Polarity Dependence of Iron Incorporation in Gallium Nitride Epitaxial Films and Theoretical Analysis Using Universal Neural Network Potential

Herein, the incorporation of iron in Ga‐polar and N‐polar gallium nitride (GaN) films grown by metalorganic chemical vapor deposition is investigated. A depth profile of iron by secondary ion mass spectrometry reveals that the iron segregated into unintentionally doped GaN in Ga‐polar GaN. However, iron segregation is not observed in N‐polar GaN. To reveal the difference between Ga‐ and N‐polar GaN, theoretical calculations by neural network potential are performed. The calculations reveal that the adsorption energy of iron on an N‐polar surface is higher than that on a Ga‐polar surface. And the calculations also reveal that iron is more stable on a Ga‐polar surface than in bulk. On the other hand, iron is more stable in bulk than on an N‐polar surface. These calculation results are consistent to experimental results and reveal the mechanism of the iron segregation.