1. X-ray spectra and images
Janez Košir

2. Contents The Ideal Spectrum Common Artifacts in XEDS Systems
Characteristic Peaks
The Continuum Bremsstrahlung Background
Common Artifacts in XEDS Systems
Escape Peaks
Internal Fluorescence Peaks
Sum Peaks
The Real Spectrum
Pre-specimen Effects
Post-specimen Effects
Coherent Bremsstrahlung

3. Contents Quality of the XEDS System Acquiring X-ray Spectra
Acquiring X-ray Images
Analog Dot Mapping
Digital Mapping
Spectrum Imaging
Post-Tagged Spectrometry

4. The ideal spectrum
X-ray spectrums consist of element-specific characteristics peaks with well-defined energies superimposed on a non- characteristic background.
They are displayed as graphs based on the X-rays energy and their intensity.
An electron beam generates two kinds of X-rays in the sample:
Characteristic X-rays
Bremsstrahlung background X-rays
Together, these X-rays comprise the X-ray spectra that we detect with the XEDS.
• Unfortunately, the spectrum generated from our sample, the spectrum detected by the XEDS system and the spectrum displayed on our computer screen are all quite different.
• In the image we can see a typical X-ray spectrum. We can clearly see element-specific peaks superimposed on a non-characteristic background.

5. Characteristic peaks
Characteristic X-rays are formed by ionizing an atom within the sample by ejecting its inner or core-shell electron. This creates a hole in the shell which can be filled with an electron from a higher shell thus bringing the atom back to its ground state.
When an electron transitions from a higher to a lower electron shell it looses energy in the form of an X-ray with a specific energy.
Each element creates X-rays with energies that are specific to that element. These energies are based on the shells from which the electron transitions to and from.
Because each element emits X-rays with energies specific to that element, we can determine the elements present in our sample based on the energies of these X-rays.
• Atoms in the sample are ionized by the interaction between the electrons from the electron beam and the electrons in an atom. When these electrons interact with each other, it typically results in an electron being ejected or “knocked out” from the atom, thus ionizing the atom.
• The energy of the X-rays corresponds to the energy difference between the higher and lower shell.
• Low-energy X-rays are typically more intense than higher-energy X-rays.
• Heavier elements have a more complex characteristic spectrum.
• It is not possible to detect X-rays with energies below 100eV as they are absorbed within the sample and the detector.
• In the image we can see how characteristic X-rays are formed through electron ejection and the atom coming back to its ground state.

6. The Continuum Bremsstrahlung Background
The X-ray spectrum background is bremsstrahlung (German for breaking) radiation arising as electrons from the electron beam are slowed down or stopped by electrostatic interactions with atomic nuclei within the sample.
The energy of the electron lost in this interaction is emitted as an X-ray without a specific energy.
The intensity of the bremsstrahlung is zero at beam energy (we can’t create more energy than the beam already has) and rises until it is infinite at zero energy.
• In the image we can see bremsstrahlung produced by a high-energy electron deflected in the electric field of an atomic nucleus.

7. Common artifacts in XEDS systems
XEDS systems introduce their own artifacts into the X-ray spectrum.
These can be separated in two groups:
Signal-detection artifacts (Escape peaks and Internal Fluorescence peaks)
Signal-processing artifacts (Sum peaks)

8. Escape peaks
Sometimes a small fraction of the incoming energy is lost within the XEDS detector and not transformed directly into electron hole pairs.
One way that this can happen is if an incoming photon, with an energy E, fluoresces a Si Kα X-ray, with an energy of 1,74 keV, which escapes from the intrinsic region of the detector.
The detector will register an apparent X-ray energy of (E - 1,74) keV instead of E.
Fortunately, escape peaks will typically amount up to only 2% of the major characteristic peaks.
• X-ray detectors are typically made from a semiconductor material (such as silicon) that work by detecting the energy of incoming photons and the amount of ionization they produce in the detector material, through electron hole pairs.
• An intrinsic region is a region that contributes to the detection of X-rays.
• The magnitude of the escape peak depends on the design of the detector and the energy of the fluorescing X-ray.
• Most XEDS software nowadays are capable of recognizing escape peaks and removing them.
• In the image we can see the escape peak in a spectrum from pure Cu. The escape peak is directly 1,74 keV below the Cu Kα peak. We can also see the difference in the peak intensities.

