New models for predicting thermophysical properties of ionic liquid mixtures

Area of Science:

• Physical Chemistry
• Chemical Engineering
• Materials Science

Background:

• Accurate prediction of physicochemical properties for ionic liquid (IL) mixtures is crucial for their diverse applications.
• Existing models often require numerous parameters and have limited applicability.

Purpose of the Study:

• To develop and validate novel semi-empirical and artificial neural network (ANN) models for predicting key properties of IL mixtures.
• To offer models with fewer adjustable parameters and broader applicability compared to existing methods.

Main Methods:

• Development of semi-empirical models, each with a single characteristic parameter determined by one experimental data point.
• Establishment of artificial neural network (ANN) models as an alternative predictive tool.
• Validation using 2304 experimental data points for binary mixtures of ILs and molecular compounds.

Main Results:

• Both semi-empirical and ANN models demonstrated high predictive accuracy, with overall average absolute deviations (AARDs) below 2%.
• The developed semi-empirical models require significantly fewer adjustable parameters than previously reported models.
• The models show applicability across a wider range of IL mixture compositions and types.

Conclusions:

• The proposed semi-empirical and ANN models provide reliable and efficient tools for predicting the physicochemical properties of IL mixtures.
• These models facilitate the design and application of ILs by simplifying property prediction.
• The reduced parameter requirement and enhanced applicability make these models valuable for researchers and engineers in the field of ionic liquids.