Twelve organic reactions (six closed shell and six radical) were studied using semiempirical, traditional ab initio and density functional methodologies. Full geometry optimizations of all species, both minima and transition states, were performed, and calculated geometries and barrier heights compared with experimental data. Our results demonstrate that although currently available density functionals tend to underestimate barrier heights, especially for radical reactions—in some cases reactions with low barriers are predicted to be essentially barrier free—they provide a significant improvement over standard methods. The adiabatic connection method recently proposed by Becke [J. Chem. Phys. 98, 5648 (1993)], in which a portion of the exact Hartree–Fock exchange is mixed in to the density functional, looks very promising.

No abstract

Drawing upon a large body of recent work covering a wide range of chemistry and comparing results obtained using local and nonlocal density functionals, as well as comparisons with traditional ab initio techniques such as Hartree-Fock and MP2, we demonstrate that hybrid HF-DFT functionals -exemplified here by Becke's original 3-parameter ACM functional -are the best density functionals currently available, certainly for organic chemistry and possibly in other areas as well. Results with the ACM functional are typically of better quality than MP2 and only marginally more expensive computationally than standard Hartree-Fock, making ACM the method of choice for accurate, cost-effective ab initio computations. The use of DensityFunctional Theory (DFT) [1,2] for the calculation of molecular structures, properties and energetics has exploded during the past few years. There have been major advances in both theory -the development of nonlocal functionals (to better model the effects of rapidly changing densities near atomic nuclei) [3-5], the advent of analytical gradients [6-10] and second derivatives [11-14] -and in the practical tools for applying the theory -computational codes such as DMOL [15], DGAUSS [16] and ADF [17] are now well established and are becoming increasingly more sophisticated; traditional ab initio packages such as GAUSSIAN [18], CADPAC [19]and TURBOMOLE [20] are including DFT capability as a major part of their functionality. In fact DFT has now become so prevalent and its applicability and the quality of its predictions so great that in the not too distant future it is likely to take over from Hartree-Fock (HF) as the "basic" ab initio technique. One has only to glance through the recent chemical literature to note the increasing number of publications with a DFT component and the growing number of groups doing work in this area.There are two major reasons for the increasing use of density functional methods in quantum chemistry. One -already mentioned in the proceeding paragraphis that the results are good. The second, and perhaps the most important reason, is that -certainly compared to the more accurate ab initio techniques -the method is fast. Depending on the implementation, DFT calculations range from being 2 or 3 times more costly than standard Hartree-Fock to an order of magnitude or more faster, with the relative speed advantage becoming even greater with increasing system size. Since DFT includes electron correlation, we have a method that is potentially much more accurate than traditional HF and yet is, at worst, no more expensive for large systems and can in fact be made many times faster.

No abstract

Our recent simple model for predicting the principal site for nucleophilic substitution in aromatic perfluorocarbons (J. Fluorine Chem. 2005, 126, 727) is generalized to include both ring-nitrogen atoms and non-fluorine ring substituents. The model is based on the relative stabilities of the Meisenheimer complexes as calculated using Hartree–Fock theory with a modest basis set. Calculations on a wide range of fluorine-containing aromatics demonstrate the general applicability of the model; in over 70 systems examined, we found only one where the predicted primary substitution site did not agree with the experimental findings. We demonstrate that criticism by Chambers et al. to the effect that the model is incapable of reproducing experimental substitution patterns, and, in particular, cannot distinguish between the different activating effects of ortho- and para-fluorines, are entirely unfounded. The observed substitution patterns for six reactions involving attack by aniline on perfluoropyridine and various non-fluorine-substituted derivatives, selected by Chambers et al. to highlight the failings of our model, are, on the contrary, accurately predicted by it.