Abstract

The use of artificial neural networks (ANNs) for the correlation and prediction of the melting temperature of ionic liquids (ILs) is analyzed in this paper. Several network architectures and two sets of data were analyzed and the results compared with others from the literature. The independent variables (those that could have an influence on the melting temperature) considered for training the ANNs were the groups forming the molecules, mass of the cation, mass of the anion, and mass connectivity index. These properties are easily available or are calculated and are provided as Supporting Information. As a measure of the accuracy of the method, the average deviation and average absolute deviation are evaluated. The results of this work and others from the literature indicate that appropriate selection of data and a good combination of architecture and variables can lead to an acceptable correlation of data, but accurate prediction is not yet possible. The lack of a clear definition of the melting temperature and the lack of knowledge on what are the properties that most affect melting are the main causes of the present incapability of accurately predicting the melting temperature of any IL.