QSPR prediction of the solubility of CO2 and N2 in common polymers

• We studied the solubility of carbon dioxide and nitrogen in common polymer.
• We calculated more than 1600 descriptors for compounds by using Drogon software.
• We selected best descriptors for gases and polymers by means of GFA.
• We constructed the linear and non-linear relationships based on QSPR approach.

In this work, a quantitative structure–property relationship (QSPR) model is presented for solubility prediction of carbon dioxide and nitrogen in polyethylene, polypropylene, polystyrene, polyvinyl acetate and poly(butylene succinate) at different temperatures and pressures. The five most important descriptors which are related to the structure of the gas molecules and the repeating unit of polymers were selected by means of a genetic function approximation (GFA) from a set of more than 1600 descriptors. The selected descriptors in addition to the temperature and pressure were used as the inputs for the artificial neural network (ANN) and adaptive neuro-fuzzy inference system (ANFIS) models to develop the required non-linear relationships. These non-linear models were assessed by some internal and external validation methods. The obtained results indicated the excellent ability of ANN and ANFIS for prediction of solubility of gases in polymers.

Figure optionsDownload as PowerPoint slide