Complications in silane-assisted GaN nanowire growth

Understanding the growth mechanisms of III-nitride nanowires is of great importance to realise their full potential. We present a systematic study of silane-assisted GaN nanowire growth on c-sapphire substrates by investigating the surface evolution of the sapphire substrates during the high temperature annealing, nitridation and nucleation steps, and the growth of GaN nanowires. The nucleation step – which transforms the AlN layer formed during the nitridation step to AlGaN – is critical for subsequent silane-assisted GaN nanowire growth. Both Ga-polar and N-polar GaN nanowires were grown with N-polar nanowires growing much faster than the Ga-polar nanowires. On the top surface of the N-polar GaN nanowires protuberance structures were found, which relates to the presence of Ga-polar domains within the nanowires. Detailed morphology studies revealed ring-like features concentric with the protuberance structures, indicating energetically favourable nucleation sites at inversion domain boundaries. Cathodoluminescence studies showed quenching of emission intensity at the protuberance structures, but the impact is limited to the protuberance structure area only and does not extend to the surrounding areas. Hence it should minimally affect the performance of devices whose functions are based on radial heterostructures, suggesting that radial heterostructures remain a promising device structure.

. XPS spectra of Ga 2p measured from the samples with growth process terminated after the high temperature annealing step (Step A), nitridation step (Step B) and nucleation step (Step C), labelled as Annealing, Nitridation and Nucleation, respectively. Ga 2p peak was only observed from the Nucleation substrate. Figure S3. XPS spectra of N 1s measured from the samples with growth process terminated after the high temperature annealing step (Step A), nitridation step (Step B) and nucleation step (Step C), labelled as Annealing, Nitridation and Nucleation, respectively.
No N 1s peak was observed from the Annealing substrate. The N 1s peak from the Nitridation substrate was fitted with one Pseudo-Voigt distribution centred at 396.93 eV, attributing to the N-Al peak. The peak from the Nucleation substrate was best represented by fitting it with two Pseudo-Voigt distributions, centred at 397.18 eV and 398.12 eV, respectively. The 397.18 eV peak is from the N-Al bond 1,2 while the 398.12 eV is likely from the N-Ga bond. 3 Figure S4. XPS spectra of Al 2p measured from the samples with growth process terminated after the high temperature annealing step (Step A), nitridation step (Step B) and nucleation step (Step C), labelled as Annealing, Nitridation and Nucleation, respectively.
All three spectra were dominated by the Al-O bonds, most likely from the sapphire substrate and the surface oxidation. The Al-O bond energy (73.89 eV) from the substrate after Annealing is slightly lower than that from the substrates after Nitridation (74.17 eV) and Nucleation (74.18 eV). This indicates that the Al-O bonds on the Nitridation and Nucleation substrates are more stable than that on the Annealing substrate. The lower Al-O bond energy for the Annealing substrate could be due to the contribution from the lattice deformation resulting from oxygen desorption.
The lower energy peak for the Annealing substrate was attributed to the Al-Al bond. 4 The lower peaks for the Nitridation and Nucleation substrates were likely from the Al-N bond. 5 Figure S5. XPS spectra of O 1s measured from the samples with growth process terminated after the high temperature annealing step (Step A), nitridation step (Step B) and nucleation step (Step C), labelled as Annealing, Nitridation and Nucleation, respectively.
All three peaks were fitted with one Pseudo-Voigt peak. The peaks were centred at 530.53 eV, 531.07 eV and 531.01 eV for the Annealing, Nitridation and Nucleation substrates, respectively. The peaks from the Annealing and Nitridation substrates were attributed to the O-Al bond. 6,7 The energy of the O-Al bond from the Annealing substrate is shifted towards lower energy, consistent with the shift of Al-O bond in Figure S4  The reflectivity of both samples started dropping immediately when TMGa was introduced into the reactor, likely due to the formation of 3D islands. For the sample without growth interruption, the reflectivity continued dropping as the growth parameters (silane introduction and temperature) were adjusted to promote nanowire growth, indicating the GaN islands remained present at the onset of Si-assisted growth in that case. When a growth interruption was employed, the reflectivity stopped dropping as soon as the TMGa flow stopped. Within 7 seconds of the growth interruption, the reflectivity recovered to its full value, suggesting a complete re-planarization of the substrate during the growth interruption.
Nanowires grew successfully in both samples, even after a growth interruption was employed to remove the island structures that formed in the nucleation step.
Although locally minor differences in the size and density of the nanowires can be observed across both samples, a statistical analysis across an area of about 1 cm 2 per sample has revealed no significant differences between them. The densities of the nanowires in both samples were measured to be ~ 510 5 cm -2 .