We report on the growth of GaN with a zinc-blende, wurtzite, or a mixed phase structure on (001)GaP and (001)GaAs substrates by a low-temperature modified molecular beam epitaxy technique. By systematically varying the incident arsenic overpressure, films grown at a moderate substrate temperature of ≊620 °C show predominately wurtzite α-GaN, zinc-blende β-GaN, or a mixed phase of the two. Films containing only the metastable phase β-GaN were achieved by using a relatively high growth temperature of ≊700 °C and with an arsenic overpressure of ≊2.4×10−5 Torr. X-ray diffraction measurements indicate an improved crystalline quality for the layers grown at ≊700 °C compared to those grown at ≊620 °C as evident by a narrower full width at half-maximum of 35 min for β-GaN, which is among the narrowest reported to date.

We report new lines in the photoluminescence (PL) spectrum of lightly Be-doped GaN. The low-temperature PL spectrum of the lightly doped sample is dominated by a transition at 3.385 eV with first and second LO phonon replicas. Power-resolved PL measurements showed that the peak at 3.385 eV narrowed in width and shifted to higher energies with increasing excitation intensity. Thus the transition is attributed to donor-to-acceptor recombination, involving a Be acceptor of optical ionization energy of between 90 and 100 meV. This is much shallower than the acceptor level of 250 meV induced by Mg doping. Increasing the doping, however, resulted in a quenching of the band-edge luminescence and the appearance of a broad transition centred around 2.4 eV which we assign to a complex involving Be. Undulations on the peak were consistent with interference effects. On increasing the doping level even further all luminescence was quenched.

We report the results of low-temperature photoluminescence measurements on GaN films grown by molecular beam epitaxy on (0001) sapphire substrates. Samples were either nominally undoped or doped with Si. The spectra are generally dominated by a sharp peak at 3.47 eV which is attributed to excitons bound to neutral donors. A much weaker peak (or shoulder) near 3.45 eV probably arises from excitons bound to neutral acceptors. On raising the temperature to 50 K, in some samples free exciton peaks can be partially resolved on the high-energy side of the main line. In others we believe that these free excitons are recaptured onto neutral acceptors, thus enhancing the low-energy side of the line. A broader emission line appears in many samples at an energy near 3.42 eV which shows significant variation in position between samples. Our data show that it represents a free-to-bound, probably a free hole-to-donor, transition. This donor has previously been associated with oxygen. Of particular interest is the fact that some samples show a second sharp peak at 3.27 eV, together with a second broader peak at about 3.17 eV (also variable in energy). The sharp peak is energetically consistent with its being either a donor-acceptor or a free electron-to-bound hole transition, but subsidiary measurements rule out both these possibilities. We suggest that it may represent an exciton bound to a deep donor or a shallow donor-bound exciton in zinc blende GaN inclusions contained within the mainly wurtzite material. We tentatively interpret the 3.17 eV line as a phonon replica of this zinc blende line, the phonon energy being perturbed by the small size of the inclusions and by strain effects within these inclusions.

No abstract

No abstract