||Reid Juday, Kewei Sun, Alec Fischer, Fernando Ponce, Hee Jin Kim, Suk Choi, Jeomoh Kim, Mi-Hee Ji, Jae-Hyun Ryou, Russell Dupuis
We have observed that low-energy electron beam irradiation (LEEBI) generates strong luminescence in the near-bandgap region in Mg-doped GaN annealed in a hydrogen-containing ambient. This emission resembles that of basal plane and prismatic faults in GaN. The samples used for this study were grown by metalorganic chemical vapor deposition (MOCVD) and consist of a 1 μm-thick Mg-doped GaN film on a 2.5 μm-thick undoped GaN epilayer on sapphire. As is required for hole conductivity in MOCVD-grown Mg-doped GaN, a post-growth thermal anneal was performed to activate the Mg acceptors. A single wafer was grown, cut into pieces, and annealed under different conditions: one in N<SUB>2</SUB> and one in N<SUB>2</SUB>(95%)/H<SUB>2</SUB>(5%) ambient atmospheres. Both samples were annealed for 3 minutes at 850 °C. Secondary ion mass spectrometry gives a Mg concentration of 3-4x10<SUP>19</SUP> cm<SUP>-3</SUP>. Cathodoluminescence (CL) was performed at 5 K to study the optical properties of the samples. Low power CL shows both samples exhibit very similar spectra; both an acceptor-bound exciton at 3.455 eV and the well-established donor-acceptor pair (DAP) at 3.289 eV are evident. However, after an energy dose of 594 keV/nm<SUP>2</SUP>, two additional recombination pathways appear in the sample annealed with hydrogen present. These are attributed to excitons bound to defects in the sample. Their CL emission energies, 3.407 and 3.384 eV, are similar to energies seen emitted from non-polar a-plane GaN, and were attributed to basal plane and prismatic plane stacking faults, respectively. We propose, however, that it is unlikely that the electron beam is creating stacking faults in the crystal. This is further suggested by the fact that very little of these emissions is seen in the sample annealed with only N<SUB>2</SUB>. As we continue to irradiate the sample in the SEM, the centers responsible for these peaks no longer become favorable for radiative recombination. We attribute both the 3.407 and 3.384 eV peaks as being related to hydrogen, which is known to be unstable, mobile, and very common in MOCVD-grown GaN:Mg. In addition, by monitoring the DAP intensity, we suggest that this peak is also related to hydrogen. LEEBI is seen to greatly enhance the luminescence of this band, most notably in the N<SUB>2</SUB>/H<SUB>2</SUB> annealed sample. Based on our results, we propose a model for the behavior of hydrogen in Mg-doped GaN during thermal annealing in a hydrogen-containing atmosphere and during electron irradiation during CL.