Toward multi-principal component alloy discovery: Assessment of CALPHAD thermodynamic databases for prediction of novel ternary alloy systems

CALPHAD databases have traditionally been developed for investigation of single-principal component alloys. With the advent of batch processing capability, engineering teams have proposed using these models to systematically explore compositional space for multiprincipal element systems. However, the uncertainty of phase equilibria predictions outside of traditional compositional bounds has yet to be evaluated. This study assesses the current capabilities of commercially available CALPHAD databases to predict phase equilibria within ternary phase space as a function of the number of full binary system descriptions contained within the thermodynamic databases, the spatial location in compositional space relative to subsystem descriptions, and the specific database used. A strong correlation was observed between the fraction of subsystem descriptions available for the free energy calculation and the accuracy of phase predictions in undefined ternary space. The accuracy of equilibria predictions degraded with increased compositional extrapolation from defined subsystems.