Er(D,T){sub 2-x} {sup 3}He{sub x}, erbium di-tritide, films of thicknesses 500 nm, 400 nm, 300 nm, 200 nm, and 100 nm were grown and analyzed by Transmission Electron Microscopy, X-Ray Diffraction, and Ion Beam Analysis to determine variations in film microstructure as a function of film thickness and age, due to the time-dependent build-up of {sup 3}He in the film from the radioactive decay of tritium. Several interesting features were observed: One, the amount of helium released as a function of film thickness is relatively constant. This suggests that the helium is being released only from the near surface region and that the helium is not diffusing to the surface from the bulk of the film. Two, lenticular helium bubbles are observed as a result of the radioactive decay of tritium into {sup 3}He. These bubbles grow along the [111] crystallographic direction. Three, a helium bubble free zone, or 'denuded zone' is observed near the surface. The size of this region is independent of film thickness. Four, an analysis of secondary diffraction spots in the Transmission Electron Microscopy study indicate that small erbium oxide precipitates, 5-10 nm in size, exist throughout the film. Further, all of the films had large erbium oxide inclusions, in many cases these inclusions span the depth of the film.
Obtaining particulate compositional maps from scanned PIXE (proton-induced X-ray emission) measurements is extremely difficult due to the complexity of analyzing spectroscopic data collected with low signal-to-noise at each scan point (pixel). Multivariate spectral analysis has the potential to analyze such data sets by reducing the PIXE data to a limited number of physically realizable and easily interpretable components (that include both spectral and image information). We have adapted the AXSIA (automated expert spectral image analysis) program, originally developed by Sandia National Laboratories to quantify electron-excited X-ray spectroscopy data, for this purpose. Samples consisting of particulates with known compositions and sizes were loaded onto Mylar and paper filter substrates and analyzed by scanned micro-PIXE. The data sets were processed by AXSIA and the associated principal component spectral data were quantified by converting the weighting images into concentration maps. The results indicate automated, nonbiased, multivariate statistical analysis is useful for converting very large amounts of data into a smaller, more manageable number of compositional components needed for locating individual particles-of-interest on large area collection media.
Spectrum imaging combined with multivariate statistics is an approach to microanalysis that makes the maximum use of the large amount of data potentially collected in forensics analysis. Here, this study examines the efficacy of using spectrum imaging-enabled microscopies to identify chemical signatures in simulated bioagent materials. This approach allowed for the ready discrimination between all samples in the test. In particular, the spectrum imaging approach allowed for the identification of particles with trace elements that would have been missed with a more traditional approach to forensic microanalysis. Finally, the importance of combining signals from multiple length scales and analytical sensitivities is discussed.
