Pure MnN and (Ga,Mn)N alloys are investigated using the ab initio generalized gradient approximation $+U$ $\left(\text{GGA}+U\right)$ or the hybrid-exchange density-functional (B3LYP) methods. These methods are found to predict dramatically different electronic structure, magnetic behavior, and relative stabilities compared to previous density-functional theory (DFT) calculations. A unique structural anomaly of MnN, in which local-density calculations fail to predict the experimentally observed distorted rocksalt as the ground-state structure, is resolved under the $\text{GGA}+U$ and B3LYP formalisms. The magnetic configurations of MnN are studied and the results suggest the magnetic state of zinc-blende MnN might be complex. Epitaxial calculations are used to show that the epitaxial zinc-blende MnN can be stabilized on an InGaN substrate. The structural stability of (Ga,Mn)N alloys was examined and a crossover from the zinc-blende-stable alloy to the rocksalt-stable alloy at an Mn concentration of $\sim 65\%$ was found. The tendency for zinc-blende (Ga,Mn)N alloys to phase separate is described by an asymmetric spinodal phase diagram calculated from a mixed-basis cluster expansion. This predicts that precipitates will consist of Mn concentrations of $\sim 5$ and $\sim 50\%$ at typical experimental growth temperatures. Thus, pure antiferromagnetic MnN, previously thought to suppress the Curie temperature, will not be formed. The Curie temperature for the $50\%$ phase is calculated to be $T_C=354\text{K}$, indicating the possibility of high-temperature ferromagnetism in zinc-blende (Ga,Mn)N alloys due to precipitates.

• Received 16 October 2008

DOI:https://doi.org/10.1103/PhysRevB.78.184109