Discovery of thermochemical energy storage materials via a hybrid data-mining approach

A key technical challenge inhibiting fossil-free energy generation, which is especially pronounced in solar plants, is the inherent intermittency. Since thermochemical energy storage represents a solution that has yet to see large-scale utilization, the goal of the study at hand has been to find suitable materials using a hybrid high-throughput approach, encompassing both data mining and first principles calculations. Through these efforts, thermal properties have been estimated for more than 50 000 mono-, bi-, and trimetallic oxides, as well as alloys, retrieved from the Materials Project database.Thereafter, close to 18 000 convex hulls were created, each representing a unique metal ratio, in order to determine the most stable phases at different oxygen chemical potentials. In turn, this allowed almost 300 000 potential transitions to be identified. After filtering out potentially toxic and rare compositions, a little over 51 000 remained, which were further reduced, to about 3000, by requiring that the phase transformation must occur at a temperature between 400 °C and 1300 °C.A thorough analysis of the results led to the discovery of several promising energy storage materials. In addition, Chromium; manganese; calcium; and magnesium were found to be associated with high reaction enthalpies. When the sensible heat was taken into account, however, light elements such as lithium; boron; iron; and sodium dominated among the top-ranking candidates. This study therefore demonstrates how high-throughput data mining can be efficiently enhanced through first-principles calculations, enabled by machine-learning interatomic potentials, to facilitate material discovery.