IMI Interdisciplinary Mathematics InstituteCollege of Arts and Sciences

Applications of Multivariate Statistical Analysis for Large Spectrum-Image Dataset and High Resolution Images

  • March 3, 2009
  • 3:30 p.m.
  • Sumwalt 102

Abstract

Elemental mapping is especially suitable for analysis of nano-scale features in materials such as fine precipitates and interfaces/boundaries because two dimensional fluctuations in composition around such small features that may be missed by conventional point or line-scan analyses in analytical electron microscopy (AEM). Such elemental distributions can be obtained by a static beam with an electron energy-filter (EF) or by a scanning beam with an X-ray energy dispersive spectrometer (XEDS) and/or an electron energy-loss spectrometer (EELS). The broad-beam EF mapping approach is useful for obtaining maps from large fields of view with relatively short acquisition times. In contrast, elemental mapping via a scanning beam requires longer acquisitions, since elemental maps need to be recorded at individual pixels sequentially. The major drawback of elemental mapping is the degradation sensitivity, particularly in the scanning-beam method. Such poor sensitivity degrades the analysis of nano-scale features in AEMs.

The poor analytical sensitivity can be offset by applying spectrum imaging (which stores a whole spectrum at individual pixels in mapping) in combination with multivariate statistical analysis (MSA). These methods succeed because the major noise components can be removed efficiently from a single spectrum acquired for a very short dwell time and then regular spectral processing is potentially applicable. The MSA noise reduction is also applicable to atomic resolution images as well. In this talk, various applications of MSA to elemental mapping and atomic resolution imaging will be explored.

© Interdisciplinary Mathematics Institute | The University of South Carolina Board of Trustees | Webmaster
USC