Thermodynamic Modelling by using Ab Initio and COSMO Calculation

In process engineering, knowledge of phase equilibria and thermochemical properties is essential for sizing multiphase reactors and unit operations used to separate products, such as liquid-vapor flash separators, decanters, distillation or liquid-liquid extraction columns. Phase equilibria play a very important role in other fields of research, such as the agri-food and pharmaceutical industries: engineers are interested in the solubility of constituents in different materials or solvents: for example, they are interested in the phenomenon of permeation of a molecule through packaging, or the solubility of polluting and harmful molecules in different media. In the field of energy efficiency, it is crucial to know the thermodynamic properties (vapor pressures, vaporization enthalpies, heat capacities...) and phase behavior of refrigerants, in order to design new thermal machines, or to determine the most appropriate working fluid for a given application. Specialists in combustion, chemical kinetics or process safety are interested in properties such as heat capacities, flash points and molecular enthalpies of formation. The European REACH regulation aims to list the physico-chemical properties of all molecules, in order to improve environmental protection. However, there are still thousands of molecules for which the physico-chemical properties are not known. This is the case, for example, of oxygenated molecules derived from biomass, for which thermodynamic data and phase behavior have never been measured. Although most experimental techniques for determining these properties exist and have been proven, experimental measurements can be costly, difficult or time-consuming.

Predictive models are therefore particularly useful in research and development, as part of a pre-study or molecular screening. For example, they can be used to perform initial process sizing, to make decisions on the choice of a new molecule such as a new drug or a new refrigerant, or of a solvent to perform an extraction. Predictive models are also widely used to validate the consistency of experimental data.

The aim of this article is to present various predictive models based on quantum calculations performed on isolated molecules (in vacuum or in a cavity), enabling thermochemical properties and phase equilibria to be determined. Quantitative Structure-Property Relationship (QSPR) methods and molecular simulation are not covered here. We will focus in particular on DFT methods ( Density Functional Theory) and COSMO ( COnductor-like- Screening MOdel), and review the main versions of these methods and associated software. We also discuss the semi-empirical methods available in the free MOPAC software, which enable pre-optimization of molecular geometries and...