COSMO-RS: From Quantum Chemistry to Fluid Phase Thermodynamics

Abstract Thermodynamic data is the fundament of any process design. While accurately measured experimental data almost always is the most reliable source of information for the thermodynamic data required for process design, there are many situations in which this data is not available. In such situations predictive models are required. Correlative models as NRTL or SAFT, or group contribution methods as UNIFAC are the traditional approaches for property prediction in chemical engineering. Correlative methods require some experimental data for exactly the compounds and mixtures under consideration and only extrapolate to other state points. UNIFAC extrapolates to new compounds, but only if the compounds are well describable by the additive group concept and do not have major intramolecular interactions. For structurally really new and demanding compounds, robust predictions require sound physical concepts for the description of the molecular interactions. The Conductor-like Screening Model for Realistic Solvation, COSMO-RS, is the most fundamental of such models, combining the sound quantum chemical polarity information with the efficient concept of surface segment interactions, also known as quasi-chemical approach. Thus, it provides reliable predictions for pure compound vapor pressures and mixture activity coefficients. COSMO-RS has been initially published more than 20 years ago, and meanwhile it has become a popular method in academia and industry. The talk will cover the basic concepts of the method, its potential and limitations, and outline several extensions and new application areas which have opened up by the various methodological improvements developed during the past 20 years.