Conductivity of Ionic Liquids: A Neural Network Approach

Abstract

A mathematical model based on artificial neural networks (ANNs) has been designed to achieve the estimation of the conductivity of three different binary mixtures containing water and one of the following imidazolium-based ionic liquids: 3-ethyl-methylimidazolium tetrafluoroborate ([emim][BF4]), 3-butyl-methylimidazolium tetrafluoroborate ([bmim][BF4]), and 3-hexyl-methilimidazolium tetrafluoroborate ([hmim][BF4]). It has been accomplished employing the composition of the mixtures (in terms of mole fraction) and the number of carbon atoms that form the lateral chain of the imidazolium cation as independent variables. All the utilized data was gathered from the literature and it corresponds to measurements that were carried out at 298.15 K. The mean prediction error for the estimation of this property was 6.42% and its R2 value 0.991. These results support that the application of nonlinear mathematical models which rely on ANNs can lead to an accurate estimation of ionic conductivity values of aqueous binary mixtures of imidazolium-based ionic liquids.