9. Internal fluorescence peaks
This characteristic peak comes from the Si in the detector dead layer.
Incoming photons can fluoresce atoms in the detector dead layer thus resulting in Si Kα X-rays entering the intrinsic region of the detector.
Because the detector can not distinguish the source of the energy a small peak is registered in the spectrum for Si Kα X- rays.
The intensity of the internal fluorescence peaks depend on the dead layer thickness and typically amounts up to 1 % of the major characteristic peaks.
• A detector dead layer is a layer above the surface of the detector, within which energy depositions do not result in detector signals.
• These peaks can pose a problem if we are trying to detect small amounts of Si within our sample.
• With the improvement of semiconductor detectors the thickness of the dead layer has decreased thus shrinking the internal fluorescence peaks.
• In the image we can see the Si internal fluorescence peak in a spectrum from pure C obtained with a Si detector. The ideal spectrum is fitted as a continuous line.

10. Sum peaks
Sum peaks occur when the input count rate exceeds the electronics ability to characterize all the incoming individual pulses thus creating a “pulse pile-up”.
Occasionally, two photons will enter the detector at exactly the same time. When this happens, the analyzer will register an energy corresponding to the sum of the two photons.
Sum peaks will typically first appear at twice the energy of the major peak.
Multiple sum peaks can be generated, causing problems in our analysis.
Sum peaks can be eliminated by lowering the input count rate.
• In the image we can see the spectrum of a pure Mg sample. The Mg K sum peak occurs at exactly twice the Mg Kα peak.

11. The real spectrum
In a perfect TEM all X-ray spectra would be characteristic only to the chosen analysis region of the sample that the beam interacts with.
In real X-ray spectrums two factors within the TEM contribute to introducing false information: Spurious X-rays and System X-rays.
These X-rays introduce small errors into qualitative and quantitative analysis.
The two factors that are responsible for these problems are:
High-voltage electrons, which generate intense doses of stray X-rays and scatter electrons in the illumination system
Thin samples, which scatter high-energy electrons and X-rays around the limited confines of the TEM stage.
Fortunately, these problems only introduce small peaks in the spectrum which typically amount up to only 1 % of the major characteristic peaks.

12. Pre-specimen Effects
The TEM electron beam produces high-energy electrons and bremsstrahlung X-rays scattered outside the main beam.
They are produced as a result of high-energy electrons interacting with column components, such as diaphragms and polepieces.
These electrons and X-rays strike the sample outside the desired analysis area thus producing spurious X-rays.
In inhomogeneous samples a significant presence of spurious X-rays means that the quantification process will give us a wrong result.
We can avoid spurious X-rays by using clean, thick, top-hat C2 diaphragms.
If the specimen is thinner than the pathway of fluorescence (a few nm), spurious X-rays will not be generated.
• Spurious X-rays are X-rays that come from the sample but are not generated by the electron probe in the chosen analysis region.
• Spurious X-rays can be avoided in several ways: * By regularly changing and cleaning the diaphragms.
* Using a very thick (several mm) Pt diaphragm with a top-hat shape and a slightly tempered bore to maintain good electron collimation.
* Using a small diaphragm just above the upper objective lens to shadow the thicker outer regions of the specimen from stray X-rays.
* By using a very thin and uniform sample.
* By using a Cs corrector in the probe-forming lens (very expensive solution).
• Stray electrons can be eliminated by using a non-beam defining spray diaphragm below the C2 diaphragm.

13. Post-specimen Effects
When electrons interact with the sample they are scattered elastically or inelastically.
Some electrons are scattered through high enough angles that they strike other parts of the sample, the support grid, the objective lens polepiece or any other part of the TEM.
This produces back-scattered electrons and system X-rays.
Some back-scattered electrons can travel directly into the XEDS detector while other may hit the sample away from the area of interest thus producing spurious X-rays.
Also, bremsstrahlung X-rays, produced in the specimen, can fluorescence characteristic X-rays from any material they strike since they possess a full spectrum of energy.
• Fortunately, in thin specimens, the scattering of these electrons is greatest in the forward direction. This means that most of the scattered electrons are gathered by the field of the lower objective polepiece and proceed into the imaging system, away from the XEDS detector.
• System X-rays are X-rays that come from parts of the TEM rather than the sample.
• Fluorescence of bremsstrahlung X-rays can be removed by using a uniformly thin foil (such as NiO) on a Cu grid.
• In the image we can see all possible sources of spurious and system X-rays from post-specimen scatter.

