Figure 1

Sketch of the sample geometry, consisting of stripes with triangular cross section. The stripes are oriented along the GaN $a$ direction and have naturally stable semipolar $\{1\overline{1}01\}$ side facets. After CL investigation of a cross section formed by cleaving (black hatched area) and of the side facets (red hatched area), a thin lamella was prepared from the same position by focused ion-beam etching.Reuse & Permissions

Figure 2

Low-temperature CL spectrum (lower part), corresponding SE micrograph (top, left), and low-temperature CL images taken at $3.32\hspace{0.5em}\mathrm{eV}$ (top, middle) and $3.41\hspace{0.5em}\mathrm{eV}$ (top, right) emission energy from the cross section of the same stripe. The spectrum was recorded at the position where the $3.32\hspace{0.5em}\mathrm{eV}$ band was found. The emission pattern at $3.41\hspace{0.5em}\mathrm{eV}$ (BSF of type $I_1$) is visible as a bright small spot together with a horizontal bright line oriented perpendicularly to the (vertical) $c$ direction. The emission at $3.32\hspace{0.5em}\mathrm{eV}$ (BSF of type $I_2$) shows up as a series of aligned bright spots horizontally elongated close to the surface. The two emission patterns are not correlated spatially.Reuse & Permissions

Figure 3

Low-temperature CL distribution around $3.32\hspace{0.5em}\mathrm{eV}$ in higher magnification ($20\hspace{0.5em}000\times$). The picture shows the same sample region as visible in Fig. 2, middle part. The spots appear in varying sizes and shapes with a slight elongation perpendicular to the GaN $c$ direction. This can be seen easily having a closer look at the spot marked with a red arrow. The dashed line denotes the sample surface.Reuse & Permissions

Figure 4

Combination of CL image recorded at $3.32\hspace{0.5em}\mathrm{eV}$ on the side facet (right) and TEM image of the cross section (left) showing a part of the thin lamella cut by FIB (Fig. 1). CL was recorded from the side facet as indicated in Fig. 1. In the TEM micrograph, BSFs are visible as almost horizontal narrow white lines. The topmost pair of them framed by a white rectangle is shown in HRSTEM micrographs (Fig. 6) and is identified as BSF of type $I_2$. For exactly this position CL emission maps show exclusively a band at $3.32\hspace{0.5em}\mathrm{eV}$. Note that the spatial resolution in CL is limited to about $200\hspace{0.5em}\mathrm{nm}$ in this image due to low magnification chosen to record simultaneously bottom and top of the side facet. The spatial distribution of the $3.32\hspace{0.5em}\mathrm{eV}$ emission is directly correlated with the $I_2$-type BSFs in the TEM micrograph.Reuse & Permissions

Figure 5

TEM micrograph of the FIB prepared lamella. The BSFs are marked with blue arrows. The topmost pair of BSFs is well separated from the others allowing their precise assignment to the emission at $3.32\hspace{0.5em}\mathrm{eV}$ (red curve).Reuse & Permissions

Figure 6

Analysis of $\langle 11\overline{2}0\rangle$ HRSTEM images (on the left) and TEM micrograph (right side) of the topmost two BSFs found in the thin FIB prepared lamella (Fig. 1). Both of them are identified as BSFs of type $I_2$. The emission at $3.32\hspace{0.5em}\mathrm{eV}$ is the only detectable one at this specific sample position, and no other defect-related emission was found here. This fact allows a one-to-one correlation between the CL signal at $3.32\hspace{0.5em}\mathrm{eV}$ and the existence of a $I_2$-type BSF.Reuse & Permissions

Figure 7

Temperature-dependent CL spectra in the spectral region of the $3.31\hspace{0.5em}\mathrm{eV}$ band on semilogarithmic scale. The labels give the nominal temperature as read from the sensor in the cold finger. The dashed lines are tangents to the high-energy tail of the spectra yielding an estimate for the apparent temperature $T_{\mathrm{app}}$. The arrows mark the expected shift of the maximum according to Pässler’s formula for the band-gap shift (Ref. 22) and adding $1/2\hspace{0.5em}{k}_{\mathrm{B}}{T}_{\mathrm{corr}}$ for the average free-carrier thermal energy [see Eqs. (1) and (2)], where $T_{\mathrm{corr}}=T_{\mathrm{nom}}+10\hspace{0.5em}\mathrm{K}$ is the estimated sample temperature calculated from the nominal sensor reading $T_{\mathrm{nom}}$ adding a correction of $10\hspace{0.5em}\mathrm{K}$. The horizontal bar marks the kinetic-energy contribution $1/2\hspace{0.5em}{k}_{\mathrm{B}}{T}_{\mathrm{corr}}$ to the shift of the maximum.Reuse & Permissions