Benchmarking Continuum Solvent Models for Keto–Enol Tautomerizations

Experimental free energies of tautomerization,
ΔGT, were used to benchmark the
gas-phase predictions of
17 different quantum mechanical methods and eight basis sets for seven
keto–enol tautomer pairs dominated by their enolic form. The
G4 method and M06/6-31+G­(d,p) yielded the most accurate results, with
mean absolute errors (MAE’s) of 0.95 and 0.71 kcal/mol, respectively.
Using these two theory levels, the solution-phase ΔGT values for 23 unique tautomer pairs composed of aliphatic
ketones, β-dicarbonyls, and heterocycles were computed in multiple
protic and aprotic solvents. The continuum solvation models, namely,
polarizable continuum model (PCM), polarizable conductor calculation
model (CPCM), and universal solvation model (SMD), gave relatively
similar MAE’s of ∼1.6–1.7 kcal/mol for G4 and
∼1.9–2.0 kcal/mol with M06/6-31+G­(d,p). Partitioning
the tautomer pairs into their respective molecular types, that is,
aliphatic ketones, β-dicarbonyls, and heterocycles, and separating
out the aqueous versus nonaqueous results finds G4/PCM utilizing the
UA0 cavity to be the overall most accurate combination. Free energies
of activation, ΔG‡, for the
base-catalyzed keto–enol interconversion of 2-nitrocyclohexanone
were also computed using six bases and five solvents. The M06/6-31+G­(d,p)
reproduced the ΔG‡ with MAE’s
of 1.5 and 1.8 kcal/mol using CPCM and SMD, respectively, for all
combinations of base and solvent. That specific enolization was previously
proposed to proceed via a concerted mechanism in less polar solvents
but shift to a stepwise mechanism in more polar solvents. However,
the current calculations suggest that the stepwise mechanism operates
in all solvents.