(448c) Thermochemistry of Organic Ions in Different Solvents

The thermochemistry of organic ions includes many chemically important quantities, such as acid dissociation constants (pK_a values), pH-dependent solubilities, and rate constants. These properties are closely related to solvation energies. Recent versions of solvation models are able to predict solvation energies of neutral molecules to well within 1 kcal/mol, and solvation energy data compilations include several thousands of entries. But the model performance and data availability for charged solutes are comparatively worse: standard errors exceed 4 kcal/mol, and published solvation energies number in the dozens or low hundreds rather than thousands.

In this talk, we address challenges in both modeling and data availability. We show how existing solvation models can be used to accurately compute relative thermochemical properties, despite their poor absolute predictive ability. Separately, we present a new semi-experimental ionic solvation free energy dataset that can be used to estimate the energies of solvated ions.

We discuss how pK_a values in different solvents can be accurately calculated with standard solvation models by reference to pK_a data in water. We then show how these pK_a calculations are accurate enough to infer the logS – pH curve of four amino acids in non-aqueous solvents using the Henderson-Hasselbalch equation. As a further demonstration, we discuss how model calculations can be used to predict rate constants of ionic reactions in different media.

We further present our experimental dataset (IonSolv) which includes 300+ experiment-based solvation free energies in water, and approximately 300 semi-experimental solvation energies for anions in acetonitrile, DMSO, or DMF. This represents the largest ionic solvation energy dataset available, and is the first to report such values in DMF. To generate the dataset, we used a thermodynamic cycle that relates the pKa, gas-phase acidity, and solvation energy of the neutral conjugate acid or base. Those values were taken from experimental data except for the solvation energies of the neutral conjugates within DMSO, DMF, and acetonitrile; those energies were calculated using COSMO-RS, which has previously been reported to predict solvation energies of neutral solutes to standard uncertainties within 0.5 kcal/mol.

This work could benefit drug screening and synthesis of ionizable compounds. The solvation of ions is directly implicated in crucial properties like solubility and log(D). Our demonstrations show how existing solvation models can be used in several such practical cases involving ions. We hope that our released solvation free energy dataset will motivate future researchers to develop even more powerful methods for investigating ionic solvation phenomena.