Thesis Statement:

The objective of this research project is to establish a classification scheme for the surfaces of epitaxially grown highly lattice mismatched semiconductor materials. These materials exhibit a cross hatched surface pattern which is not yet well defined or understood. Initial studies performed for this project have shown that various growth conditions affect the cross hatch pattern, and so the focus of this research will be to study this relationship. Part of the research will be to determine what kinds of measurements are useful for characterizing surfaces and relating surfaces to growth conditions. Ultimately, electrical data that has already been collected by Professor S.M. Lord's colleagues at Penn State will be used to try and resolve a relationship betwee the cross hatched surfaces, growth conditions, and electrical characteristics of the samples being studied.

Background:

The reference to growth of semiconductor materials is not biological in any way. It implies an epitaxial or layer by layer deposition of one or more elements on a substrate. A substrate is a semiconductor wafer that serves as a foundation on which to build layers of other materials. Choices of substrates are limited since there are only a few materials commercially available. One state of the art growth technique is Molecular Beam Epitaxy (MBE), a process which is capable of depositing layers of material as thin as only a few atoms. The benefit of such a process is a high degree of uniformity across an entire wafer. This uniformity is particularly desirable for fabrication of device arrays.

One of the most significant problems in epitaxial growth processes such as MBE is mismatched lattice parameters of the materials to be interfaced. The lattice parameter is simply the spacing between the atoms of a material. If the spacing between two different materials is not exactly the same, there will be defects known as dislocations at the interface, and thus degraded electrical properties.

One solution to this problem is to use a CGBL (Compositionally Graded Buffer Layer) structure to reduce the strain at the interfaces of each deposited layer (1). It is possible to deposit layers of a material, InGaAs for example, with each subsequent layer being of a slightly different composition of In and Ga. The slight change in composition results in a slight change in the lattice parameter, so that as each layer is deposited, it will have a lattice parameter very close to the layer below. The result is a relaxed structure with few threading dislocations.

Significance:

The materials that will be studied are semiconductors used to make photodetectors. Photodetectors are devices that convert light energy into electrical current. The materials used to make the photodetectors are sensitive to light of a specific energy or energy range. When light with high enough energy is incident on the material, it is possible to measure the current produced an thus detect the light. The important matereial property of a semiconductor that determines what energy is required to created current is known as the bandgap energy Eg. Different semiconductors have different bandgap energies, and therefor are sensitive to different light energies (2).

The problem with photodetectors is that there is not a simple material for every desirable energy. For instance, silicon (the most well known semiconductor), is only usable for a small range of energies. So to detect other energies of light, it is necessary to combine various materials to get the appropriate bandgap energy to detect that type of light. However, most of the materials that need to be combined have different lattice constants and therefore have many dislocations when interfaced. This is why the CGBL technique is used to relax these interfaces through gradual lattice constant changes.

One of the most prevalent effects of highly mismatched epitaxy is the cross hatch surface pattern, as opposed to the relatively smooth (i.e. flat) surfaces of ordinary semiconductor materials. This pattern varies with growth conditions and is not yet fully explained. A major issue and concern is to find a set of conditions under which to grow materials which can be fabricated into devices with optimized electrical characteristics. It has already been shown that different growth conditions result in variations of the cross hatch pattern, so it is important to determine what types of measurements will be useful in describing how the surfaces vary. "Although the cross hatch has been well-known for 20 years, quantitative studies of the associated surface roughness have not been pursued (3)." Measurements such as roughness will be used to deduce a relationship between surfaces and electrical data, thereby providing a system of characterization for these materials. A successful classification scheme will provide researchers with the knowledge needed to produce devices such as photodetectors with enhanced electrical performance.

Methodology:

The primary characterization tool used in this study will be a relatively new surface imaging technique called Atomic Force Microscopy (AFM). The atomic force microscope was introduced in 1985 by Gerd Binnig and Christof Gerber of the IBM Zurich Research Laboratory, along with Calvin F. Quate of Stanford University (1). This technique is particularly useful in that it is non-destructive and offers imaging at very high resolution. AFM incorporates the use of a very small probe which is physically dragged across the surface of the sample. A laser diode produces a light beam which is directed onto a cantilever to which the probe is attached. A photodetector module is used to collect the light reflected off of the cantilever as it scans across a sample surface. The photodetector has four quadrants and is able to detect the change in the deflected light's position as the cantilever is deflected by the sample surace, thus providing a way to physically map the surface features of the sample with great resolution.

The samples supplied for this study have been grown with different buffer thicknesses as well as different buffer layer and active layer temperatures. The initial phase of the investigation will be to attain numerous images of every sample with computer imaging software. The types of measurements (e.g. roughness and line spacing) that will be useful in characterizing the cross hatched surfaces will be determined. The computer software will be used for image capturing as well as measurement and analysis. The next step will be to analyze all the data collected for the samples and begin to draw conclusions about how the cross hatch changes for the various growth conditions. To arrive at a good method for classification it will be necessary to find distinct differences between various measurements for different samples. It will be important to take measurements on several areas of each sample area to insure that the results are statistically reliable.

The final step in the project will be to associate the surface findings to electrical data that has already been measured at Penn State. Ideally, it will be possible to link one or more of the growth conditions to an improvement in electrical performance. If this is achieved, it will be possible to define which growth conditions or what surface features result in improved material performance.

Summary:

The objective of this research will be to attain a better understanding of the cross hatched surface characteristics of a highly mismatched CGBL epitaxial film. It is important to understand how the surface is related to growth conditions as well as how electrical performance is affected by the surfaces of these materials. AFM is an ideal technique for this study and preliminary images taken at Bucknell have suggested that suitable images can be obtained. If a correlation between electrical properties and surface features can be made, it could be used in fabricating real devices with improved electrical and optical properties.

 
1 J. Tersoff, Appl. Phys. Ltt. 62, 693 (1993).
 
2 Jasprit Singh, Optoelectronics: An Introduction to Materials and
Devices, McGraw Hill Companies Inc., New York, 1996, pg. 146-148.
 
3 J.W.P. Hsu, E.A. Fitzgerald, Y.H. Xie, P.J. Silverman, and
M.J. Cardillo, Appl. Phys. Ltt. 61, 1293 (1992).
 
4 H. Kumar Wickramasinghe, Scientific American, 261, 99 (1989).
 
back to the main page
to the EE senior design page
last updated 7/9/97