(16d) Automatic Mechanism Generation for Liquid-Phase Reaction Kinetics

Understanding liquid phase reaction mechanisms is important for several applications, including atmospheric chemistry and hydrocarbon processing. Automatic mechanism generation is a useful tool for learning about these systems, as their reaction networks can contain thousands of reactions and are difficult to deduce by hand. The Reaction Mechanism Generator (RMG)¹is an open-source software tool for predicting kinetic models of large reaction networks. RMG uses chemical knowledge to propose elementary chemical reactions and their rates. The new version of RMG written in Python, RMG-Py, has been used for gas phase mechanism generation up until this point; we have now introduced thermodynamic and kinetic corrections for liquid-phase reactions into RMG-Py.

For every molecule, five solute descriptors are determined using a molecular structure group additivity method first implemented by Platts et al. (1999)². These parameters describe specific interactions between solute and solvent, including electrostatic interactions, cavity formation and dispersion, and hydrogen bond acidity and basicity. Using these solute descriptors, as well as solvent descriptors, partition coefficients between solute and solvent are calculated using the Linear Solvation Energy Relationships (LSERs) of Abraham et al. (1993)³.  The partition coefficients along with the solute and solvent descriptors are then employed to calculate the Gibbs free energy of solvation for all species. These have been compared to known solvation thermodynamics for molecules in the Minnesota Solvation Database (MNSOL)⁴. Enthalpies of solvation can be calculated using a similar method from Mintz et al. (2006)⁵ and together with the Gibbs free energy of solvation, can be used to calculate the entropy of solvation.⁶

To account for diffusion limitations on reaction kinetics, we calculate an effective rate constant for each reaction, which depends on the intrinsic gas-phase reaction rate and the diffusivities of reacting species.  The diffusivities are calculated using the Stokes-Einstein equation, with the reactants’ diameters estimated via group additivity.

The effect of the solvent on the intrinsic reaction rate is also now estimated.  For certain types of reactions, rates of reaction are shown to increase in solvents of increasing dielectric constant. This effect is due to an increased charge transfer in the transition state, meaning the electrostatic stabilizing effect of the solvent is greater in magnitude for the transition state than the reactants, lowering the reaction barrier height⁷. For each reaction family in RMG, the electrostatic effect of the solvent on the activation energy has been identified and implemented, and the reaction rate constants modified.

Using these modified thermodynamics and kinetics, complete reaction mechanisms can be automatically generated for liquid-phase systems. Rate estimates will be more accurate than previously due to the inclusion of kinetic solvent effects. Further work includes more specifically changing the reaction barrier height based on the chemical identity of the reacting species.

References

1. RMG - Reaction Mechanism Generator. Open-source software project, 2013, http://rmg.sourceforge.net/
2. Platts et al., 1999. J. Chem. Inf. Comput. Sci., 39: 835-845.
3. Abraham et al., 1993. Chem. Soc. Rev., 22: 73-83.
4. Marenich et al., Minnesota Solvation Database-version 2009; University of Minnesota, Minneapolis, 2009, http://comp.chem.umn.edu/mnsol.
5. Mintz et al., 2006. J. Chem. Inf. Mod., 47(1): 115-121.
6. Jalan et al., 2013. J. Phys. Chem. B. 117(10): 2955–2970. 
7. Rinaldi et al., 2004. J. Chem. Phys., 120(5): 2343-2350.