Chair for Materials Science and Engineering

Advanced quantitative metallography – machine learning assisted detection of dendrites and 3D extension of the high resolution RB-EBSD method. The research of Alexander Richter aims at improving the quantitative metallographic characterization of dendrites in dast Ni-base superalloys. The work has two parts. In the first part, machine learning algorithms are applied to identify dendrites in large montages which were obtained in an automatic procedure using optical microsopy. In the second part, Pascal Thome’s high resolution RVB EBSD method (angular resolution > 0.1°) is extended to provide 3D information. In the experimental part of this work closely spaced (e.g. 0.2 mm) parallel cross sections are produced perpendicular to the growth direction of dendrites. These are then being analyzed using our RVB-EBSD method. The crystallographic information retrieved from all cross section is then consolidated in order to obtain a spatial distribution of crystallographic orientations in the cast microstructure. In combination with solidification experiments, this will allow to a deeper insight into the growth mechanisms of dendrites.