Prediction of possible CaMnO3 modifications using an ab initio minimization data-mining approach

We have performed a crystal structure prediction study of CaMnO3focusing on structures generated by octahedral tilting according to group–subgroup relations from the ideal perovskite type (Pm\overline 3 m), which is the aristotype of the experimentally known CaMnO3compound in thePnmaspace group. Furthermore, additional structure candidates have been obtained using data mining. For each of the structure candidates, a local optimization on theab initiolevel using density-functional theory (LDA, hybrid B3LYP) and the Hartree-–Fock (HF) method was performed, and we find that several of the modifications may be experimentally accessible. In the high-pressure regime, we identify a post-perovskite phase in the CaIrO3type, not previously observed in CaMnO3. Similarly, calculations at effective negative pressure predict a phase transition from the orthorhombic perovskite to an ilmenite-type (FeTiO3) modification of CaMnO3.