Comparison of multistandard and TMS-standard calculated NMR shifts for coniferyl alcohol and application of the multistandard method to lignin dimers.

Coniferyl alcohol is a monomeric building block of lignin, the second most abundant biopolymer. During lignification, the monomer forms a variety of linkages through free radical additions. A large NMR database has been constructed that reports the (1)H and (13)C chemical shifts for thousands of lignin oligomers. Herein, Boltzmann averaged (1)H and (13)C GIAO NMR calculations were performed on coniferyl alcohol and four of its dimers, β-O-4, β-β, β-5, and 5-5, to compare the calculated chemical shifts with experiment. Six B3LYP/6-311++G(d,p) energy-minimized conformational isomers of coniferyl alcohol were subjected to single-point GIAO NMR calculations. Initially, four NMR shift calculation methods were compared: three were performed using the TMS-standard method at the HF/6-311+G(2d,p), B3LYP/6-311+G(2d,p), and mPW1PW91/6-31G(d) theory levels, and the fourth was performed with a multistandard approach using a mPW1PW91/6-31G(d) theory level. For the multistandard method, benzene was used as the standard for aromatic C and H atoms and methanol was used for aliphatic C and H atoms. The hydroxyl-H of methanol was used as the standard for hydroxyl-H atoms. The Boltzmann averaged results for six conformers showed that the multistandard method is more accurate for coniferyl alcohol and its dimers than the often used TMS-standard method, based on the mean unsigned, root-mean-squared, and maximum errors, as well as linear correlations between observed and calculated values. The (13)C results were more accurate than the (1)H results, due to poorer agreement between calculated hydroxyl-H results and observed data. Further Boltzmann-averaged, multistandard NMR calculations compared the (13)C and (1)H chemical shifts with experiment for the four stereoisomers of the β-O-4 dimer, as well as the 5-5, β-5, and β-β dimers of coniferyl alcohol. The (13)C results correlated well with experiment (r(2)>0.99) for all dimers and showed small statistical errors, compared with experiment. The correlation with experiment for (1)H NMR was generally inferior to the (13)C NMR results for the dimers.