Quantitative data processing of parallel recorded electron energy-loss spectra with low signal to background

TLDR

In order to achieve detection sensitivity limited only by counting statistics, the nonuniform spatial response of the array detector and potential noise added by the collection system must be critically considered. Several techniques were tested to mitigate the nonuniform response, including normalization with an experimentally measured response curve, electronic differentiation to reduce background, dynamic gain correction, and spatial averaging by repositioning spectra. The multielement array detector was shown to perform well for 100‑keV electron energy‑loss spectroscopy, exhibiting low added noise, a DQE of 0.87, and preserving the statistical information of the spectra.

Abstract

The performance of a multielement array detector for measuring small spectral features superimposed on a large background has been examined. In order to have the sensitivity of detection limited only by counting statistics, two factors have to be critically considered: the nonuniform spatial response of the array detector, and the possible addition of noise by the collection system. For the special case discussed here, energy-loss spectroscopy of 100-keV electrons, the added noise was small. The detective quantum efficiency (DQE) was measured to be DQE=0.87. Several techniques reducing the effect of the nonuniform response were tested: normalizing the spectra with an experimentally measured response curve, electronic differentiation to reduce the background, dynamic gain correction, and two methods of experimentally averaging the spatial response by measuring a sequence of repositioned spectra. Preservation of the statistical information present in a representative energy-loss spectrum is shown to be feasible.