Text

theses

Gallium nitride (GaN) thin film is an attractive material for manufacturing optoelectronic device applications due to its wide band-gap and superb optoelectronic performance. The reliability and durability of the devices depend on the quality of thin films. Metal-organic chemical vapor deposition (MOCVD) process is a common technique used to fabricate high-quality GaN thin films. The deposition rate and uniformity of thin films are manipulated by controlling operating conditions and reactor geometry configurations. In this study, the epitaxial growth of GaN thin films on sapphire substrates (AL2O3) was carried out in two commercial MOCVD systems, a vertical rotating disk MOCVD reactor, and a close-coupled showerhead MOCVD reactor. Material characterizations have been done using Atomic Force Microscopy (AFM), X-ray diffraction (XRD), Scanning Electron Microscope (SEM), and Raman scattering to examine the surface morphology and crystal quality of GaN thin films. The growth rate and uniformity of GaN thin films are simulated based on a three-dimensional computational fluid dynamics (CFD) model. Transport phenomena and chemical kinetics of the GaN growth process are performed using a reduced chemistry model, which contains 17 gas phase, and 8 surface species participating in 17 gas phase and 17 surface reactions. Numerical simulation of the single wafer and multi-wafers reactors have performed. A comprehensive study of the influence of operating variables, including rotation rate of the susceptor, susceptor temperature, inlet velocity, the reactor pressure, and precursor concentration ratio, on the GaN growth process is carried out. Operating parameters that have significant effects on the growth rate and uniformity of GaN thin films are identified. The reactor pressure and flow rate of trimethylgallium (TMG) have a significant effect on the deposition rate. A high-quality thin film is obtained when pure H2 is used as a carrier gas. The high flow rate of pure N2 gas enhances the growth of GaN thin films at high reactor pressure. However, it decreases the uniformity of the GaN thin film and promotes carbon contaminations. Thus, using an appropriate mixture of H2 and N2 as a carrier can improve the deposition rate and quality of GaN thin films. The inlet design has a significant effect on improving the reactant species utilization and increases the growth rate. The proper distance between the inlet and the susceptor aids to decrease the temperature gradient and improve the stability of the flow above the rotating susceptor.
The optimization of GaN deposition rate and uniformity in the MOCVD process have represented in a surrogate model. Surrogate-based optimization is an effective technique to alleviate expensive computation experiments with fewer sample points. The response surface from simulation data with minimum error variance estimation is generated using the Kriging method. The optimization of GaN deposition is performed as a deterministic problem, without taking into consideration the uncertain input parameters and the corresponding output response. Also, the optimization under uncertainty of design variables is considered. Multi-objective optimization using a multi-objective genetic algorithm carried out to find optimal solutions. The results reveal that the proposed optimization formulation can generate Pareto frontier of conflicting objectives, thus providing reliable trade-off solutions for decision-makers.
The optimization of GaN deposition rate and uniformity in the MOCVD process have represented in a surrogate model. Surrogate-based optimization is an effective technique to alleviate expensive computation experiments with fewer sample points. The response surface from simulation data with minimum error variance estimation is generated using the Kriging method. The optimization of GaN deposition is performed as a deterministic problem, without taking into consideration the uncertain input parameters and the corresponding output response. Also, the optimization under uncertainty of design variables is considered. Multi-objective optimization using a multi-objective genetic algorithm carried out to find optimal solutions. The results reveal that the proposed optimization formulation can generate Pareto frontier of conflicting objectives, thus providing reliable trade-off solutions for decision-makers.

ETD_10129

Ph.D.

Includes bibliographical references

doi:10.7282/t3-6cxt-ph56

ETD doctoral

The author owns the copyright to this work.