Assessment of experimental bond dissociation energies using composite ab initio methods and evaluation of the performances of density functional methods in the calculation of bond dissociation energies.

Composite ab initio CBS-Q and G3 methods were used to calculate the bond dissociation energies (BDEs) of over 200 compounds listed in CRC Handbook of Chemistry and Physics (2002 ed.). It was found that these two methods agree with each other excellently in the calculation of BDEs, and they can predict BDEs within 10 kJ/mol of the experimental values. Using these two methods, it was found that among the examined compounds 161 experimental BDEs are valid because the standard deviation between the experimental and theoretical values for them is only 8.6 kJ/mol. Nevertheless, 40 BDEs listed in the Handbook may be highly inaccurate as the experimental and theoretical values for them differ by over 20 kJ/mol. Furthermore, 11 BDEs listed in the Handbook may be seriously flawed as the experimental and theoretical values for them differ by over 40 kJ/mol. Using the 161 cautiously validated experimental BDEs, we then assessed the performances of the standard density functional (DFT) methods including B3LYP, B3P86, B3PW91, and BH&HLYP in the calculation of BDEs. It was found that the BH&HLYP method performed poorly for the BDE calculations. B3LYP, B3P86, and B3PW91, however, performed reasonably well for the calculation of BDEs with standard deviations of about 12.1-18.0 kJ/mol. Nonetheless, all the DFT methods underestimated the BDEs by 4-17 kJ/mol in average. Sometimes, the underestimation by the DFT methods could be as high as 40-60 kJ/mol. Therefore, the DFT methods were more reliable for relative BDE calculations than for absolute BDE calculations. Finally, it was observed that the basis set effects on the BDEs calculated by the DFT methods were usually small except for the heteroatom-hydrogen BDEs.