Thermosalient Phase\nTransitions from Machine Learning\nInteratomic Potential

We\ndeveloped an accurate machine learning interatomic potential\nfor the thermosalient molecular crystal <i>N</i>-2-propylidene-4-hydroxybenzohydrazide.\nThis crystal exhibits one of the largest mechanical responses during\nits thermosalient phase transition. Leveraging the speed of our developed\npotential, we performed Gibbs free energy calculations that successfully\npredict phase transitions in good agreement with experimental observations.\nAdditionally, our model accurately captures the phenomenon of negative\nlinear thermal expansion preceding the thermosalient phase transition.\nWe show that the energy barrier exists at phase transition temperature\nand that this energy is purely elastic, elucidating the physical reasons\nfor the thermosalient effect.