Microsc. Microanal. Microstruct. 2, 107-127 (1991)
DOI: 10.1051/mmm:0199100201010700

Multispectral linear filtering of high resolution EELS images by geostatistics

Pinnamaneni Bhanu Prasad^{1, 2}, Dominique Jeulin¹, Colin Daly¹, Claudie Mory³, Marcel Tencé³ et Christian Colliex<sup>3

1</sup>  Centre de Géostatistique, Ecole des Mines, 35 Rue Saint Honoré, 77305 Fontainebleau, France
²  LERMAT, ISMRa, Université de Caen, 14032 Caen Cédex, France
³  Laboratoire de Physique des Solides associé au CNRS, Bâtiment 510, Université Paris-Sud, 91405 Orsay Cedex, France

Abstract
Elemental mapping obtained from EELS spectra suffers from a relatively poor signal/noise ratio due to low counting statistics when looking for the ultimate spatial resolution of the instrument. This paper is devoted to the application of geostatistical techniques for filtering EELS images: multivariate optimal linear filters improve the S/N ratio by a factor 5 to 10 in the case of uranium and terbium bright field images, without apparent loss of spatial resolution. A systematic use of this method applied to EELS images would help progressing in the direction of single atom identification with this technique.

PACS
0250 - Probability theory, stochastic processes, and statistics.
0778 - Electron, positron, and ion microscopes; electron diffractometers.
3480 - Electron scattering.

Key words
Electron energy loss spectrum -- High-resolution methods -- Geostatistics -- Multispectral detection -- Linear filtering -- Data acquisition -- Data analysis -- Optimal filtering -- Kriging


© EDP Sciences 1991