Solid-liquid equilibrium of the GaInGaSb system and artificial neural network prediction

The GaInAsSb solid-liquid equilibrium was studied by experimental measurement and by computer prediction. The liquidus and solidus data are determined by differential thermal analysis and by liquid-phase epitaxial growth respectively. Artificial neural networks, trained by use of the experimental data, have been used to predict liquidus and solidus data. The cross-validation method is also employed to get rid of overfitting.