The density functional theory （DFT） is the most popular method for evaluating bond dis- sociation enthalpies （BDEs） of most molecules. Thus, we are committed to looking for alternative methods that can balance the...The density functional theory （DFT） is the most popular method for evaluating bond dis- sociation enthalpies （BDEs） of most molecules. Thus, we are committed to looking for alternative methods that can balance the computational cost and higher precision to the best for large systems. The performance of DFT, double-hybrid DFT, and high-level com- posite methods are examined. The tested sets contain monocyclic and polycyclic aromatic molecules, branched hydrocarbons, small inorganic molecules, etc. The results show that the mPW2PLYP and G4MP2 methods achieve reasonable agreement with the benchmark val- ues for most tested molecules, and the mean absolute deviations are 2.43 and 1.96 kcal/mol after excluding the BDEs of branched hydrocarbons. We recommend the G4MP2 is the most appropriate method for small systems （atoms number≤20）; the double-hybrid DFT methods are advised for large aromatic molecules in medium size （20≤atoms number≤50）, and the double-hybrid DFT methods with empirical dispersion correction are recommended for long-chain and branched hydrocarbons in the same size scope; the DFT methods are ad- vised to apply for large systems （atoms number〉50）, and the M06-2X and B3P86 methods are also favorable. Moreover, the differences of optimized geometry of different methods are discussed and the effects of basis sets for various methods are investigated.展开更多
文摘The density functional theory （DFT） is the most popular method for evaluating bond dis- sociation enthalpies （BDEs） of most molecules. Thus, we are committed to looking for alternative methods that can balance the computational cost and higher precision to the best for large systems. The performance of DFT, double-hybrid DFT, and high-level com- posite methods are examined. The tested sets contain monocyclic and polycyclic aromatic molecules, branched hydrocarbons, small inorganic molecules, etc. The results show that the mPW2PLYP and G4MP2 methods achieve reasonable agreement with the benchmark val- ues for most tested molecules, and the mean absolute deviations are 2.43 and 1.96 kcal/mol after excluding the BDEs of branched hydrocarbons. We recommend the G4MP2 is the most appropriate method for small systems （atoms number≤20）; the double-hybrid DFT methods are advised for large aromatic molecules in medium size （20≤atoms number≤50）, and the double-hybrid DFT methods with empirical dispersion correction are recommended for long-chain and branched hydrocarbons in the same size scope; the DFT methods are ad- vised to apply for large systems （atoms number〉50）, and the M06-2X and B3P86 methods are also favorable. Moreover, the differences of optimized geometry of different methods are discussed and the effects of basis sets for various methods are investigated.
