Examining the PM6 semiempirical method for pKa prediction across a wide range of oxyacids

Sierra Rayne,Kaya Forest,Ken Friesen

doi:10.1038/npre.2009.2981.1

Abstract

AbstractThe pK~a~ estimation ability of the semiempirical PM6 method was evaluated across a broad range of oxyacids and compared to results obtained using the SPARC software program. Compound classes under consideration included acetic acids, alicyclic and aromatic heterocyclic acids, benzoic acids, boronic acids, hydroxamic acids, oximes, peroxides, peroxyacids, phenols, α-saturated acids, α-saturated alcohols, sulfinic acids, α-unsaturated acids, and α-unsaturated alcohols. PM6 accurately predicts the acidity of acetic and benzoic acids and their derivatives, but is less reliable for alicyclic and aromatic heterocyclic acids and phenols. α-Saturated acids are reliably modeled by PM6 except for polyacid derivatives with α-alcohol moieties. α-Saturated alcohols only appear to yield reliable PM6 results where an α-hydroxy or α-alkoxy moiety is absent. Carboxylic acids with simple α-alkene unsaturation are well approximated by PM6 except where alkyne α-unsaturation or α-carboxylation are also present. The PM6 and SPARC methods exhibit approximately equal pKa prediction performance for the acetic, alicyclic, and benzoic acids. SPARC outperforms PM6 on the peroxides, peroxyacids, phenols, and α-saturated acids and α-saturated alcohols. pKa values for boron, nitrogen, and sulfur oxyacids do not appear to be reliably estimated by either the PM6 or SPARC methods. The findings will help guide the potential appropriateness of results from the PM6 pK~a~ estimation method for waste treatment and environmental fate investigations.

Highlights

Predicting the acidity constant of compounds is a critical task in designing waste treatment methods and understanding the environmental fate of both contaminants and natural compounds
We have shown that the PM6 method likely underestimates the pKa values for perfluorinated carboxylic acid contaminants [21] and likely overestimates the pKa values for perfluorinated sulfonic acid contaminants [22]
A comparison of pKa values for 68 compounds from the source validation set in the MOPAC 2009 manual [14] and the current work is shown in Figure 2 and given in Electronic Supplementary Material Table S1

Summary

Introduction

Predicting the acidity constant (pKa value) of compounds is a critical task in designing waste treatment methods and understanding the environmental fate of both contaminants and natural compounds. Historical approaches to pKa estimation typically involved linear free energy relationships [1,2,3], of which the Hammett-type correlations are perhaps the best known [4], and fragment-type, one- and two-dimensional, and topological/connectivity index methods [5,6]. These methods often have difficulty dealing with geometrical isomers, new substituent types, and intramolecular hydrogen-bonding effects due to the lack of dependence on three dimensional optimized molecular structures. We investigate the pKa predictive capacity of the PM6 method across a wide range of carbon and non-carbon oxyacids (including environmental contaminants, natural products, industrial compounds, and medicinally active substances) in the hope of better defining the applicability domain of this computational approach

Methods

Results

Conclusion