The SAFT and cubic models are often compared in the open literature. Usually, limitations and strengths of both types of equations of state (EoSs) are ascribed to their theoretical foundations. Little attention has been paid until now to the influence of parameterization choices on the performance of these EoSs. By parameterization, it is here meant the way or method used to attribute values to EoS component-dependent parameters. SAFT-type and cubic EoSs for pure compounds use two different parameterization methods: SAFT parameters are generally regressed on vapor pressure and liquid density data, whereas cubic models use as input parameters experimental values of the critical temperature, pressure, and acentric factor and are generally parameterized to exactly reproduce these three properties. As a consequence of these parameterization methods, SAFT EoSs overestimate critical pressures, whereas cubic EoSs fail in reproducing accurately liquid density data. To discuss the influence of parameterization methods on EoS performances, three different parameterization methods are considered and applied to both SAFT and cubic models. It is shown that by application of the same parameterization method, SAFT and cubic EoSs are equivalent as they exhibit similar accuracies for the correlation of vapor pressure, enthalpy of vaporization, heat capacity, and liquid density data. This study led us to conclude that EoSs requiring three component-dependent parameters cannot provide a simultaneous accurate prediction for all of the properties of a pure fluid. It is highlighted that the key point to get high accuracy (for nonassociating pure compounds) is to work with an EoS equivalent to a four-parameter corresponding-states principle, i.e., that requires the specification of four macroscopic properties for a given pure compound. Eventually, we have considered a volume-translated PC-SAFT EoS version involving four input parameters: the three classical ones (m, σ, and ε) and a volume-translation parameter (c). Such an EoS shows excellent performances: for no less than 587 nonassociating compounds, vapor pressure data are predicted with an error of 1.5%, liquid heat capacity and enthalpy of vaporization are predicted with an error of 3%, and liquid density data are predicted with an error of 4%; critical temperatures and pressures are exactly reproduced.

The Gibbs energy of solvation measures the affinity of a solute for its solvent and is thus a key property for the selection of an appropriate solvent for a chemical synthesis or a separation process. More fundamentally, Gibbs energies of solvation are choice data for developing and benchmarking molecular models predicting solvation effects. The Comprehensive Solvation-CompSol-database was developed with the ambition to propose very large sets of new experimental solvation chemical-potential, solvation entropy, and solvation enthalpy data of pure and mixed components, covering extended temperature ranges. For mixed compounds, the solvation quantities were generated in infinite-dilution conditions by combining experimental values of pure-component and binary-mixture thermodynamic properties. Three types of binary-mixture properties were considered: partition coefficients, activity coefficients at infinite dilution, and Henry's-law constants. A rigorous methodology was implemented with the aim to select data at appropriate conditions of temperature, pressure, and concentration for the estimation of solvation data. Finally, our comprehensive CompSol database contains 21 671 data associated with 1969 pure species and 70 062 data associated with 14 102 binary mixtures (including 760 solvation data related to the ionic-liquid class of solvents). On the basis of the very large amount of experimental data contained in the CompSol database, it is finally discussed how solvation energies are influenced by hydrogen-bonding association effects.