Predicting polymer degradation mechanisms and susceptibility is a significant challenge due to the vast quantity of unique polymers and the long time scales required to complete comprehensive degradation studies, sometimes on the order of decades. Alternatively, high throughput analysis and prediction of polymer degradation presents an exciting opportunity to leverage far faster quantum chemistry reaction analysis programs and machine learning to accurately predict relative degradation behavior and, inversely, stability. By utilizing Yet Another Reaction Program (YARP), a reaction enumeration and transition state search algorithm, to identify degradation reactions classified in multiple motifs, experimentally relevant degradation pathways were identified and classified in a fraction of the time required by classical experiments. Furthermore, identification of conserved pathways across multiple materials unlocked tangible similarities across data sets that machine learning leveraged to accurately predict relative degradation susceptibility of materials outside of training. This demonstrated an exciting and promising breakthrough in achieving accurate on-demand predictive metrics for polymer degradation in a variety of environments.
