Modelling study on phase equilibria behavior of ionic liquid-based aqueous biphasic systems

• 17,449 experimental binodal data points covering various IL-ABS at different temperatures are collected. Optimal parameters of a well-known mathematical description for each the obtained. A machine learning model (ANN-GC) with reliable predictive performance is developed. Some main issues that govern phase equilibria behavior IL- ABS discussed. The ability to predict crucial relevance in early design stage biphasic liquid-liquid systems. Ionic liquid-based aqueous systems (IL-ABS) have demonstrated superior many applications such as recovery bio-products and recycling hydrophilic ILs from solutions. In order better utilize these novel systems, modelling studies on carried out this work. First, IL database developed our previous work extended unconventional total, 171 (278.15 K-343.15 K) Then, all involved correlated using popular three-parameter optimal Afterwards, we try build linear group contribution (GC) IL-ABS, but it fails due high complexity For reason, finally turn applying algorithm, i.e., artificial neural network (ANN), nonlinear GC purpose. This gives mean absolute error (MAE) 0.0175 squared correlation coefficient (R 2 ) 0.9316 13,789 training points, 3,660 test they 0.0177 0.9195, respectively. results indicate proposed ANN-GC model, some extent, capable IL-ABS. Besides efforts building models, also discuss ABS, which could be guidance