Spectroscopy and Raman imaging of inhomogeneous materials

This thesis is aimed at developing methodologies in Raman spectroscopy and imaging. After reviewing the statistical instruments which allow treating giant amount of data (multivariate analysis and classification), the study is applied to two families of well-known materials which are used as models for testing the limits of the implemented developments. The first family is a series of carbon materials pyrolyzed at various temperatures and exhibiting inhomogeneities at a nm scale which is suitable for Raman-X-ray diffraction combination. Another results concern the polishing effect on carbon structure. Since it is found to induce Raman artifacts leading to the overestimation of the local structural disorder, a method based on the use of the G band width is therefore proposed in order to evaluate the crystallite size in both unpolished and polished nano-graphites. The second class of materials presents inhomogeneities at higher (micrometric) scales by the example of uranium dioxide ceramics. Being well adapted in terms of spatial scale, Raman imaging is thus used for probing their surfaces. Data processing is implemented via an approach combining the multivariate (principal component) analysis and the classical fitting procedure with Lorentzian profiles. The interpretation of results is supported via electron backscattering diffraction (EBSD) analysis which enables us to distinguish the orientation effects of ceramic grains from other underlying contributions. The last ones are mainly localized at the grain boundaries, that is testified by the appearance of a specific Raman mode. Their origin seems to be caused by stoichiometric oxygen variations or impurities, as well as strain inhomogeneities. The perspectives of this work include both the implementation of other mathematical methods and in-depth analysis of UO₂ structure damaged by irradiation (anisotropic effects, role of grain boundaries). (author)