IN SITU STRESS MEASUREMENTS DURING MOCVD GROWTH OF ALUMINUM GALLIUM NITRIDE ON SILICON CARBIDE

Open Access
Author:
Acord, Jeremy Daniel
Graduate Program:
Materials Science and Engineering
Degree:
Doctor of Philosophy
Document Type:
Dissertation
Date of Defense:
May 24, 2007
Committee Members:
  • Joan Marie Redwing, Committee Chair
  • Elizabeth C Dickey, Committee Member
  • Christopher L Muhlstein, Committee Member
  • Mark William Horn, Committee Member
Keywords:
  • METAL ORGANIC CHEMICAL VAPOR DEPOSITION
  • III-V NITRIDES
  • STRESS MEASUREMENT
  • EPITAXY
Abstract:
Al(x)Ga(1-x)N is a wide band gap compound semiconductor material with the desirable properties of high chemical and thermal stability, as well as a direct band gap. This makes Al(x)Ga(1-x)N particularly suitable for fabrication of deep-ultraviolet (UV) optoelectronic devices. However, there are currently no readily available bulk group III-nitride substrates, hence growth must be carried out on foreign substrates, such as sapphire or silicon carbide. This leads to epitaxial misfit between the film and substrate, as well as strain due to mismatch in the coefficients of thermal expansion (CTE) during a temperature change. Cracking is often attributed to these differences between film and substrate, and also between Al(x)Ga(1-x)N layers of different compositions. The important role of growth-related stress is less obvious and not as well understood in this process. Due to the dynamic nature of these stresses, it is often difficult to understand their behavior from post-growth measurements. Hence it is valuable to study these stresses using an in situ measurement technique. To fabricate devices that are active at various UV wavelengths, a range of Al(x)Ga(1-x)N compositions must be deposited, often within the same device structure. Empirical observations have shown that Al(x)Ga(1-x)N layers deposited on such substrates have a tendency to crack that increases with the aluminum content of the alloy and with the silicon doping level needed to make n-type conducting material. Therefore the primary goals of this study were twofold: to develop a fundamental understanding of the mechanisms contributing to stress evolution during deposition of Al(x)Ga(1-x)N thin films by metalorganic chemical vapor deposition (MOCVD) on SiC substrates, and to understand the dependence of cracking on the Al-content of Al(x)Ga(1-x)N. In situ substrate curvature measurements performed with a Multibeam Optical Stress Sensor (MOSS) were used throughout this study to further this goal. It is noteworthy that no prior literature reports exist studying MOCVD of Al(x)Ga(1-x)N on SiC using in situ techniques. The first portion of this study investigated the changes in the stress evolution as the Al-fraction of the Al(x)Ga(1-x)N thin films was varied. Gallium-rich alloys were observed to follow the pattern reported previously in the literature determined by post-growth techniques. GaN deposited on AlN buffer layer initiated growth in compression due to epitaxial misfit that was partially relaxed by formation of misfit dislocations near the interface. The growth stress relaxed with increasing GaN thickness until the surface became largely stress-free at growth temperature after growth of approximately 3 µm. As a consequence of this direct measurement, it was determined that cracking in thick GaN layers is due solely to the CTE stress, and not due to any growth stress. In contrast, the growth stress in Al-rich Al(x)Ga(1-x)N deposited on AlN was strongly determined primarily by the stress and defect structure of the AlN buffer layer, in combination with a small epitaxial misfit. For sufficiently high Al-fraction, the AlxGa1-xN growth stress was tensile from the start of growth. For compositions in between these two extremes, the stress evolution contained elements of both behaviors. This was termed “Al(x)Ga(1-x)N” type behavior. As with Ga-rich films, the initial growth stress was compressive. However, the stress evolved into tension toward the end of growth. The mechanisms from the two previously discussed regimes were determined to be active, in addition to a mechanism described in the literature as “dislocation effective climb”. This mechanism is active when pre-existing threading dislocations with an edge component propagate into the Al(x)Ga(1-x)N epilayer, becoming inclined as they do so leading to reduction in the initial compressive stress and eventually generation of tensile stress. This regime is also sensitive to the stress and structure of the AlN buffer layer. In the remainder of the study attempts were made to understand and modify the effect of the buffer layer. It was determined that high (>10,000) NH3 / Al-precursor ratio produced AlN buffer layers with improved quality, relative to lower ratios. This translated into reduced dislocation density and decreased tensile growth stress in subsequently grown Al(x)Ga(1-x)N epilayers. Compositionally graded buffer layers were investigated to attempt to increase the mean compressive stress at the end of epilayer growth. This technique was successful for GaN epilayers, with the thickest graded buffer layers producing the greatest effect. However, as the Al-fraction of the epilayer increased, the effectiveness of the graded buffer layer decreased. This was attributed to the rapidly decreasing strain energy available. One preliminary experiment demonstrated that dislocations become inclined when SiH4 was introduced during growth of Al(0.28)Ga(0.72)N, leading to increased tensile growth stress.