Are calculated enthalpies of formation sometimes more reliable than experimental? A test on alkyl substituted benzoic acids

Energies of 20 alkyl‐substituted benzoic acids were calculated at the levels B3LYP/6‐311+G(d,p)//B3LYP/6‐311+G(d,p) and MP2/6‐311+G(d,p)//MP2/6‐311+G(d,p); the pertinent enthalpies at 298 K were calculated at the same levels. Comparison with experimental enthalpies of formation Δ f H ° g (298) was carried out in terms of isodesmic reactions, that is, in the relative values. Of the four calculated quantities, the DFT enthalpies yielded best correlation with the standard deviation of 2.1 kJ mol −1 , near to the experimental uncertainty; the DFT energies are only slightly worse and the MP2 enthalpies or energies much worse. However, the DFT method overestimated systematically the substituent effects and had to be calibrated. Comparison with the experimental gas‐phase acidities was less telling and the fit was worse because both methods overestimated the substituent effects. Extending the base in selected examples did not give better results. Although the systematic deviations are evidently due to the imperfections of the theoretical models, individual big deviations should be attributed to experimental errors or to the abnormal behavior of certain compounds at the experimental conditions. From this point of view, three examples of the so‐called long‐range effect claimed in the case of different benzoic acid derivatives, always for substituents in the meta position, must be refused as unproven because the experimental energies were not confirmed by calculations. © 2006 Wiley Periodicals, Inc. J Comput Chem 5: 571–577, 2006