<i>Ab initio</i> thermodynamics of transition metal oxides for thermochemical energy storage.

(Co_xMn_1−x)₃O₄ is a promising candidate material for solar thermochemical energy storage. Modelling this system is challenging due to the varied sources affecting the free energy, and the prohibitive amount of configurations needed in the configurational entropy calculation. We present an accurate prediction of the experimental hausmannite-spinel SG in the case of (Co_xMn_1−x)₃O₄, using machine learning to extend an ab initio dataset of hundreds of structures, and including many different entropic contributions to the free energy. We then compare three different machine learning approaches for sampling the configuration space, and assess the accuracy of model predictions via the experimental phase diagram. Finally, we discuss how species mixing can increase the energy storage capacity of redox reactions with transition metal oxides and decrease their embodied energy cost. In the case of (Co_xMn_1−x)₃O₄, a simple calculation suggests possible storage capacities over 30% higher than that of pure Co₃O₄, in agreement with previous measurements. Theoretical considerations like these may be useful when trying to determine if thermochemical energy storage could replace the current use of molten salts in concentrated solar power generation.