The dopant chemical space in LaMnO3 (LMO) is systematically
explored using first-principles computations. We study a range of
cationic dopants including alkali, alkaline earth metals, 3d, 4d,
and 5d transition metal elements without and with an adjacent O vacancy.
A linear programming approach is employed to access the energetically
favorable decomposition pathway and the corresponding decomposition
energy of doped LaMnO3. The decomposition energy is then
used to classify the dopants for stability, site preference and tendency
of O vacancy formation. We find that La site doping is more favored
compared to Mn site doping. We also identify dopants previously not
considered, such as K, Rb, Cs, and In, which lead to stable doped
LMO and are also excellent O vacancy formers. Employing data mining
techniques, we identify the dopant features that are critical to the
stability of a doped oxide.