Reaction barriers of radical hydrogen abstractions

The classical and vibrational adiabatic (zero-point corrected) reaction barriers obtained by various DFT methods for the reaction CH4 + 30 - CH3 + OH are listed in Table 6. Results obtained by conventional correlated ab initio methods as well as the experimental value are included for comparison purposes. The DFT calculations employed the 6-31G(d,p) basis set and the (50,194) grid, with fully optimized geometries. The PMP4 and POL-CI calculations [72] were carried out with larger, triple-zeta-quality basis sets. 

Method Classical barrier Zero-point energy Vibrational adiabatic barrier 

It is surprising to see that the gradient-corrected methods B-VWN and B-LYP, which exhibited the best performance in the previous study, underestimate the reaction barrier by up to 11 kcal/mol. On the other hand, 

B-null gives a barrier in perfect agreement with experiment, much better than PMP4 and as good as POL-CI, which is particularly puzzling given the extremely poor results obtained with B-null in the case of atomization energies. For the methods S-null, S-VWN and S-LYP, no transition structure was found on the potential surface, and thus these methods incorrectly predict a barrierless reaction. 

---