Machine Learning Predicts NO Release in Nitroaromatics

Ever wondered how certain chemicals break down when hit by light? Scientists used a smart computer model to figure this out for nitroaromatic compounds. These compounds are known to release a gas called NO when they break down. The scientists used something called Gaussian process regression, a type of machine learning, to predict the energy surfaces of these compounds. This helps understand how the compounds move and change when light hits them. The model looked at two ways the NO could be released: through a process called roaming or through something called the oxaziridine mechanism. By simulating these processes on a simplified energy surface, the scientists found that the way NO is released depends mainly on how the compound moves on a specific energy level, called T1. When they compared their results with real experiments, they found a good match, showing that their model was on the right track.