Hide menu

Understanding magnesium impurities in gallium nitride

In short: Light-emitting diodes and laser diodes rely on positive charge carriers known as holes. These holes are provided by impurities, and for gallium nitride is magnesium the only impurity that is useful for this purpose. The exact properties of the magnesium impurity have long been controversial, partly due to instability of its spectral features. We are able to spectrally identify isolated magnesium impurities as well as impurities near faults in the stacking sequence of atomic layers in the gallium nitride crystal. These experimental results are in conflict with recent theoretical predictions of magnesium in gallium nitride.

Light-emitting diodes and laser diodes rely on two types charge carriers, negative electrons and positive holes. The electrons and holes are provided by impurities intentionally incorporated into the semiconductor crystal to substitute the original atoms. Magnesium is the only impurity that is useful for producing holes in gallium nitride. In the best cases a hole concentration of 2⋅1018 cm-3 has been demonstrated at room temperature, just about sufficient for light-emitting diode or laser diode applications.

Impurities in semiconductors are conveniently studied by optical spectroscopy. The impurities emit light – luminescence - when the semiconductor at low temperature is exposed to light of high photon energy or bombarded with electrons. Optical spectra related to the magnesium impurity have long been controversial, partly due to the instability of the spectral features [1]. Another peculiarity is the presence of two magnesium-related peaks instead of one in the high-energy part of the spectrum [1] (see peaks ABE1 and ABE2 in the figure).

Cathodoluminescence spectrum of m-plane GaN:Mg ([Mg] = 1E19 cm-3) grown on a bulk GaN substrate.

We have made a detailed study of both optical and structural properties in gallium nitride samples grown on native gallium nitride substrates. This guarantees a very low density of crystal defects (dislocation density in the 106 cm-3 range), and in addition a very narrow width of the spectral peaks (~1-2 meV). In this case the strain related energy shift of the peaks is absent in the spectra, enabling confident interpretation of the spectral features.

Structural studies based on transmission electron microscopy show that the magnesium impurities introduce faults in the stacking sequence of atomic layers above a concentration of 1018 cm-3. The spectral features of these stacking faults appear in below the ABE1 and ABE2 (see figure). Our explanation of the two magnesium-related peaks is the following: ABE1 related to the isolated magnesium impurities, while ABE2 is due to magnesium impurities positioned nearby stacking faults, so that the latter perturb and lower the energy level of the impurities. The instability of the magnesium-related spectra is now reinterpreted as due to an unstable nonradiative defect level.

Interestingly the structural studies show that magnesium clustering in these samples only occurs at high concentrations (≥ 1020 cm-3), and then as regions with up to ten percent magnesium. This is different from previous studies of samples grown on sapphire, where magnesium segregation into metallic pyramidal defects are reported at magnesium concentrations above 2⋅1019 cm-3.

Our experimental results are in conflict with recent theoretical calculations of the energy levels and optical transitions for the magnesium impurity in gallium nitride. Lyons et al. [2] claim that the spectral signature of magnesium is a broad peak at low energy (2.9 eV). This idea has to be wrong, e. g. since that broad feature occurs only at high magnesium concentrations in samples grown by the MOCVD method, while the above ABE1 and ABE2 features occur in all magnesium incorporated materials.

[1] B. Monemar et al., Phys. Rev. Lett. 102, 235501 (2009)
[2] S. Lyons et al., Phys. Rev Lett. 108,156403 (2012); A. Alkauskas et al., Phys. Rev. Lett. 109, 267401 (2012)

Details of this work are described in the following two journal articles:

Properties of the main Mg-related acceptors in GaN from optical and structural studies
Journal of Applied Physics 115, 053507 (2014)     doi:10.1063/1.4862928

Authors: B. Monemar, P. P. Paskov, G. Pozina, C. Hemmingsson, J. P. Bergman, S. Khromov, V. N. Izyumskaya, V. Avrutin, X. Li, H. Morkoç, H. Amano, M. Iwaya and I. Akasaki

Luminescence of Acceptors in Mg-Doped GaN
Japanese Journal of Applied Physics 52, 08JJ03 (2013)    doi:10.7567/JJAP.52.08JJ03

AuthorsBo Monemar, Sergey Khromov, Galia Pozina, Plamen PaskovPeder BergmanCarl Hemmingsson, Lars Hultman, Hiroshi Amano, Vitaliy Avrutin, Xing Li, and Hadis Morkoç


Bo Monemar, professor
Phone: +46 (0)13 28 17 65
E-mail: bom@ifm.liu.se


  • Knut and Alice Wallenberg Foundation (KAW)
  • Swedish Energy Agency
  • Linköping Linnaeus Initiative for Novel Functional Materials (LiLi-NFM)


Responsible for this page: Fredrik Karlsson
Last updated: 02/11/14