RADICAL-ADDITION TO ALKENES - FURTHER ASSESSMENT OF THEORETICAL PROCEDURES

Ab initio molecular orbital calculations at a variety of levels of the ory have been carried out for a number of prototypical radical additio n reactions with a view to determining a level of theory suitable for predicting reliable barriers. Closest agreement with experimental barr iers is achieved with a variant of the recently introduced CBS-RAD pro cedure. At this level, the mean absolute deviation from experimental b arriers for methyl radical additions in solution is just 1.4 kJ mol(-1 ). A second high-level theoretical procedure examined is a variant of G2(MP2,SVP), corresponding effectively to QCISD(T)6-311+G(3df,2p) ener gy calculations on QCISD/6-31G(d) optimized geometries and incorporati ng scaled B3-LYP/6-31G(d) zero-point vibrational energy corrections. A t this level, the mean absolute deviations from the experimental barri ers is significantly larger at 7.7 kJ mol(-1), the calculated barriers being consistently too high. The effect of quadruple excitations is f ound to be small. The considerably less expensive B3-LYP/6-311+G(3df,2 p)//B3-LYP/6-31G(d) procedure performs quite well, with a mean absolut e deviation of about 5.6 kJ mol(-1). Solvent effects were estimated us ing the SCIPCM model. For a dielectric constant of 2 (nonpolar medium) , the effect on barrier ranges from -1.1 to +1.1 kJ mol(-1), while for a dielectric constant of 40 (polar medium), the effects range from -3 .0 to +2.8 kJ mol(-1).