Passing a thin specimen, primary electrons can loose energy when they collide with electrons from the atoms of the material. With an energy filter these inelastically scattered electrons can be dispersed as a function of their energy and then detected using either a linear diode array or a CCD camera. As the excited electron will take some kinetic energy with it (which the primary electron must supply) one gets ionization edges on top of an approximately exponentially decaying background, rather than discrete peaks in x-ray microanalysis. This means that the background in front of the edges needs to be extrapolated and subtracted before the net signals can be evaluated. Because of the improved energy resolution of a typical electron spectrometer (<1eV) compared to an x-ray spectrometer (>100eV) electron energy-loss spectra (EELS) often exhibit details that allow not only the chemical composition to be determined, but also provides information on the nature of the chemical bonding, in particular the bond symmetry and the oxidation state of transition metal ions. This information can be gathered with sub-nm spatial resolution if a focussed electron beam is used, which makes EELS ideally suited for a chemical analysis of defects and interfaces. In our group we focus on a quantitative study of planar defects and diffusion phenomena at interfaces.

As an example of our work, a 6nm thin oxide layer on top of a 65nm germanium silicon layer is shown. The figure depicts

(a) a bright field overview image with the entrance aperture to the spectrometer
(b)-(h) energy loss (horizontally) as a function of position in real space (vertical)