Local electronic structure of Cu-doped GaN investigated by XANES and x-ray linear dichrosim

X-ray absorption near edge structure (XANES) and x-ray linear dichroism (XLD) measurements of a series of Cu-doped GaN samples with different doping concentrations were investigated at the Ga and Cu $K$ edges to get insight into the preferred occupation site of Cu in the GaN matrix. The XLD data of the Cu $K$ edge clearly demonstrates that Cu is incorporated in the GaN film. $\gamma$-Cu${}_9$Ga${}_4$ compounds that formed on the samples during the GaN growth process could be partially removed by etching with HNO${}_3$. At both absorption edges, a clear increase of the XLD signal intensity could be observed after etching as a consequence of the removal of the $\gamma$-Cu${}_9$Ga${}_4$ compounds. By performing simulations of XANES and XLD spectra, which are compared to the experimental data, it was possible to show that most Cu in the GaN film is incorporated on Ga or interstitial sites and only a minor fraction on N sites.

Figure 1

Schematic representation of part of a GaN $4\times 4\times 4$ supercell. (a), (b), and (c) show how the Cu atoms are placed in the GaN matrix on interstitial, N, and Ga sites, respectively.

Figure 2

SEM image of sample F (2.69%) before etching (a) and after etching (b). On the left sides of the pictures, the surface coverage of the precipitates is estimated by approximating the structures’ area with the area of multiple small circles. For this sample, a 88% decrease in the surface coverage of the surface structures is calculated from the pre-etched to the post-etched condition of the sample.

Figure 3

The measured XLD spectra depicted here were normalized to the background at the lowest and the highest energies of the spectrum. (a) and (b) are the XLD spectra of the Ga $K$ edge prior to and after etching of the samples, respectively. (c) and (d) show the XLD spectra of the Cu $K$ edge before and after etching of the samples, respectively. Bragg peaks are marked by “$\times$”.

Figure 4

The intensity increase of the XLD peaks at the Ga $K$ edge between etched and unetched samples is shown as a function of the nominal doping concentration. The increase is due to the removal of $\gamma$-Cu${}_9$Ga${}_4$ compounds (see text for discussion).

Figure 5

Comparison of the measured Ga $K$-edge XLD signal to the simulated ones. (a) shows the undoped spectra, while (b) displays the measured graph of sample C (1.35%) in comparison to calculated spectra for different Cu site possibilities.

Figure 6

(a)–(c) depict the Cu $K$-edge XLD simulation results for 1.56% Cu-doped GaN on Ga, N, and interstitial sites, (d) shows the experimental data of sample C (1.35%). As a guide to the eye, the two main peaks present in (d) are marked by dotted lines throughout the other graphs.

Figure 7

Experimental XANES spectra of the Ga (a) and Cu (c) $K$ edge for the differently doped samples. (b) and (d) show the experimental XANES data of sample C (1.35%) along with calculated spectra for Cu on different sites for the Ga and Cu $K$ edges, respectively.

Figure 8

Fitting of the experimental XANES graph of sample C (1.35%) with different calculated Cu site positions.

Figure 9

(a) shows the simulated XLD data of GaN:Cu with a mixture of the possible sites (Ga, N, interstitial) that Cu can occupy, according to the same fractions of contributions as described for the XANES spectra in Fig. 8. (b) displays the experimental XLD curve of sample C.

Figure 10

Comparison of the XLD signals of strained $\gamma$-Cu${}_9$Ga${}_4$ (a) and (c) and the experimental curve (b) of sample C (1.35%) at the Cu $K$ edge. None of the calculated spectra correspond well with the experimental one, indicating the absence of strain in $\gamma$-Cu${}_9$Ga${}_4$.

Figure 11

Fitting of the experimental XANES graph of sample C (1.35%) with different Cu site positions, where the $\gamma$-Cu${}_9$Ga${}_4$ compounds are subject to 15% biaxial compressive strain.