Reliable and Versatile Model for the Density of Liquids Based on Additive Volume Increments

A new procedure is put forward to predict the standard density (ρ) of pure fluids, using a newly compiled database, including 8870 organic and metal–organic compounds. This model is simpler, easier to apply, and more general than existing methods, being applicable to a broad range of compounds, including elements C, H, N, O, F, Cl, Br, I, B, Al, Si, P, S, Ge, Se, Sn, Pb, and Ti. The average relative error of <1.7% demonstrates an accuracy similar to that of group contribution (GC) methods focused on restricted chemical families, and much better than obtained using quantitative structure–property relationships. The small number of fitting parameters involved (47), compared to GC methods, is made possible through a physically motivated decomposition of the molar volume into additive contributions. The model provides new insight into the dependence of ρ on structural features, such as atom radii and rings. We provide a script allowing for straightforward application of the model, and the extensive database specifically compiled for this work to foster further development.