Application of neural network molecular modeling for correlating and predicting Henryʹs law constants of gases in [bmim][PF6] at low pressures

Ionic liquids, due to their unique properties, have aroused great interests within chemical engineering, chemistry and environmental sciences. Solubility of gases in ionic liquids has been investigated experimentally by several researchers and different modeling techniques have been employed to correlate obtained experimental data. Almost all proposed modeling methods require tuned adjustable parameters which have been optimized based on experimental data. Without experimental data and tuned adjustable parameters, none of recommended modeling methods can be used confidently for estimating solubility of gases in ionic liquids. s manuscript, Henryʹs law constants of carbon dioxide, carbon monoxide, argon, oxygen, nitrogen, methane and ethane in 1-butyl-3-methylimidazolium hexafluorophosphate has been modeled by neural network technique. Gas molecular weight, gas acentric factor (sphericity of gas molecule), reduced temperature and absolute pressure have been employed as network inputs, and Henryʹs law constant has been correlated accurately. In addition to precise modeling, the new method has the capability of predicting the Henryʹs law constant of a specific gas based on experimental data points of other gases.