Extracting information from sequences of spatially resolved EELS spectra using multivariate statistical analysis

The sophisticated acquisition procedures now available in spatially or time resolved spectroscopies produce large amounts of data which require effici ent automatic data processing procedures. In an electron energy loss spectr oscopy (EELS) line-spectrum sequence, all individual spectra can be process ed independently. However, it is better to consider the data set as a whole and to process it as such. Multivariate statistical analysis (MSA) can be used to identify several basic sources of information as contributing throu gh a linear combination to the overall experimental data set. In this paper we show that MSA can be applied to spatially resolved spectroscopy, but of ten necessitates that extension from orthogonal to oblique analysis be perf ormed. The aim is then to identify fine structures characteristic of EELS e dges and to extract the signal specific of the interface. In the first sect ion we present the overall methodology. Then it is illustrated through a si mulation. Finally, we apply the method to the study of Si-SiO2 and SiO2-TiO 2 interfaces. We show that MSA constitutes a useful tool to estimate compon ents present at the interface and the corresponding concentration profile o f these components. (C) 1999 Elsevier Science B.V. All rights reserved.