Splicing dual-range EELS spectra: Identifying and correcting artefacts

In dual or multiple range electron energy loss spectroscopy, splicing the low loss spectra together with core loss ranges allows Fourier-log deconvolution of the entire energy range. However, because of the huge intrinsic dynamic range in EELS, the intensity at the splice point in a low loss spectrum is typically small, meaning that even weak or subtle artefacts can have big effects. Three main sources of artefacts in a Gatan GIF Quantum system have been investigated: non-uniformity of energy dispersion caused by aberrations in the spectrometer optics, stray scattering in the detector chamber, and small differences between the responsivity of the different detector quadrants. We present methods to measure, quantify and correct these artefacts. Ideally, the ratio for scaling at the splice should be the ratio of integration times. Prior to correction, the scaling factor is found to be about 15 % less than the exposure or time ratio and is dependent on the specimen thickness. After correction, the discrepancies are less than 0.5 %. This allows quantitative comparison of data taken at different points in time, even after major system changes, provided suitable artefact-correction datasets are taken. Whilst the detail is specific to one particular instrument, the principles are also applicable to newer spectrometers, including those with direct electron detectors.