1. Low-Loss and No-Loss Spectra
Jani-Petteri Jylhä

2. Scattering
Spectrum contains both inelastically and elastically scattered electrons
Zero-loss peak contains primarily elastic electrons scattered forward
Also some electrons, that have suffered small energy losses
Forming images and DPs with zero-loss electrons have advantages especially with thicker specimen
Low-loss (up to ~50 eV) region contains electrons, that have interacted with outer shell electrons of atoms
Low-loss region can be used to reveal information about electronic structure of specimen

3. Spectrum
Most understandable and more easily modelled at single scattering
This requires very thin specimen and high acceleration voltages
Usually sample is thicker causing plural-scattering
Deconvolution routines is used to remove plural-scattering effects
Deconvolution involves risk of introducing artifacts into part of the spectrum, where it is used

4. Typical energy losses
EELS is used to study energy losses of principal scattering processes
These consist of single-electron scattering
Interband and intraband transitions at 2-20 eV
Plasmon interactions at 5-30 eV
Inner-shell ionizations at eV
Phonon excitation also happens, but losses too small to be separated from ZLP
Typical energy angles can be estimated as a function of energy loss as

5. Zero-loss peak (ZLP)
Predominant feature of spectrum only if specimen is thin enough for EELS
Consists primarily of electrons that have retained incindent-beam energy
ZLP actually has electrons with energy loss below resolution limit
Mostly those that excited phonons
Phonon-scattered electron don’t have useful information but heat up the specimen
ZLP can cause saturation of detector causing ghost peak in spectroscopy
ZLP should be deflected from detector if it is not needed
In microscopy ZLP is very useful when forming image or DP where most of energy-loss electrons have been excluded
However, also forming images with ZLP filtered out and using selected energy-loss electrons is useful

6. Zero-loss images and DP
If all energy-loss electrons are filtered, elastic image or DP is formed
This removes chromatic-aberration effects and increases resolution of thick specimen and enhancing contrast
Filtering has positive effect on all forms of TEM-image contrast, but especially for thick biological or polymeric specimen
Inelastic scattering is stronger than elastic scattering in these
Also diffraction contrast can be enhanced since inelastically scattered broaden the excitation error reducing diffraction contrast
Also contrast tuning can be used
Energy-loss window is selected and tuned to find best contrast
Useful for both low- and high-loss regions of spectrum

7. Low-loss spectrum After ZLP, next major peak is caused by plasmon
Spectrum is relatively featureless apart from plasmon peak
Spectrum still contains useful data containing lot of counts
Beam electrons have interacted with condution and/or valence bands
Low-loss spectrum can be used to fingerprint materials
Dielectric constant of specimen can be determined from low-loss spectrum
Great interest of semiconductor industry
For best low-loss spectroscopy FEG is required with high resolution and dispersion spectrometer and monochromator

8. Plasmons
Plasmons are longitudinal wave-like oscillations occurring when a beam electron interacts with electrons in the conductance or valence band
Plasmons have lifetime of about s
Plasmons dominate in materials with free-electron structure
Specimen is too thick for EELS if first plasmon peak is more than 1/10 of the zero-loss intensity
Correlations between plasmon energy and elastic properties, hardness, valence-electron density, and cohesive energy exists
Interest for this part of spectrum is rising for studing nanoscale materials

9. Single-electron excitations
High energy beam electron can transfer enough energy to single electron in valence band to change its orbital state
These single-electron interactions result in inter- and intraband transitions for valence electrons
Energy losses for these go up to ~25 eV
Intensity variation can be used to fingerprint particular phase
Interactions with molecular orbitals produce characteristic peaks in spectrum, which may cause shifts in the plasmon peak causing intensity variations
Valence band electron may be given sufficient energy to escape the nucleus making secondary electron (SE)
These are used to form topographic images in SEM and STEM
Typically SE requires < 20 eV to escape

10. The band gap
If region between ZLP and intensity preceding plasmon peak has no interband transitions, forbitten-transition region can be seen
This is band gap between valence and conduction band in semiconductors and insulators
Need for sub-nanometer resolution imaging of the band gap will increase
This is caused by advancing of sub-nanometer-scale semiconductor technology
Low-loss EEL images are only way of visualizing this electronic property