Exploring the Reaction Network of Acetic Acid in Supercritical Water via Machine Learning Interatomic Potential

A quantum-trained reactive machine-learning potential maps acetic-acid oxidation in high-temperature, high-pressure supercritical water. Evaluation against one conventional model shows better recovery of experimentally grounded pathways involving short-lived fragments and complete oxidation. Its barrier estimates depend on the training data's coverage of barrier-region structures.