Comparison of the Luminous Efficiencies of Ga- and N-Polar ${\mathrm{In}}_x{\mathrm{Ga}}_{1-x}\mathrm{N}/{\mathrm{In}}_y{\mathrm{Ga}}_{1-y}\mathrm{N}$ Quantum Wells Grown by Plasma-Assisted Molecular Beam Epitaxy

We investigate the luminescence of Ga- and N-polar ${\mathrm{In}}_x{\mathrm{Ga}}_{1-x}\mathrm{N}/{\mathrm{In}}_y{\mathrm{Ga}}_{1-y}\mathrm{N}$ quantum wells grown by plasma-assisted molecular beam epitaxy on freestanding GaN as well as 6$H$-SiC substrates. In striking contrast to their Ga-polar counterparts, the N-polar quantum wells prepared on freestanding GaN do not exhibit any detectable photoluminescence even at 10 K. Theoretical simulations of the band profiles combined with resonant excitation of the quantum wells allow us to rule out carrier escape and subsequent surface recombination as the reason for this absence of luminescence. To explore the hypothesis of a high concentration of nonradiative defects at the interfaces between wells and barriers, we analyze the photoluminescence of Ga- and N-polar quantum wells prepared on 6$H$-SiC as a function of the well width. Intense luminescence is observed for both Ga- and N-polar samples. As expected, the luminescence of the Ga-polar quantum wells quenches and redshifts with increasing well width due to the quantum confined Stark effect. In contrast, both the intensity and the energy of the luminescence from the N-polar samples are essentially independent of the well width. Transmission electron microscopy reveals that the N-polar quantum wells exhibit abrupt interfaces and homogeneous composition, excluding emission from In-rich clusters as the reason for this anomalous behavior. The microscopic origin of the luminescence in the N-polar samples is elucidated using spatially resolved cathodoluminescence spectroscopy. Regardless of well width, the luminescence is found to not originate from the N-polar quantum wells but from the semipolar facets of $\vee$-pit defects. These results cast serious doubts on the potential of N-polar ${\mathrm{In}}_x{\mathrm{Ga}}_{1-x}\mathrm{N}/{\mathrm{In}}_y{\mathrm{Ga}}_{1-y}\mathrm{N}$ quantum wells grown by plasma-assisted molecular beam epitaxy for the development of long-wavelength light-emitting diodes.

Figure 1

Band profiles of the Ga-polar sample $A$ (a) and the N-polar sample $B$ (b). In both figures, we also show the calculated electron (solid blue line) and hole (solid red line) wave functions for the second QW.

Figure 2

Low-temperature (10 K) $\mu \text{−}\mathrm{PL}$ spectra under resonant excitation (413 nm) of samples $A$ and $B$. The samples are measured side by side with an excitation density of 200 and $2000\mathrm{kW}/{\mathrm{cm}}^2$ for samples $A$ and $B$, respectively. The sharp feature at 2.82 eV for sample $B$ originates from second-order optical phonon scattering in the GaN substrate. The spectra are vertically shifted to line up their background in order to account for the different integration times used during the measurements.

Figure 3

Characteristic AFM images of Ga- (a) and N-polar (b) ${\mathrm{In}}_x{\mathrm{Ga}}_{1-x}\mathrm{N}/{\mathrm{In}}_y{\mathrm{Ga}}_{1-y}\mathrm{N}$ QWs on $\mathrm{SiC}\{0001\}$. The images shown are taken from samples $E$ (a) and $J$ (b).

Figure 4

Examples for experimental and simulated triple-crystal $\omega \text{−}2\theta$ scans across the GaN 0002 reflection of Ga- (a) and N-polar (b) ${\mathrm{In}}_x{\mathrm{Ga}}_{1-x}\mathrm{N}/{\mathrm{In}}_y{\mathrm{Ga}}_{1-y}\mathrm{N}$ QWs on $\mathrm{SiC}\{0001\}$. The scans are taken from samples $C$ (a) and $G$ (b).

Figure 5

Low-temperature (10 K) $\mu \text{−}\mathrm{PL}$ spectra of Ga- (a) and N-polar (b) ${\mathrm{In}}_x{\mathrm{Ga}}_{1-x}\mathrm{N}/{\mathrm{In}}_y{\mathrm{Ga}}_{1-y}\mathrm{N}$ QWs with different widths. The arrows indicate the shift of the ${\mathrm{In}}_x{\mathrm{Ga}}_{1-x}\mathrm{N}$ peak with increasing QW width.

Figure 6

(a),(b) Experimental (squares) and simulated (diamonds) transition energy $E$ of the QW PL band as a function of the QW width $d$ for the Ga- and N-polar samples, respectively. (c),(d) Experimental (squares) and simulated (diamonds) intensity $I$ of the QW PL band for the Ga- and N-polar samples, respectively. The latter is obtained by scaling the calculated electron-hole overlap as described in the text.

Figure 7

WBDF micrographs of sample $I$ showing the entire sample structure (a) and only the PAMBE grown layers (b) and (c). The diffraction conditions are indicated in each figure. Mixed dislocations are visible in all of the images. $a$-type TDs are visible only in (a) and (c); $c$-type TDs are visible only in (b). The small arrows indicate the position of the $\vee$-pit defects.

Figure 8

STEM HAADF micrographs of the entire QW region (a) and the fifth QW (b) of sample $I$. In (b), the arrows indicate the position of the QW or QB interface. Both micrographs were taken along the $[11\overline{2}0]$ direction. (c) STEM HAADF intensity profile along the growth direction for the fifth QW. The red dashed lines indicate the interfaces between the QW and the QBs.

Figure 9

(a) HRTEM image of the QW region of sample $I$ taken along the $[1\overline{1}00]$ direction. The inset shows with higher magnification the interface between the third QW and the subsequent QB. (b) Color-coded map of the local lattice parameter $c$ distribution ranging from 500 pm (blue) to 550 pm (yellow).

Figure 10

Histograms of the local lattice parameter $c$ measured in a QB and two different QWs of sample $I$. The solid blue lines represent Gaussian fits to the experimental data.

Figure 11

Representative top-view monochromatic CL and superimposed SEM images of Ga- (a) and N-polar (b) ${\mathrm{In}}_x{\mathrm{Ga}}_{1-x}\mathrm{N}/{\mathrm{In}}_y{\mathrm{Ga}}_{1-y}\mathrm{N}$ QWs. The images shown are taken from samples $C$ (a) and $G$ (b). The CL images are taken at 10 K. For the N-polar sample $G$, the luminescence observed stems exclusively from the semipolar facets formed around TDs.