文摘
Density Functional Theory DFT) is a widely used method in quantum mechanics for modeling atoms and molecules. Commonly used DFT functionals have many shortcomings that include a poor description of dispersion,molecular geometries,exchange-repulsion,and hydrogen-bond interactions. To improve the quality of DFT,one popular idea is to apply empirical corrections to existing density functionals. Such an approach is both conceptually simple and computationally affordable. Despite many successful applications,most existing DFT empirical correction methods focus only on the dispersion corrections. In this thesis,we introduce system-specific empirical corrections to DFT. Our method not only provides corrections for dispersion,but also addresses problems such as deficiencies with molecular geometries,exchange-repulsion,and hydrogen bonding. The empirical correction,named "supplemental potential" SP),is created by fitting the force differences between a functional and a high quality post-Hartree-Fock method. We tested the performance of SPs for three types of systems: water,methane-water,and molecular crystals. For the water system,the Becke-Lee-Yang-Parr BLYP) functional description of the water potential energy surface PES) can be improved to coupled-cluster quality with our water SP. For H2O)n n=1-6),the relative cluster energies,cluster binding energies,and optimized energy structures are correctly predicted with the water SP correction. The developed methane-water SP is able to improve the BLYP PES to coupled-cluster quality in the study of methane water system. In the molecular crystal studies,the DFT-SP method correctly predict the most stable crystal structures among the sets of low-energy polymorphs,for four out of five studied organic molecules.