The heat of formation of the CH3CO radical has been determined on seve
ral occasions(1-16) (see Table 1). The experimental literature before
1992 supports a value of about -5 kcal/mol for the heat of formation.
Yadav and Goddard(17) studied acetaldehyde and its dissociation using
relatively low levels of theory. While the calculations illustrated th
e character of the potential energy surface, they were incapable of ac
curately determining the heat of formation of CH3CO. The more accurate
calculations of Francisco and Abersold(15) support a heat of formatio
n of around -5 kcal/mol, especially if one takes their value from sche
me 1 (-4.9 kcal/mol) in preference to their average value. That is, th
eir reaction which involves breaking a C-H bond is expected to be more
accurate than their scheme which involved breaking a C-Cl bond, becau
se it is easier to describe a C-H bond than a C-Cl bond. Also in 1991,
Radom and co-workers(16) computed the C-H bond energy in acetaldehyde
using the G1 approach.(18) Their bond energy (at 0 K) was 3.8 kcal/mo
l larger than the experimental value (derived from a heat of formation
(13) at 298 K of -5.4 kcal/mol). Because the G1 approach is usually ac
curate to +/-2 kcal/mol, they suggested that the acetyl radical heat o
f formation was several kcal/mol smaller in magnitude than experiment.
Unfortunately, they did not pursue this suggestion as the acetyl radi
cal was only a minor aspect of their study. Recently, Niiranen et al.(
14) determined a heat of formation of -2.39 +/- 0.29 kcal/mol for CH3C
O from a kinetics study of the reaction CH3CO + HBr. This value suppor
ts the suggestion of Radom and co-workers that the older values are to
o large in magnitude. In this work we determine the heat of formation
of CH3CO using high levels of theory in conjunction with large basis s
ets. In addition, we determine all of the other bond energies in CH3CH
O using the G2(MP2) approach.(19) The G2(MP2) approach combines a high
ly accurate method in small basis sets with a more approximate method
in a large basis set and an empirical correction, and it is therefore
a very cost effective method of computing bond energies accurate to ab
out +/-2 kcal/mol.