Electrical, Optical, Structural, and Analytical Properties of Very Pure GaN

Document Type

Conference Proceeding

Publication Date


Find this in a Library

Catalog Record


Present hydride vapor phase epitaxial growth of GaN on Al2O3 can produce material of very high quality, especially in regions of the crystal far from the substrate/epilayer interface. In the present study, we characterize a 248-μm-thick epilayer, which had been separated from its Al2O3 substrate and etched on top and bottom to produce flat surfaces. Temperature-dependent Hall-effect data have been fitted to give the following parameters: mobility μ(300) = 1320 cm2/V-s; μ(peak) = 12,000 cm2/V-s; carrier concentration n(300) = 6.27 × 1015 cm−3; donor concentration ND = 7.8 × 1015 cm−3; acceptor concentration NA = 1.3 × 1015 cm−3; and effective donor activation energy ED = 28.1 meV. These mobilities are the highest ever reported in GaN, and the acceptor concentration, the lowest. Positron annihilation measurements give a Ga vacancy concentration very close to NA, showing that the dominant acceptors are likely native defects. Secondary ion mass spectroscopic measurements show that ND is probably composed of the common donors O and Si, with [O] > [S1]. Transmission electron microscopy measurements yield threading dislocation densities of about 1 × 107 cm−2 on the bottom (N) face, and < 5 × 105cm−2 on the top (Ga) face. Photoluminescence (PL) spectra show a strong donor-bound exciton (D°X) line at 3.47225 eV, and a weaker one at 3.47305 eV; each has a linewidth of about 0.4 meV. In the two-electron satellite region, a strong line appears at 3.44686 eV, and a weaker one at 3.44792 eV. If the two strong lines represent the same donor, then ED,n=1 – ED,n=2 = 25.4 meV for that donor, and the ground-state activation energy (EC – ED,n=1) is (4/3)25.4 = 33.9 meV in a hydrogenic model, and 32.7 meV in a somewhat modified model. The measured Hall-effect donor energy, 28.1 meV, is smaller than the PL donor energy, as is nearly always found in semiconductors. We show that the difference in the Hall and PL donor energies can be explained by donor-band conduction via overlapping donor excited states, and the effects of non-overlapping excited states which should be included in the n vs. T data analysis (charge balance equation).


Presented at the 2002 MRS Fall Meeting, Boston, MA.

Copyright © Materials Research Society 2003.



Catalog Record