Atomic AEM - poissonian problems from gaussian probes!

Abstract

In a previous paper, the authors have described the engineering requirements needed to detect one atom in the analytical electron microscope (AEM) by using x-ray microanalyis. Whilst the requirements to achieve this goal cannot be specified at present for a particular instrument, the specification for machines being developed by Vacuum Generators have a calculated minimum detection limit (MDL) of fewer than 4 atoms. At these detection limits the usual Gaussian statistics which have applied in AEM give way to Poissonian statistics. This paper will look at some of the interesting consequences of AEM at the atomic level.The energy dispersive x-ray spectrometers (EDS) used in AEM have percentage detection limits usually quoted as about 0.1 wt. %. For this to equal 1 atom as the MDL, the analysed volume, defined by the probe diameter and the specimen thickness, must contain about 1000 atoms. For a field emission gun (FEG) on an AEM, sufficient current (1nA) can be obtained in a small enough probe (1nm FWHM) to allow analysis from a volume containing 1000 atoms (assuming adequate x-ray detection sensitivity) if the sample is about 20 atoms thick.