Improvements in the characterization of polycrystalline thin films: Microchemistry, microtexture, and microstructure

Materials properties often depend critically on microstructure, especially in polycrystalline thin films. Improvements in characterization techniques are necessary to improve our understanding of structure-property relationships. In this work, we develop improvements for the quantitative measurement of grain boundary segregation, crystallographic texture, and the grain size distribution.; Low-magnification X-ray mapping has been developed in the analytical electron microscope (AEM) to allow large numbers of grain boundaries to be probed in parallel rather than in series as is conventionally done. Segregation measurements on an Al-4 wt.% Cu thin film metallization material show wide variation in the magnitude of Cu segregation, which may have implications for its electromigration (EM) reliability. This wide variation implies that typical measurements from ∼10 grain boundaries may be insufficient to properly characterize a material.; Electron backscatter diffraction (EBSD) has been used to characterize the texture of Al and Cu metallization materials, and quantitative comparison to X-ray diffraction (XRD) shows good agreement. The data processing applied to the raw EBSD data has a substantial effect on the result, and typical analyses have particular problems when applied to strongly-textured thin film materials. Improved analysis software has been developed which overcomes several conventional limitations, resulting in improved sensitivity to minor texture components compared to XRD and conventional EBSD analysis. A total of ∼1 vol% of unexpected (100)- and (221)-textured grains were observed in Al films with improved processing, while neither conventional EBSD nor XRD detected them. This may have implications for the EM reliability of these materials.; Automated microstructural analysis software has been developed and applied to quantify the grain size distribution of over 8000 grains from transmission electron microscope (TEM) images of an Al thin film. This involved improvements to both the TEM image acquisition and the subsequent data analysis, including the development of several novel image processing algorithms. Automated analysis has been shown to give results consistent with manual analysis, but with the added advantage of consistency. Improved statistics (compared to typical populations of ∼100 grains) are necessary for better quantification of a material's behavior, especially in the tails of the distribution which are often important for thin film properties such as EM resistance.