STUDENT PROJECTS

Introduction

AlGaN is a wideband semiconductor with potential for applications in optical-electronic applications. The band gap of the material can be modified by adjusting the Al content in the material. The chemical composition of the material is generally written as Al_xGa_(1-x)N, where x can range from 0 to 1. Zero signifies that the material is GaN and x=1 signals that the material is AlN.

During the formation of the material the flow rate (flux) of the Al organic can be adjusted. It is currently unknown how such an adjustment will affect the final Al content in the material. The purpose of this research is to find a relationship between the Al flux used during formation of a crystal and the Al content present in the final product. X-ray diffraction and Rutherford Backscattering are used to analyze the samples.

Procedure
Five AlGaN crystals were grown using MOCVD. Each was grown from a Sapphire substrate and not all samples were of the same structure. Each sample varied in layer thickness and in the flux used in the formation of the AlGaN layer. The Al flux was allowed to vary, but the GaN flux was held constant at 26 mL/min.

The five samples have the structures seen to the right.

The AlN interlayer in all samples containing it was less than 15nm thick while the GaN and AlGaN layers were greater than 500nm.

Analysis
X-ray diffraction of the samples was carried out using a theta-2 theta scan. The range of data considered was in the 2 order, so for Bragg’s Law, n=2, which reduces to lambda=d*sin(theta). Further, reflection was from the (002) planes, hence the distance between the reflecting planes was just equal to the lattice constant c of the wurzite crystal structure. So, given that a peak appeared for each layer, thus yielding a value for theta, a calculation of the lattice constant c for each material could be made. Next, by using the equation C_AlxGa(1-x)N=x*C_AlN+(1-x)*CGaN, one could calculate the Al content x.

In the x-ray data the AlN data never appeared so in order to make the calculations of x, the value of the lattice constant for AlN was assumed to be 4.982 Angstroms.

The data for the x-ray diffraction is as follows.

From this there seems to be no trend in the data. Sample 4 is yields by far the largest value of x while sample one yields the median value, yet sample 1 and 4 have equal flux values. Further, the first three samples have similar structures yet there does not seem to be any sort of trend apparent in those data points.

The Rutherford Backscattering data seems to yield a little more interesting results. RBS analysis for these samples is very practical because the Ga atoms are so much larger than the Al or N atoms in the material. So, the ion beam can travel through the material and make a very noticeable pattern of the Ga content in the material. As the ion beam reflects off of Ga atoms deeper in the material the signal of the reflected beam has less energy. This relationship is linear.

RBS data for sample one is shown in Figure 1. This is for sample one which has an AlGaN layer thickness of around 530 nm and a GaN layer thickness of 690 nm. The amount of Al flux used during formation was 35 mL/min. The Al content for this sample was calculated to be from .19 to .25.

The black data represents the Ga content in the material and the green line is a simulation of the data. The right vertical line represents the surface of the AlGaN material (the reflected beam has the highest energy here). The small vertical line at about 1.65 MeV represents the interface between the GaN and AlGaN layers. There should be a dip in the data at this point due to the AlN interlayer, however, it was not detected. The left vertical line near 1.0 MeV represents the interface between the GaN and the Sapphire substrate.

Calculations for the Al content in each sample was carried out by a pre-written program. These values can be calculated by knowing what the atomic numbers are for the incident beam and the nuclei being hit. For this case the incident beam was He ions and the nuclei were of the Ga atoms.

The data from the RBS analysis is as follows. The Chi value gives an indication of the quality of the crystal structure.

There is no obvious trend in the data, however, there is perhaps more to say than there was with the x-ray diffraction data. If sample 3 could be ignored a fairly nice relationship between the flux and the Al content might arise, however, this is illogical to do as the highest quality sample is sample 3 as indicated by its low Chi value.

Discussion
Below is a comparison of the RBS data and the X-ray data.

As was briefly mentioned above, there was no AlN interlayer detected in the RBS data. However, there does seem to be some indication of the layer although not as expected. In all of the samples there is a fairly sharp transition from the GaN layer to the AlGaN layer. In sample 4, however, which does not contain a thin AlN layer, the transition is much more soft. The RBS data for sample 4 is shown in Figure 2 and this feature can be compared to other RBS data throughout this paper.

This sample is also unique in that the surface of the AlGaN layer begins to taper off as the layer becomes thicker. There are two proposed explanations for this, both of which may account for this effect. Because of the lack of the AlN layer, the AlGaN layer may have many dislocations. Thus, as the layer is grown thicker, the structure of the layer falls apart leading to such a drop. The other explanation is that there may just be a critical thickness to which the AlGaN layer can be grown. None of the other samples were grown with as thick of a AlGaN layer. Over the thickness range that is comparable to the other samples the layer seems to show similar characteristics to the other samples, however, beyond that thickness is where the Ga content in the material seems to drop off critically.

Getting back to the initial purpose of the research, let us examine sample 1 and 4 more closely. Both crystals were grown with a flux of 35 mL/min, however, the x value calculated in sample 4 is greater. This is easily explained. The drop off in the AlGaN layer of sample 4 indicates an increase in the amount of Al in that layer, and thus the x value calculated for that sample was higher. Had that drop off not been there, the x values for sample 1 and 4 would have most likely been much more similar.

Sample 2, which had a flux of 46 mL/min is shown in Figure 3. In the RBS data the AlGaN layer curve has a very large slope indicating that the Al content in the material is varying greatly. As the layer is grown thicker, the Al content in the crystal increases for this higher value of flux. This seems reasonable. It can further be explained in that Al rich AlGaN has a smaller lattice constant that Ga rich AlGaN. So, near the interface between GaN and AlGaN layers, the AlGaN tries to have a larger lattice constant by being Ga rich. However, as the layer is grown the strain is relaxed and the Al content in the material increases.

Although the explanation given above for explaining the sample 2 data and the effects of higher flux, samples 3 and 5 which have a flux of 58 do not show similar behavior. The data for sample 3 and 5 is shown in Figure 4 and Figure 5 respectively. There seems to be some other effect taking place at the higher flux regime. More data is needed between flux of 46 and 58.

The values for the Al content calculated for these two samples vary drastically. This is a very strange phenomenon and cannot be explained by relaxing of strain as the crystal layer is grown. There appears to be some effect concerning the layer thickness. In sample 5 there are two thinner AlGaN layers as opposed to one thick layer.

Conclusion
So far there is no determining evidence of a clear relationship between the Al flux used during formation and the Al content present in the final product. However, that is not to say that such a relationship does not exist. The two 35 flux samples yielded similar results and the results would have been more similar had the AlGaN layer not had a jump in Al content near the surface. There simply needs to be more samples made with flux values between 35 and 58.

The most interesting phenomenon is that of layer thickness. It seems to play some role in the 35 flux samples, but for the 58 flux samples it drastically affects the Al content in the material. There needs to be more research on this phenomenon. Samples with equal flux need to be made with varying layer thickness and visa versa.