Simple yet accurate prediction of liquid molar volume via their molecular structure

Molecular structure based study of liquid molar volume can be helpful in better understanding of liquid behaviors among a wide variety of scientific phenomena. Here, the most responsible molecular parameters for the pure liquid molar volume of organic and simple inorganic chemicals were determined through a robust quantitative molecular structural analysis. Using rank based ant system (RBAS) which is an encouraging optimization program, 1741 pure liquid molecules comprising 76 diverse chemical classes along with more than 3200 calculated molecular parameters for each liquid molecule were terminated in a simple yet accurate theoretical model with high accuracy. The result of three-parametric RBAS-MLR model is able to reach a prediction performance far better than the existing ones with a squared correlation coefficient (R2) of 0.97 and root mean square error (RMSE) of 15.34 cm3/gmol for the entire dataset. Based on chemical structural similarities, distribution of the average absolute relative errors were calculated and sorted.

► We developed a molecular structure based model for liquid molar volume.
► The 3 molecular descriptors were realized as the most responsible descriptor.
► The squared correlation coefficients for the RBAS-MLR model was 0.97.
► The model validity tests and error distribution analyzes were checked and the results were also physically presented.