DFTB對H2，O2，H2O與H2O2的自動參數化

Abstract

我的論文致力於兩個重要的問題：（i）開發一組Slater-Koster參數，能夠精確地再現H2O2中的二面角，並且能夠以高準度預測H2，O2，H2O和H2O2的所有分子性質，（ii）通過測試，報告和指出程式錯誤以及提供合適的解決方案來協助開發自動DFTB參數化工具包（ADPT）。 我們的參數已經成功地計算出H2，O2，H2O和H2O2的分子性質。優化後的參數所計算的H2O2二面角僅與參照數據相差0.1度，並同時精確的計算出H2，O2和H2O的基態幾何形狀，平衡鍵長的誤差小於0.001埃。另外，在測試與修繕ADPT後，此工具包已能為沒有DFTB參數化經驗和曾經手動進行參數化的用戶提供了更多的靈活性與可用性。 針對水簇（water clusters）的測試，我們的參數計算的氫鍵結合能之平均絕對誤差為4.29 kcal/mol。建議水二聚體（water dimer）中所有可能的分子間運動之能量分佈納入參數化時所使用的參照數據，將能改善我們的參數在計算水簇結合能與基態幾何形狀的誤差。

My Thesis is devoted to 2 important issues: (i) developing a set of the Slater-Koster parameters capable of reproducing accurately dihedral angle in H2O2 and simultaneously capable to predict all molecular properties of H2, O2, H2O, and H2O2 with high fidelity and (ii) assisting the development of the automatic DFTB parameterization toolkit (ADPT) by testing, reporting and locating bugs, and suggesting suitable solutions. The molecular properties of H2, O2, H2O, and H2O2 have been satisfactorily reproduced with our parameters. The optimized parameters give an error in the dihedral angle of H2O2 of −0.1 degree and simultaneously reproduce accurate equilibrium geometries for H2, O2, and H2O, with the errors smaller than 0.001 angstrom. After including the changes proposed by me, ADPT provides more flexibility to users who have no experience with DFTB parameterization and those who used to do the parameterization manually. The optimized parameters have been additionally tested against the ab initio data for water clusters. The mean absolute error for the binding energy per hydrogen bond computed with our parameters is 4.29 kcal/mol. The suggestion for improving the performance of our parameters is based on including in the reference data the energy profiles for all possible intermolecular motions in water dimer.