Practical Electron Microscopy and Database

The nominal energy dispersion in EELS system (e.g. shown on Gatan Digital Micrograph window) may not be accurate. For instance, 2 %  error is not unusual for many EELS systems. Assuming 0.2 eV/channel of nominal energy dispersion is used, the actual energy dispersion can be, for example, 0.196 eV/channel. Therefore, the maximum inaccuracy of 8 eV exists if there are totally 2000 channels in the overall EELS profile of 400 eV. In many cases, accurate analyses of core losses, e.g. for valence states, are based on calibration of energy dispersion, which can be done, e.g. for mineral scientists and engineers using Fe L_2,3 and O K edges from hematite. The position of the Ti L₂ edge can also be used as an energy calibration. A 4 nm thickness of pure Ti film would be sufficient for this calibration without predominant plural scattering.

Furthermore, it is very common that an extra calibration step of the peaks of the other elements (e.g. Fe and Ti edges mentioned above) is performed against π* peak of K edge of amorphous carbon (C) because the C π* peak is accurately locked at 285 eV.