(317e) Theoretical Prediction of Organic Reaction Yields and Predicting Retrosynthetic Pathways with Network Chemistry

Free energy and extent of the reaction (yield) are predicted by a thermodynamically guided approach based upon estimation of free energies of compounds. The non-ideality of the system and the effect of solvent are taken into account through the fugacity coefficient calculated by Perturbed-Chain Induced-Polar Statistical Associating Fluid Theory (PCIP-SAFT) molecular theory. A structure-based formulation is defined for the Gibbs free energy and the yield is obtained from the relationship between the free energy of reaction and equilibrium constant considering temperature effects. The model is trained using a database containing 10000 cited reactions. Regression of computed Gibbs free energy and yield of reaction versus experimental values resulted in a coefficient of determination (R2) of about 0.95. The absolute average deviation of the predicted yield is ~0.15 that has been validated against sets of 2000 reactions. Finally, the proposed theory is implemented to predict retrosynthetic pathways of chemical reactions with Chematica which is the most advanced network chemistry package.