The technique uses the fact that the GaN RHEED pattern is less intense when there is unreacted Ga on the sample surface disturbing its periodicity. A good way to visualize what happens to the RHEED pattern when Ga starts building up on the sample surface is to consider a perfect GaN surface under Ga and NH_3 flux, at a substrate temperature that is low enough so that the sum of the evaporating Ga flux Gamma_(Ga evap) and incorporating Ga flux Gamma_(Ga inc) is less than the incoming Ga flux Gamma_(Ga in).
Under these conditions the first layer of Ga starts building up, trying to maintain the periodicity of the underlying GaN lattice. The Ga molecules though move quickly between preferential sites due to their high mobility at the substrate temperatures used. During the buildup of the first layer the intensity of the RHEED pattern drops sharply until the whole surface is covered with Ga. The remaining RHEED pattern intensity is due to Ga atoms residing longer at preferential sites given by the underlying crystalline GaN surface than in-between. In addition, the RHEED beam probes to some extend beyond the Ga layer and still "sees" some of the underlying crystalline GaN surface.
Subsequent layers of Ga will be progressively disordered approaching the random ever changing arrangement of the liquid state. This causes the weak remaining RHEED pattern to further decrease in intensity until the RHEED beam fails to "see" any periodicity dictated by the crystalline GaN surface, and the surface itself.
If a high enough substrate temperature is chosen so that all the Ga hitting the surface will reevaporate or form GaN with the available NH_3 the surface will only be covered by some fraction of full Ga coverage at any time. In this situation the RHEED pattern intensity varies as shown in Figure 28a and Figure 29, decreasing upon opening of the Ga shutter by a fixed amount and remaining at that level until the shutter is closed again. This phenomena will be termed a `Dip' from now on, and the recovering of the original intensity after the Ga shutter is closed will often be termed `Recovery'. The smaller the amount of Ga on the surface available to react, disturbing the periodicity of the surface seen by the RHEED beam, the smaller the decrease in intensity of the pattern. It is suggested that the decrease in RHEED intensity is proportional to the fraction of Ga coverage of the GaN surface.
If the Ga flux and substrate temperature is adjusted in such a way that the surface is always fully covered by 1 monolayer of Ga atoms, then the incoming flux equals the outgoing flux, given by the sum of the evaporating flux and the incorporating flux. In this situation the intensity goes as shown in Figure 28b and Figure 29, dropping to a level that indicates complete Ga coverage of the surface. This situation will be called the balance point from now on. A low balance point simply means that a combination is chosen of low Ga flux and low substrate temperature, and a high balance point means that a combination of high Ga flux and high substrate temperature is chosen. Note that every Ga and NH_3 flux combination has a unique substrate temperature at the balance point, and vise versa. Note that the NH_3 flux can be neglected as discussed later.
If the Ga flux is too high for a given substrate temperature and NH_3 flux, or the substrate temperature is too low for a given Ga and NH_3 flux, Ga starts accumulating on the GaN surface. As explained above, the RHEED pattern intensity drops sharply to a level indicating full Ga coverage of the GaN surface and then continues to degrade as shown in Figure 28c and Figure 29. When closing the Ga shutter the excess Ga evaporates from the surface until the intensity level indicating a completely Ga covered surface is reached, after which the intensity rises quickly to the level indicating that all excess Ga left exposing the crystalline GaN surface as shown in Figure 28c and Figure 29.
It is noted that it was found that if growth is conducted just below the balance point GaN continues to form, but in a 3D polycrystalline fashion with preferred orientations, and recovery of the crystalline GaN surface is not possible after closing the Ga shutter if growth took place over an extended time.
It was also found, as discussed later, that the incorporating Ga flux Gamma_(Ga inc) is much less than the evaporating Ga flux Gamma_(Ga evap). Therefore the balance point does not change noticeable with or without NH_3 flux Gamma_(NH_3 in).
Figure 28: Optimizing Ga Flux using RHEED Intensity (Sketch)
Figure 29: Optimizing Ga Flux using RHEED Intensity
Table of Contents