Correct Structural Phase Stability of FeS2, TiO2, and MnO2from a Semilocal Density Functional

The structural phase stabilities of iron disulfide (FeS2), titanium dioxide (TiO2), and polymorphs of manganese oxides (MnO2) are known to be challenging problems for the semilocal exchange-correlation (xc) approximations of the Density Functional Theory. Popular and nonempirical semilocal xc functionals fail badly for at least one of these solids. For example, the strongly constrained and appropriately normed (SCAN) meta-GGA xc functional fails to predict the correct ground state for FeS2 (pyrite and marcasite polymorphs) and TiO2 (rutile and anatase polymorphs) structures but yields accurate formation energies and the correct ground state of all MnO2 polymorphs. In this work, we find that the recently proposed meta-GGA constructed from the cuspless hydrogen exchange hole density (MGGAC) [Phys. Rev. B 2019, 100, 155140] can predict the correct ground state of polymorphs of FeS2, TiO2, and MnO2. Moreover, when compared to the experimental results, we also obtain the quantitatively good agreement from the MGGAC functional for the structural properties and bandgaps of those solids, and this agreement indicates the feasibility of this method.