There seems to be a lower limit to the substrate temperature at which GaN of good quality can be grown, which is around T_sub=730°C. For AlN this temperature seems to be around 900°C. All attempts so far to grow material at the balance point below that lower limit substrate temperature caused the film to become polycrystalline with preferred orientations or even amorphous as indicated by RHEED.
RHEED also indicates that GaN decomposition becomes significant at substrate temperatures above approximately 900°C, as discussed in an earlier section. When growth is stopped the streaky RHEED pattern becomes 3-D, indicating that the surface becomes rough due to GaN evaporating from the substrate. The streaky RHEED pattern can be recovered by growing on top of the rough surface.
It was also observed that a 3-D surface stays 3-D at a relatively low substrate temperature and close to the balance point, but becomes 2-D at higher substrate temperatures and above the balance point. This might be explained by the higher diffusion rate at high substrate temperatures and a step flow or island growth mode, while at high growth rates close to the balance point there is not enough time for diffusion to occur to permit smooth growth. A streaky 2-D pattern usually stays that way even at relatively low substrate temperatures and close to the balance point, indicating that the diffusion time is sufficient to even out unregular parts of the surface, which is not possible on extremely rough surfaces.
It was also observed that the growth quality at the balance point depends on how the growth conditions were achieved, as discussed in an earlier section about `Dip' hysteresis. If growth is conducted after any growth below the balance point, even though the surface was recovered by giving it time to evaporate excess Ga, the RHEED pattern seems to degrade slowly over time slightly above the balance point (about 20°C). This is the condition where a `Dip' is observed. On the other hand, if growth is conducted under the same conditions slightly above the balance point (about 20°C), but previous growth was conducted way above the balance point (about 70°C or more), the RHEED pattern does not degrade. This is the condition when no `dip' is observed. More studies are needed to confirm those observations and to obtain quantitative results.
It was further found that samples grown above the balance point featuring a streaky 2-D pattern indicating good crystalline quality were insulating, while samples grown slightly below the balance point featured a high n-type carrier concentration, which was especially true for samples grown at low substrate temperatures where growth seems to be worse. The n-type carrier concentration seems to be a function of the crystal quality. Samples that grew fairly bad (somewhat below the balance point) were 10^18 n-type with a low mobility around 20, while samples grown slightly below the balance point were roughly 10^17 n-type with a much improved mobility of around 80. Samples grown above the balance point are usually insulating and no mobilities could be determined.
The reason why the samples were insulating could be due to their low content of Nitrogen vacancies, but there is some concern that this effect is caused by Antimony (Sb) contamination in our system (from a previous spill), since Sb is known to be a deep level dopant which could be able to compensate n-type GaN. Sb showed up before on Auger scans, but not on scans taken lately. It should also be mentioned that some Sb contamination showed up on the surface using RBS, which could have been deposited though after the cryoshroud was shut down in the growth chamber while the sample was still in the chamber, causing the still warm sources to evaporate Sb contamination on the cryoshroud. A SIMS study conducted did not detect any Sb though. Further studies are planned to get a grip on this issue as well as the cleanup of the spill if possible.
During growth GaN on c-plane sapphire experiences a weak 2x2 reconstruction. Under NH_3 flux this reconstruction becomes increasingly intense upon cooling of the sample and a sharp transition to 1x1 occurs at a temperature depending on jet conditions. Further studies are planned. It was also noticed that samples left overnight in the growth chamber at room temperature reconstruct 3x3.
Judging from RHEED images it was found that AlN grows well on c-plane sapphire, while growth is worse on a-plane sapphire. More studies are needed though. The growth procedure established involves growth of an AlN buffer layer on sapphire until the RHEED pattern does not improve anymore. Further studies are needed to obtain crystal quality data as a function of AlN buffer layer thickness. No studies were conducted so far on exploring GaN growth initiation with a GaN buffer layer as used by several groups. Koverspike et. al. (Ref. 45) found that GaN grown on an AlN buffer layer has superior crystalline properties than when grown on a GaN buffer layer.
It has been observed that the RHEED electron beam pushes the balance point up by about 10-20°. There are also indications that the growth rate might be slightly higher a few degree above the balance point than at the balance point. III-Nitride RHEED oscillations were so far not observed in our group, or by any other group, which is attributed to an insufficient surface quality.
Table of Contents