In their comment, Siegle et al. present Raman and photoluminescence ~PL! data from samples grown either on GaAs by molecular beam epitaxy ~MBE! or on sapphire by metal organic vapor phase epitaxy and hydride vapor phase epitaxy. They observe electronic Raman scattering ~ERS! peaks like those reported in Ref. 2 and claim, however, our assignments to be incorrect. Siegle et al. report peaks at 60 ~a!, 95 ~b!, 102.5 ~g!, 125 ~d!, and 250 cm ~e! in addition to the ERS peaks at 151 ~A*!, 189 (A), 218 ~B*!, and 237 cm (B) discussed in Ref. 2 and assert that all these peaks have the same origin. In fact, we also observe peaks b and d in the spectra of our GaN samples. According to our model, we expect a very similar Raman resonance behavior for different donor species with sufficiently low binding energy. Therefore, the argument based on the resonance behavior is invalid in attributing all the observed peaks to the same origin. If the same origin is assumed for all peaks, their relative intensities should be nearly constant for different samples. This is by far not the case as shown, e.g., in Fig. 1 of Ref. 2 for peaks A and B vs A* and B* as well as for peaks b vs d in Fig. 1 ~compare also Figs. 1 and 2 in Ref. 1!. In contrast to the ERS peaks discussed in Ref. 2, peak d has been observed also in p-type GaN and shows a weak temperature dependence. Therefore, it is not even clear whether peak d is due to ERS. Siegle et al. further claim as a main statement that the ERS peaks are only observable for GaN grown on GaAs. However, we observe the ERS peaks reported in Ref. 2 also in GaN layers after removal of the GaAs substrate @Fig. 1~c!# demonstrating that they do not originate from GaAs. It is furthermore known that N impurities in GaAs introduce a resonant energy level in the conduction band which is not occupied under normal conditions. Moreover, peak B* was observed previously by absorption spectroscopy and, more recently, peaks A* and B* by Raman scattering from hexagonal GaN grown on sapphire. These findings completely rule out any connection with GaAs and therefore refute the statement of Ref. 1. The observability of the ERS peaks rather depends on the donor concentration and the Fermi level in the GaN layer. For too small donor concentrations, the intensity of the corresponding Raman excitations vanishes below the detection limit. At very large concentrations, a pronounced formation of impurity level bands is expected to destroy the observability of individual electronic transitions. In addition, the resonance enhancement should be only observable for sufficiently high concentrations of the involved deep level defect. In this respect, arguments based
Read full abstract