Atomic structures and chemical states for active and inactive Mg dopant sites in GaN

抄録

INTRODUCTION

Gallium nitride (GaN) has many unique properties, such as a wide direct band gap, high thermal conductivity, extremely stable chemical characteristics, and strong radiation resistance. For such GaN-based devises, proper doping is necessary to improve electrical conductivity.[1] In the case of n-type GaN, Si is widely used to form the shallow donor states in the bandgap of GaN. In the case of p-type GaN, on the other hand, Mg is usually used due to the formation of shallow acceptor levels in the bandgap of GaN. For the atomic structure and the chemical states of active and inactive dopant sites in GaN, various doping sites have been proposed in previous studies. However, the atomic structure and chemical state of the active and inactive dopant sites of GaN have not yet been clarified, owing to the lack of direct evidence. The local atomic structures and chemical states of the dopant atoms should be clarified to obtain GaN-based devices with high performance.

In the present study, we employed photoelectron holography (PEH) to obtain 3D local atomic structure around the Mg dopant in GaN. In addition, by employing the SPEA-MEM/SPEA-L1 method to experimentally obtained PEH for the dopant atoms in GaN, we tried to directly observe and visualize the local 3D atomic structure for the active and inactive Mg dopant sites for GaN.

EXPERIMENTAL

Mg-doped GaN (0001) substrates were prepared by a hydride vapor phase epitaxy method with an epitaxial layer thickness of 100 nm. The Mg dopant concentration was 2.1 x 10²⁰ cm^-3. Hall effect measurements revealed a hole concentration of 2.6 x 10¹⁹ cm^-3 for Mg dopant.

The PEH measurements were performed at the BL25SU beamline of SPring-8. We used the Scienta-Omicron DA30 as an electron analyzer. The incident photon energies were changed so that Ga 3p and Mg 2p core level kinetic energies became ~ 800 eV for the PEH simulations.

RESULTS AND DISCUSSIONS

We measured the Mg-KLL Auger spectra of Mg-doped GaN. In the spectra, two peaks were observed; peak α and peak β. After annealing at 800 ℃, the areal intensity of peak β increased while that of peak α decreased. According to previous studies, high-temperature annealing can activate a part of the inactive dopant sites, increasing the hole concentration. Therefore, peak β should be the active site of the Mg-dopant. By measuring all azimuth angles and polar angle of such components. Then we obtained PEHs for components α and β, which is shown in Fig. 1. As can be seen, component α does not have any clear hologram patterns or Kikuchi lines, whereas component β exhibits clear hologram patterns. These patterns are very similar to the simulated PEH for Mg atom substituting a Ga atom in GaN (Mg_Ga). Since component β is the active site in Mg-doped GaN, Mg_Ga should be the active site in Mg-doped GaN. According to our photoelectron spectroscopy, the intensity ratio of component β is about 27%. This is very close to the electrical activation rate of 26.2% obtained from the Hall effect measurements. Accordingly, we can conclude that the active dopant site in Mg-doped GaN is Ga_Mg. From analysis of the PEH for component α, we found that the component α may be attributed to Mg_Ga-V or Mg_Ga-H. [2]

References

[1]. Tang Jingmin and Yamashita Yoshiyuki, ACS Appl. Electron. Mater. 3, 4618 (2021).

[2]. Tang Jingmin et al., ACS Appl. Electron. Mater. 4, 4719 (2022).