Logo image
面內外整合型電子光斑量測系統之建置與驗證
Thesis

面內外整合型電子光斑量測系統之建置與驗證

魏守勤
Masters, 國立清華大學, 動力機械工程學系
2010

Abstract

電子光斑 四點彎折 相平移法 相位重建 光路設計 Electronic Speckle Pattern Interferometry ESPI Four-Point Bending Phase-Shifting Phase Unwrapping
As the progress of semiconductor manufacturing technology, even the mechanical components in MEMS (Micro Electro Mechanical System) are fairly small and shrink to micro-scale. Accompanied with the size reduction, how to measure the deformation of a package or a device becomes an issue. Therefore, this study proposes an improvement of the ESPI (Electronic Speckle Pattern Interferometry) light path design that switches between ESPI in-plane/out-of-plane measurement without moving and rotating any optical devices. And phase-shifting is built in the light path as the phase extraction method and phase map quality enhancement. In order to verify the feasibility and reliability of the novel light path design, four-point bending test with silicon specimens is performed. Numerical results obtained by finite element software (ANSYS®) and the theoretical solution from Euler-Bernoulli beam equation is also brought in for affirmation. By comparing the three results with different methods, the 2D ESPI light path design can be validated. However, there are still lots of singular points (noise) in the phase map acquired from phase-shifting, so the sine/cosine image smoothing algorithm is introduced in image post-processing. And the quality guided phase unwrapping algorithm is implemented for generating the continuous phase map. Because the linear relationship between continuous phase map and displacement map is established when light path is set, the whole in-plane/out-of-plane displacement map is obtained as the continuous phase map is obtained. Compare the experiment results with FEM and theoretical results, the errors of all tests are lesser than 5%, and the standard variation is minor . It shows that the light path proposed in this study is feasible and stable.

Metrics

1 Record Views

Details

Logo image