactionssignificantly contributed to the total interactions between CO2and aromaticmolecules. Because C5H5N is the part of Zeolitic Imidazolate Frameworks (ZIFs),our work is helpful in the study of CO2adsorption in ZIFs.2. We developed reliable and accurate force field, which can precisely describe thephysical and chemical properties of fluoroalkanes and chloroalkanes by fitting theQM and experimental data of them. The guiding ideology of the TEAM forcefield development is transferability, extensibility, accuracy, and modularized. Inthe TEAM force field, the definition of the atom type is based on the structure ofthe atom, which can guarantee the accuracy and completeness of both moleculardynamics simulation and the transfer of the force field parameter. To capture thebond parameters and charge parameters, we adopted least square method to fit theQM data of molecular structure, energies and Mulliken charge. And theexperimental data of densities and vapor enthalpies in standard state were used toobtain the van der Waals parameters. This developed force field can be veryhelpful to study the properties and applications of fluoroalkanes andchloroalkanes in the future.3. We predicted the SPs of seventeen common amorphous polymers by using classicMD. It was found that the calculated SPs could be only agreeable withexperimental data when the polymers were about at5Polymerization Degree (PD)in the room temperature. The values of SP decreased with the increment of PD.This could be explained that the intramolecular interaction, which should beovercome in the process of solution, became a larger component of the totalovercome interactions as the PD growed. However, the MD simulation usually calculated the intermolecular interaction and neglected the intramolecular part.Therefore, the classic MD could not accurately predict the SPs of polymers. Tosolve this problem, considering the physically significance, we proposed amathematical formula to revise the calculated values of SP. This formula requiresthe polymer SP at1PD; and the chain length, radius of gyration (ROG), SP of thepolymer at n PD. The results indicated a good agreement of revised SPs withexperimental one for a majority of polymers. For other polymers, the revised SPsare higher than experimental ones. We supposed this overestimation has theconnection with the crystallinity of polymer itself. To verify our supposal, wechose polytetrafluoroethylene, which is representative for the polymers owinghigh crystallinity, to run MD simulations in a high temperature. This temperatureis higher than the melting point of polytetrafluoroethylene, and supposed to breakits high crystallinity. The revised SPs in the high temperature showed muchsmaller deviations with experimental one than the deviations in the roomtemperature, confirming our supposal that crystallinity can greatly affectcalculated and revised values of SP. This work has important implications forresearch on polymers, and solves the problem of underestimations of SPs in MDsimulation.