A new 2D-self-calibration method with large freedom and high-precision performance for imaging metrology devices

When calibrating 2D (or 3D) metrology systems you need to rely on a traceable artefact for the calibration. However if the system you intend to calibrate has smaller uncertainties than the uncertainty of the reference artefact, the uncertainty of the instrument will be dominated by the artefact and not by the instrument. The only way to reveal the performance of the instrument is then to use self-calibration, i.e. a calibration without any externally verified references, except a 1D traceable measurement between two points on an artefact. Already in 1997, Mikael Raugh developed the rigorous mathematics for self-calibration of a 2D metrology stage, based on a lattice structured artefact. The original method and subsequent later improvements have in common that the problem is solved by using some assumptions regarding the artefact used in the calibration; like that the locations of the marks in the lattice are approximately known. There are also other constrains in the mathematical solution that limits its practical use in the industry. In this paper the application of a new general self-calibration algorithm is presented giving a large freedom to the positioning of the artefact, and also less demands on the 2D-structure on it. Rather than being based on rigorous mathematics requiring very exact positioning of the artefact, our algorithm is using a numerical iterative technique to minimize all overall errors. The algorithm is an enhancement of the self-calibration method already published by P. Ekberg et al. The algorithm has successfully been tested by simulations and by using real data from a white light interference microscope, yielding X, Y precision of few nm. The algorithm has also been used for separating distortions in ordinary low cost camera based systems opening up possibilities for accurate measurements in images. In the latter case the images can be compensated for most errors, like barrel or pin-cushion distortions, as well as perspective effects due to the angle of the camera relative the object.