Due to the long withdrawal period and good storage capacity, thermal energy storage has proven to be a central element for making the future energy system more flexible. Phase change materials enable high storage densities at low temperature differences by exploiting the enthalpy of transformation upon phase transitions, thus, requiring an application-specific transition temperature, rendering the need for sophisticated modelling methods such as Calphad [1].

Calphad modelling has also been employed to predict phase equilibria and thermodynamic, electrochemical and physical properties of materials for Li-ion batteries [2]. However, the traditional Calphad approach relies on experimental phase diagram and thermodynamic data impeding the design of new electrode materials where experimental data is lacking.

Ab initio calculations at 0 K have become a very popular additional source of data. In the last decade, this has led to the development of huge repositories with calculated data such as materialsproject.org or oqmd.org, containing data for many hundreds of thousands of compounds.

Here, it will be shown how the 0 K phase stability data from the complete databases materialsproject.org and oqmd.org can be extrapolated to relevant temperatures. This is done by ML models for Cp(T) and S(T) trained on high-quality Calphad data taken from the FactSage databases. In addition, enthalpy of formation consistency modifications are used to make the enthalpy of formation consistent to existing Calphad databases. The models have led to the development of aiMP and aiOQ databases (www.gtt-technologies.de) [3].

Utilization of aiMP and aiOQ as well as further Calphad databases for the application cases of phase change materials for thermal energy storage and for Li-ion batteries are demonstrated.