Chemometrical analysis of substituent effects. XIII. Comparison of substituent effects on dissociation and chemical shift in C-13 NMR spectra of mono- and disubstituted benzoic acids

The C-13 chemical shifts have been measured of the carboxyl carbon atoms fo r all the 2-, 3-, and 4-substituted benzoic acids with H. CH3, CH3O, F, Cl, Br, I, and NO2 substituents, as well as for all 3,4-, 3,5-, and 2,6-disubs tituted benzoic acids with combinations of CH3, CH3O, Cl (or Br). NO2 subst ituents and for symmetrically 2,6-disubstituted derivatives with Et, EtO, P rO, i-PrO, and BuO substituents. The chemical shifts of carboxylic group ca rbon atoms of the 3- and 4-substituted derivatives show correlation only wi th the substituent constants sigma(1). For the 2-substituted derivatives wa s found the dependence only on sigma(1) and on the ir constant describing s teric effects (s = 0.122, R = 0.996, without the CH3 derivative which has a distinct anisotropic effect). The substituent effects on the carboxylic ca rbon chemical shift show additivity with 3.4-. 3.5, and 2,6-substituents, a nd the 2,6-disubstituted derivatives show a linear synergic effect of subst ituents due obviously to the steric hindrance to resonance. Application of the principal component analysis to the data matrix involving all the combi nations of mono- and disubstitution involving, the above-mentioned substitu ents has proved an identical substituent effect from ail the positions on t he chemical shift described by one latent variable, steric effects and anis otropic behaviour of methyl at the 2 and 2,6 positions being predominantly described by the second latent variable (with the total explained variabili ty of 99.5%). Comparison of substituent effects on the chemical shift of ca rboxylic carbon with that on the dissociation constant measured in the same solvent has confirmed the anisotropy due to ortho methyl group, the ortho halogen substituents in monosubstituted derivatives also hating a different effect, The dependence of chemical shift on pK(a) was not very close for t he derivatives studied (s = 1.005, R = 0.690). The inclusion of anisotropy of ortho alkyl group by means of an Indicator variable improved the correla tion (s = 0.533, R = 0.925), and omitting of 2-F, 2-Cl, 2-Br, and 2-I subst ituents gave a regression without deviating paints (s = 0.352, R = 0.968).