| Summary: | 碩士 === 國立中央大學 === 光電科學與工程學系 === 107 === In recent years, the rapid development of optoelectronic technology has made the demand for high-quality optical components higher and higher. In order to achieve these goals, optical films on components have become more and more important. Therefore, an excellent simulation software can not only optimize for the desired target value before coating, but also analyze the difference between the actual value and the ideal value for the failed film after coating.
In this research, simplex algorithm is selected to optimize optical thin films. It is well known that the simplex algorithm mainly converges to the optimal solution based on the three mechanisms: reflection, contraction and expansion. Therefore, the movement of vertices has a great effect on finding an optimizing solution. In this thesis, using the adaptive parameters which is the most effective way to improve it. The rate of successful optimization can enhance 10% by modified simplex algorithm (MSA). Besides, the optimized value can be situated in 0.08 deviation of the target.
When the spectrum of real coating is different from simulation, reverse engineering (RE) can effectively estimate the film thickness difference between these two. A simple spectral fitting can achieve the coefficient of determination, denoted R2 or r2 and pronounced R squared (RSQ) equal to 1.
Intelligent layering simplex algorithm (ILSA) can be used in the problems that the difference between the initial spectrum and the target is large. Its layering method is a new concept that has higher accuracy of optimization in the desired spectrum. Most important of all, this way can avoid unreasonable thickness that thinner than 5nm and the optimized value can be situated in 0.23 deviation of the target. ILSA is a more realistic algorithm within thin film coating regime.
|
|---|
