Detailed kinetic models provide useful mechanistic insight into a chemical system. Manual construction of such models is laborious and this has led to the development of automated methods for exploring chemical pathways. These methods rely on fast, high throughput estimation of species thermochemistry and kinetic parameters. Most automatic mechanism generation tools have been developed to model gas phase systems.
We discuss methods for extending the idea of automatic mechanism generation to solution phase reactions using linear solvation energy relationships (LSERs) for estimation of solvent effects on thermochemical and rate parameters. Specifically, we employ and validate the Abraham model (using descriptors estimated by group additivity) to estimate the free energy of solvation for several solute and solvent classes. First order temperature dependence of the solvation free energy is modeled by decomposing it into enthalpic and entropic parts. Two approaches to perform this decomposition are discussed: one derived from a hard sphere model of liquids and another recently proposed empirical models by Mintz and co-workers. Ease of implementation, extensibility to different solvents and accuracy of estimates are used as metrics for comparison.
The descriptor based approach can also be extended to model kinetic parameters as suggested by existing empirical correlations of solvent effects on elementary reaction rates. Continuum solvation models provide an alternative method for determining solvent effects in cases with little or no experimental data. We discuss solvent effects on free radical reactions relevant to hydrocarbon oxidation using previous experimental work and continuum solvation calculations. In addition to kinetic solvent effects, diffusion-limits of various reactions are estimated on the fly using known expressions from literature.
These tools have been incorporated into the open source Reaction Mechanism Generator (RMG) software and applied to generate kinetic models of AIBN initiated hydrocarbon oxidation in different solvents. The computer-generated models accurately replicate the solvent effects on oxidation measured experimentally, and can be used to predict the behavior in new solvents and at different temperatures.