Calculating accurate proton chemical shifts of organic molecules with density functional methods and modest basis sets.

The purpose of this paper is to convince practitioners of (1)H NMR spectroscopy to consider simple quantum chemical calculations as a viable option to aid them in the assignment of their spectra. To this end, it is demonstrated, on a test set of 80 conformationally stable molecules of various kinds carrying different functional groups, that, in contrast to what is claimed in the literature, large basis sets are not needed to obtain rather accurate predictions of (1)H NMR chemical shifts by quantum chemical calculations. On the other hand, modeling the solvent by an SCRF-type calculation may improve certain predictions significantly. The best accuracy/cost ratio is provided by GIAO calculations in chloroform as a solvent with the specially parametrized WP04 functional of Cramer et al. using the cc-pVDZ or 6-31G** basis set, closely followed by similar calculations with the ubiquitious B3LYP functional (both predict (1)H chemical shifts with an average deviation of ca. 0.12 ppm, if the results are scaled linearly). A slightly higher accuracy can be attained by adding diffuse functions to the basis set, but going to the triple-zeta basis sets which have invariably been used hitherto in calculations of chemical shifts does not lead to any improvement. The popular increment schemes such as those implemented in the ChemDraw or ACD programs do not do nearly as well and are often incapable of correctly distinguishing stereoisomers.