Assessment of density functional theory for model SN2 reactions:: CH3X+F- (X = F, Cl, CN, OH, SH, NH2, PH2)

The B3LYP, BLYP, and BP86 variants of density functional theory, in conjunction with double and triple zeta basis sets, have been examined for S(N)2 reactions of the type CH3X + F- --> CH3F + X- (X = F, Cl, CN, OH, SH, NH2, and PH2), using the CCSD(T) method in combination with the TZ2Pf+dif and aug-cc-pVTZ basis sets as a reference for comparison, along with experimental calibrations. The functionals perform modestly well, with some preference for B3LYP, in describing the structures of the stationary points, nonetheless exhibiting bond distance deviations as large as 0.24 Angstrom and bond angle deviations as large as 39 degrees. Regarding the energetics, the three functionals perform best for ion-molecule complexation energies (E-X,Y(w)), on average deviating by 1.3 kcal mol(-1). However, the pure functionals are not able to characterize the reaction energies (E-X,Y(0)) and particularly the net activation barriers (E-X,Y(h)) with the same accuracy, with underestimations as large as 11.0 kcal mol(-1) for S(N)2 barriers. The hybrid B3LYP functional significantly outperforms the pure functionals for these same energetic quantities, better approximating the coupled cluster reference by over 4 kcal mol(-1). Still however, B3LYP is only marginally satisfactory. In fact, all of the functionals give S(N)2 transition states which are anomalously too low in energy, revealing the need for reconstructions and reparametrizations for accurate treatments of these Pervasive reactions. Comparison of CCSD(T)/TZ2Pf+dif with CCSD(T)/aug-cc-pVTZ shows the need to use high level extrapolation schemes to describe the energetics, of these types of reactions to within one kcal mol(-1).