Quantitative image analysis of superconductor thin film microstructure: The use of conditional, multiparametric, shape-analysis algorithms

Abstract

We have developed a novel approach for quantifying the microstructure of granular thin films using digital image processing and analysis. In the past, conventional scanning electron microscopy of thin films has generated qualitative information on the surface topography and film microstructure. However, when coupled to digital image analysis, the amount or degree of surface contours (i.e., granularity) in SEM micrographs can be quantified in a rapid and reproducible manner. Briefly, SEM micrographs are digitized and the edge boundaries on the film surface are enhanced by a gradient filter; granularity is then quantified by calculating the %AREA covered by the edges with respect to the entire field. Objects of a particular shape, such as phase impurity particles, can be selectively deleted from the image using a specific sequence of shape analysis algorithms and parameter values. In this manner, the contributions of edges from the phase impurity particles is minimized in the final measurement of real surface contours. Statistical analysis of the data yields quantitative information concerning variations in microdomains within single thin films and can detect statistically significant differences among samples. This method is being used in the characterization of the microstructure of superconducting thin films for optimization of their electrical and magnetic properties.