Correlations between experimental and theoretical adiabatic ionization energies for organic compounds and rate constants for atmospheric reactions with hydroxyl radicals

Adiabatic ionization energy (AIE) calculations were performed at the AM1, PM3, PM6, PDDG, HF/QZVP, and B3LYP/QZVP levels of theory on 722 atmospherically relevant organic compounds with available experimental kOH . From the starting set of molecules, a final suite of 114 monoand polyfunctionalized compounds provided converged neutral and cationic geometries without imaginary frequencies for all six levels of theory. NIST evaluated AIEs were available for 54 compounds, providing mean absolute AIE prediction errors of 0.31 (AM1), 0.28 (PM3), 0.50 (PM6), 0.36 (PDDG), 1.22 (HF/QZVP), and 0.20 eV (B3LYP/QZVP). Modest correlations were found between the experimental (r=-0.68, SE=0.81) and computationally estimated (r=-0.77 [AM1], -0.75 [PM3], 0.83 [PM6], -0.79 [PDDG], -0.83 [HF/QZVP], and -0.82 [B3LYP/QZVP]; SE=0.75 [AM1], 0.78 [PM3], 0.66 [PM6], 0.73 [PDDG], 0.67 [HF/QZVP], and 0.68 [B3LYP/QZVP]) AIEs and the corresponding experimental log kOH . Univariate AIE versus kOH correlations are of lower predictive ability than state-of-the-art multivariate techniques, and are limited by the inability to calculate reliable AIEs for large numbers of atmospherically relevant compounds due either to convergence failures at various levels of theory or the presence of imaginary frequencies for converged cationic geometries.