High-Temperature Growth of Bulk Gallium Nitride by the Sodium Flux Method
The progress in optoelectronics and power electronics devices and their industrial applications has been remarkable since the beginning of the 21st century. Gallium nitride (GaN) based devices have shown high efficiency and good reliability, but their full potential is hindered by the shortage of native substrates. The Na-flux technique has the ability to produce high quality and low cost bulk GaN. At UCSB, a novel Na-flux system was designed by Von Dollen et al. to monitor GaN growth in-situ. In addition, they demonstrated the fastest growth rate (>50 μm/h) reported for this technique. However, the main challenges such as low structural quality, high impurity, and opaqueness were usually noticed in the grown GaN. Furthermore, the poor control of growth and the low reproducibility of experiments limited the reliability of the system. The focus of this dissertation is on understanding the reasons behind these challenges, finding possible solutions, and testing the capability of the system to monitor and control the growth in-situ.
Higher structural quality and lower impurity transparent GaN growth was noticed on some regions of the seed related to its position within the system. First, higher structural quality was linked to the switch from 3D to 2D growth mode as was evident from the optical microscope images. For 2D growth, the full width half maximum (FWHM) of the (0002) and (112 ̅0) rocking curves were as low as 89 and 44 arcseconds respectively. This switch – from 3D to 2D – occurred at a late growth stage and the reason for the observed crystal quality improvement is speculated to be decreased thermodynamic driving force as growth progressed. Second, in optically transparent regions, concentrations of oxygen, sodium, and carbon were as low as 7×1016, 2×1016, and 5×1016 atoms/cm3, respectively, while the concentration of molybdenum was below the detection limit. In addition, the area, the thickness, and the reproducibility of transparent regions were increased by controlling the seed position and omitting the addition of carbon (a common additive in the Na-flux method). Furthermore, the opaqueness was found to correlate with oxygen impurity as it was established by secondary ion mass spectrometry (SIMS) analysis. GaN growths which contain oxygen concentrations above 1019 atoms/cm3 were highly absorbing regardless of the presence of other impurities. Thus, the effect of three oxygen getters – i.e. magnesium, aluminum, and titanium – on the GaN transparency was investigated but was not significant. However, polycrystalline GaN was found to preferentially form on titanium sheets; and the reproducibility of growth conditions was enhanced with titanium sheets in the Na flux reactor. The effective solution to grow transparent GaN was by depleting oxygen impurity in the system via extended GaN growths. Extended growth typically consisted of two consecutive single growth experiments; and it was found that the all second experiments were thermodynamically limited in contrast with many of the first or single experiments. Third, the second experiments of extended growths were used to inspect the thermodynamics of GaN growth in-situ using the UCSB system. The ability of the system to determine equilibrium pressure and temperature for a specific sodium-gallium melt composition was demonstrated. Furthermore, positive deviation of the melt from ideality was noticed and the gallium activity and activity coefficient were calculated based on experimental data. Finally, some changes and improvements to the system – based on the outcomes of this dissertation – were suggested for better monitoring and controlling of GaN crystal growth.