Evaluating transition metal oxides within DFT-SCAN and SCAN + U frameworks for solar thermochemical applications
Using the strongly constrained and appropriately normed (SCAN) and SCAN+U approximations for describing electron exchange-correlation (XC) within density functional theory, we investigate the oxidation energetics, lattice constants, and electronic structure of binary Ce-, Mn-, and Fe-oxides, which are crucial ingredients for generating renewable fuels using two-step, oxide-based, solar thermochemical reactors. Unlike other common XC functionals, we find that SCAN does not over-bind the O 2 molecule, based on direct calculations of its bond energy and robust agreement between calculated formation enthalpies of main group oxides versus experiments. However, in the case of transition-metal oxides (TMOs), SCAN systematically overestimates (i.e., yields too negative) oxidation enthalpies due to remaining self-interaction errors in the description of their ground-state electronic structure. Adding a Hubbard U term to the transition-metal centers, where the magnitude of U is determined from experimental oxidation enthalpies, significantly improves the qualitative agreement and marginally improves the quantitative agreement of SCAN+U-calculated electronic structure and lattice parameters, respectively, with experiments. Importantly, SCAN predicts the wrong ground-state structure for a few oxides, namely, Ce 2 O 3 , Mn 2 O 3 , and Fe 3 O 4 , while SCAN+U predicts the right polymorph for all systems considered in this work. Furthermore, the SCAN+U framework, with an appropriately determined U, will be required to accurately describe ground-state properties and yield qualitatively consistent electronic properties for most transition-metal and rare earth oxides.