Caractérisation automatique des structures de dislocations par MEB

In order to predict the behavior of the materials, it is important to make an effort for optimizing the precision in measuring values which characterize the material microstructure. A parameter of metals is known to give many problems in that sense, since difficulties in its measurement are found and its value is many times measured with a poor precision. The concerned micro-structural propriety is the dislocation density. Main techniques for measuring the dislocation density are the Transmission electron microscopy (TEM) and the Electron Channeling Contrast Imaging (ECCI), thanks who a direct visualization of dislocations can be achieved. However, these methods require a lot of time and a not negligible background in crystallography since dislocations are visible just in specific orientations (Two Beam conditions). Moreover, a significant fraction of dislocations might be invisible under certain TB conditions, leading to an error in their measurement. Other global techniques, which measures dislocation density with an indirect way, are present. However, they behave large uncertainties linked both to the mathematical model of the indirect measurement and to assumption that have to be made on the nature of dislocations. This master’s thesis aims to develop a new method for the characterization of dislocations for a meaningful number of grains and without any consideration of grain orientations prior to the observations. This method is expected to characterize the nature of defects and to give quantitative information about their density in briefer times of those reached with TEM or ECCI characterization. It will be tried to reach this goal by melting together two existing techniques: ECCI and eCHORD, an analytical method developed by MATEIS microscopy researchers (from INSA Lyon) within the last few years. Electrons CHanneling ORientation Determination (eCHORD) uses a set of backscattered electrons images, obtained with different angles between sample and electrons beam, to automatically map the crystallographic orientation of grains in a polycrystal. In short, the contrast in these images changes with the angle and an intensity profile is obtained for each pixel belonging to a characteristic zone of the sample; this profile is effectively a signature of the orientation of the grain. The interesting point is that, while performing an eCHORD observation, each grain execute a 360° rotation, therefore, for each grain, different two beam (TB) conditions will be found. This will be used for visualizing dislocations without a previous study of the orientation and avoiding the problem of their invisibility under certain TB conditions. It will be tried to analyze these stack of images with a Clustering program in order to obtain exploitable data for studying both the nature and the density of dislocations. In short, thanks to an analysis of rotational profiles, the program allows the discrimination of zones with different behaviors such as different grains, different phases and crystallographic defects, letting to distinguish matrix pixels from dislocations pixels. Hence, in optimal conditions it should be possible to obtain useful data and to determinate a value of dislocation density or to individuate invisibility criterions for detected dislocations. A close examination of experimental parameters and a development of tools for data processing will be done. The resulting technique takes the name of disCHORD.