More applications for every \(g^\mathrm{E}\) model
Binary and ternary phase equilibria through the convex envelope method (CEM) are now available with UNIFAC 2.0, mod. UNIFAC 2.0 and both COGNAC variants, not only with HANNA. Excess enthalpies are now available with mod. UNIFAC 2.0 and both COGNAC variants as well, and NRTL parameters can now be fitted with four parameters and a fixed α to any of the models.

EVE: New mutual diffusion coefficient model added
Added an integration of EVE (Enhanced Vignes Equation), which predicts the composition-dependent mutual diffusion coefficient of a binary mixture in both the Maxwell–Stefan and the Fick convention. EVE builds on two models already on MLPROP: HADES supplies the infinite-dilution coefficients that anchor the curve, HANNA the thermodynamic factor. As with HADES, only the pure-component viscosities are required. Pre-print coming soon.

COGNAC: New activity coefficient model added
Added an integration of COGNAC, a graph neural network that writes the complete mod. UNIFAC (Dortmund) parameter table for the queried molecules, with every atom as its own group. Both residual closures are available, and COGNAC can also be combined with GRAPPA for VLE predictions or used as a source for fitted NRTL parameters. Pre-print coming soon.

HADES: New self-diffusion coefficient model added
Added an integration of the HADES model (Hybrid Architecture for Consistent Diffusion Coefficient Estimation Across Solvents), which predicts composition-dependent self-diffusion coefficients in mixtures from the pure-component viscosities alone. Pre-print coming soon.

HANNA update
Updated the HANNA model with new training data, also including excess enthalpies. The paper is published in Nature Communications. Excess enthalpy predictions for binary mixtures with HANNA are now available.

ESE: New diffusion coefficient model added
Added an integration of the Enhanced Stokes-Einstein (ESE) model for binary diffusion coefficients at infinite dilution. Pre-print available on arXiv.

Added CEM-based phase equilibrium predictions
MLPROP now enables the prediction of binary and ternary phase equilibria including VLE and LLE through the convex envelope method (CEM) using HANNA and GRAPPA.

New HANNA Model Integration
MLPROP now features the new HANNA model that was additionally trained to LLE data and is applicable to multi-component mixtures. The old HANNA model is still available as HANNA (legacy).