Hybrid density functional methods empirically optimized for the computation of ¹³C and ¹H chemical shifts in chloroform solution

Two hybrid generalized-gradient approximation density functionals, WC04 and WP04, are optimized for the prediction of ¹³C and ¹H chemical shifts, respectively, using a training set of 43 molecules in chloroform solution. Tests on molecules not included in the training set, namely six stereoisomeric methylcyclohexanols and a β-lactam antibiotic, indicate the models to be robust and moreover to provide results more accurate than those from equivalent B3LYP, PBE1, or mPW1PW91 calculations, particularly for the prediction of downfield resonances in nuclear magnetic resonance spectra. However, linear regression of the B3LYP, PBE1, and mPW1PW91 predicted values on the experimental data improves the accuracy of those models so that they are comparable to WC04 and WP04.