The use of neighbourhood intensity comparisons, morphological gradients and Fourier analysis for automated precipitate counting & Pendellösung fringe analysis in X-ray topography

¹ Vision Systems Group, Research Institute for Networks & Communications Engineering (RINCE), Dublin City University, Dublin 9, Ireland
² Microelectronics Group, RINCE, Dublin City University, Dublin 9, Ireland
³ Optoelectronics Laboratory, Helsinki University of Technology, PO Box 3500, 02015 TKK, Espoo, Finland
⁴ Institute für Synchrotronstrahlung (ISS), Forschungszentrum Karlsruhe GmbH, Postfach 3640, D-76021 Karlsruhe, Germany

Corresponding author: patrick.mcnally@rince.ie

Received: 30 June 2003
Accepted: 31 December 2003
Published online: 15 July 2004

Abstract

Crystal distortions modify the propagation of X-rays in single crystal materials, and X-ray topography can be used to record these modifications on a film thus providing images of the distributions and nature of defects, dislocations, strains, precipitates, etc. in semiconductors. Small variations of contrast, which often need to be analysed can be rendered invisible. Furthermore, artefacts in the films must be removed. This study examines the use of advanced image analysis techniques applied to a selection of X-ray topographs in section transmission mode: (i) the automated counting of oxygen-related precipitates and (ii) the enhancement of Pendellösung fringes. The technique also succeeds in removing unwanted features in the original x-ray topographs such as vertical streaking due to collimating slit phase contrast and strain features near the surface due to the presence of integrated circuit process strains.