Training machine learning models based on the structural formula for the enthalpy of vaporization and sublimation and a thorough analysis of Trouton’s rules

Abstract

To convert a calculated value of the enthalpy of formation from the gas phase to room temperature, the enthalpy of sublimation and vaporization are needed. Numerous methods to predict the enthalpies are available, ranging from complex quantum methods to simple empirical correlations to the melting or boiling points. In this study, two machine learning models have been introduced to quickly predict the enthalpy of sublimation and vaporization. To put the results into perspective, a comparison with two empirical models correlating the named enthalpies to the melting/boiling temperature is made.