Radical ring opening reactions constitute critical steps in several industrially significant synthetic pathways, such as natural product synthesis and combustion reactions. Accurately predicting the enthalpies of these reactions is of great importance for designing reaction pathways. In light of this, a machine learning model for predicting bond dissociation enthalpies of both the ring-opening radical formation step and the radical propagation hydrogen-abstraction step of the two step ring opening reaction was developed.<br/>Initially, reaction enthalpies were collected from the existing body of literature to perform benchmark tests.[1] These were then supplemented with enthalpies calculated using DFT methods to create a training and test set to train neural network based deep learning algorithms. Several models were tested with a random sample from the database used for model training and the remaining values used for model validation. Comparison with existing models was performed to ensure reliability and accuracy for a range of reaction substrates. Ultimately, model results were used to gain insights into specific chemical structural features influencing the enthalpy of ring opening reactions.

Machine learning based model for ring opening bond dissociation enthalpies

ACS Spring 2022

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.