Department of Materials Science and Engineering
Advisor: Subhash Mahajan
11:00 AM, Friday, September 23, 2016
1003 Kemper Hall
III-nitride semiconductors – gallium nitride, aluminum nitride, indium nitride, and their alloys – are critical to our energy future. High efficiency, high brightness light emitting diodes made with III-nitrides are replacing the Edisonian light bulb, and high power, III-nitride transistors are poised to drastically reduce power conversion losses across the grid. Since the development of growth methods for high quality gallium nitride on sapphire in 1989 by Akasaki, Ammano, and Nakamura, progress in this field has been extremely rapid. However, due to the unavailability of large diameter, native substrates, III-nitride devices are still typically grown on silicon carbide, sapphire, or silicon wafers. (111) silicon is considered by many to be the most favorable substrate for commercial applications because of its low intrinsic cost and availability in large wafer sizes. III-nitride growth on silicon also presents opportunities for the direct integration of III-nitride devices with silicon microelectronics and offers additional benefits such as easy processing, the ability to modify substrate conductivity, and improved thermal management compared to III-nitride on sapphire technology.
Unlike sapphire and silicon carbide, (111) silicon has a positive lattice mismatch and a negative thermal expansion mismatch with the III-nitrides. These properties cause tensile strain to accumulate in epitaxial layers during growth and can induce cracking when films are grown with thicknesses greater than about 1 μm. Gallium also reacts with silicon at high temperatures. This reaction causes a roughening of the substrate surface and limits the thermal stability of gallium nitride – silicon interfaces. These issues have been circumvented, in part, through the use of aluminum nitride buffer layers, which protect the substrate from gallium and induce compressive strain in overgrown gallium nitride. The growth of smooth buffer layers with low defect densities, however, is difficult due to an extremely large lattice mismatch between aluminum nitride and (111) silicon (+19%). It was found that exposing the substrate to a small dose of aluminum prior to aluminum nitride growth can reduce surface roughness and improve the crystal quality of the buffer, however the mechanisms responsible for this have remained controversial and, to a large extent, misunderstood.
In this talk, the structure and morphology of aluminum predoses after ammonia exposure will be discussed and a model will be presented for the mixing of aluminum, silicon, and nitrogen under prescribed conditions. Subsequent low temperature and high temperature aluminum nitride growth will be examined with an eye toward explaining the anomalous aluminum nitride – silicon interface structures observed by others. Dislocations in buffer layers grown both with and without aluminum predoses will be shown and it will be proposed that enhanced dislocation motion and associated annihilation reactions lead to improved crystal qualities and surface topographies in buffers grown with a predose. Consistent with III-nitride growth on alternative substrates, threading dislocations with line directions normal to the basal plane were found to terminate within highly defective, low temperature nucleation layers. Possible mechanisms for their formation, along with supporting evidence, will be discussed.