This paper presents an automatic alignment system based on machine vision method. A high-speed gray pattern match algorithms is proposed based on the combination of sequential similarity detection algorithm (SSDA) and multiresolution pagoda structure algorithm (MPSA). A dynamic system calibration model suitable for the algorithm automatic alignment system is established to relate the pixel coordinate in CCD to the physical coordinate, which is based on Tsai's two-step algorithm but with the help of precise positioning of the wafer stage in X-Y directions. A lot of experiments conducted on a machine vision experimental platform confirm that the proposed technique is feasible and effective. The pattern match algorithm is demonstrated to achieve an error less than one-twentieth pixels, while the computation time is shorter than 200ms when using a large pattern image with 320×320 pixels. The absolute alignment error is illustrated to be lower than 200nm within a large field of view of 1mm×1mm after the platform is calibrated using the proposed dynamic calibration method.

Date Published: 18 May 2009
PDF: 7 pages
Proc. SPIE 7284, 4th International Symposium on Advanced Optical Manufacturing and Testing Technologies: Design, Manufacturing, and Testing of Micro- and Nano-Optical Devices and Systems, 72840I (18 May 2009);

Published in SPIE Proceedings Vol. 7284:
4th International Symposium on Advanced Optical Manufacturing and Testing Technologies: Design, Manufacturing, and Testing of Micro- and Nano-Optical Devices and Systems
Sen Han; Masaomi Kameyama; Xiangang Luo, Editor(s)