Multiscale Methods for the Dimensional Characterization of Surfaces and Interfaces

Autor: Yibo Zou

ISBN: 978-3-95900-133-5

Dissertation, Leibniz Universität Hannover, 2017

Herausgeber der Reihe: Eduard Reithmeier

Band-Nr.: imr 01/2017

Umfang: 166 Seiten, 71 Abbildungen

Schlagworte: thermal spray coatings, multiscale methods, surface measurement, surface characterization, data stitching

Kurzfassung: In this work, multiscale methods are developed for measuring and characterizing the typical surface textures (porous surfaces and stochastic roughness features) of thermal spray coatings in the field of 3D optical metrology. The purpose is to allow a more efficient, fast and reliable surface measurement and surface characterization. Different objective lenses of a confocal laser scanning microscope and a white light interferometer are used for data acquisition. In order to conduct a more robust data processing, an improved stitching method is developed. The developed algorithm takes additional tilt errors into account and exhibits good applicability in stitching typical engineering surfaces.
For the analysis of porous surfaces, the measurement and characterization are conducted by using different objective lenses. Firstly, different lenses are assigned with different functions: lenses with low magnifications are used to investigate 
the distribution of the pores, while lenses with higher magnifications are utilized to measure the 3D geometry of the porous microstructures. Afterwards, the influences of different magnifications on the pores’ 3D parameters under different resolutions are studied. The results show that some parameters are variant, while other parameters are invariant to the applied lenses, depending on the physical properties of the lenses, the measurement principles of microscopes as well as the concrete type of the parameters.
For the analysis of stochastic roughness features, fractal methods combined with conventional roughness characterization are applied. The surface measurement is also conducted by multiscale methods. Both box-counting method and power spectrum method are implemented for the numerical calculation of the fractal dimension. The results show that the investigated samples exhibit isotropic textures and the estimated fractal dimension from multiscale images holds the scale-invariance property. Moreover, no correlation has been found between conventional roughness parameters and the fractal dimension.