14. Post-specimen Effects
To determine the contribution of the TEM to the spectrum, a sample with a low atomic number is inserted into the microscope. In addition to the sample peaks, the spectrum will also show the various instrumental contributions.
We can reduce the effects of scattered radiation in several ways:
Keeping the sample at a zero tilt. This way the sample will have a minimum interaction with the Bremsstrahlung X-rays and backscattered electrons.
Surrounding the sample with a material that has a low atomic number (Z) such as beryllium (Be). This will also remove any characteristic peaks from the spectrum due to the microscope constituents.
• We can tilt the sample up to 10° as the background intensity will not be measurably increased.
• Ideally, all solid surfaces in the TEM stage region, that could be struck by scattered electrons, should be shielded with beryllium. Unfortunately these modifications are very rare.
• Peaks that come from elements that are part of the TEM itself cannot be observed within the sample if there is only a small amount present.
• In the image we can see an XEDS spectrum of boron, which also includes all peaks due to system X-rays.

15. Coherent Bremsstrahlung
In thin single-crystal samples, we can generate a bremsstrahlung X-ray spectrum with small peaks known as coherent bremsstrahlung (CB).
CB arises due to coulomb interactions of the beam electrons with the regularly spaced atomic nuclei in a crystal sample.
As the electrons proceed through the lattice, close to a row of atoms each bremsstrahlung-producing event is similar in nature and so the resulting radiation tends to have the same energy.
The intensity of the CB peaks is greatest when the beam is close to a low-index zone axis.
CB peaks will move depending on the acceleration voltage and sample orientation. This is not the case for characteristic peaks.
• Unfortunately, CB peaks can not be removed entirely.
• CB peaks are a problem only if you are trying to detect small amounts of specific elements in the sample.
• In the image we can see CB peaks in a spectrum for pure Cu.

16. Quality of the XEDS system
The quality of the XEDS system is determined by measuring the peak-to-background (P/B) ratio and the detector efficiency.
The PB ratio is based on the intensity of the characteristic peaks compared to the background. This ratio will increase with high energy peaks.
The detector efficiency is a measure of how many counts per second are collected, detected and processed by the XEDS system.
It will depend on the sample thickness, probe current and angle of collection by the detector.
• The detector efficiency will also decrease with increasing kV due to the decrease in ionization cross section.
• In the image we can see the P/B ratio in a Cr thin film.

17. Acquiring X-ray spectra
The standard way of gathering X-ray spectra is by using the spot mode.
The spot mode works by positioning the electron beam on a feature in the sample.
This is done by condensing the electron beam with the C2 lens and adjusting the C1 lens until the beam is small enough to interact only with the feature we want to analyze.
Spectrum-Line profiles are a variation of the spot mode where a series of spot analysis are taken linearly across a feature of interest.
This builds up a set of spectra which reveals the composition profile across the interface
Limitations of both spot and line profiles can be solved by gathering a full spectrum in every pixel of the TEM image, producing compositional images or maps.
• For spot mode to work, both the probe position and the feature in the image have to stay stationary long enough to gather a spectrum with sufficient counts.

18. Acquiring X-ray images
X-ray images can be acquired through different mapping methods:
Analog dot mapping
Digital mapping
Spectrum imaging
Post-tagged spectrometry
The biggest challenges in obtaining these images is that the sample will drift over time as well as changes in the beam current.

19. Analog dot mapping
Dot maps are the original method of acquiring qualitative X-ray images.
They work by selecting a specific (or a range of) energy in the X-ray spectrum, scanning the beam across an area of interest and when the XEDS registers an X-ray of the selected energy it records it and displays it as a dot in the image based on the location and intensity.
The changes in intensities reflect the changes in the number of X-rays detected.
These maps are not directly quantifiable since the background cannot be removed.
• Changes in the sample thickness will also produce changes in the image intensities.
• These maps can be refined by gathering multiple X-ray maps, assigning colors to different X-rays and overlaying the maps to give an indication of relative composition changes.
• In the image we can see a TEM image of a Pd particle (A), an analog dot map using the Pd L signal (B) and an early digital map using the Pd L signal with a removed background signal (C).

20. Digital mapping
Digital mapping works by collecting X-rays from multiple channels or windows thus acquiring several maps simultaneously.
If one or more of the maps contains the bremsstrahlung background intensity than quantitative maps can be produced.
Once a digital map of the sample is acquired, we can go back into the map and extract quantitative data from any region.
Digital mapping can reveal relatively small composition variations and also removes the problems with foil-thickness variations that analog dot mapping has.
• In the image we can see a quantitative digital map showing enrichment of Al at grain boundaries in an electro-migrated sample.

21. Spectrum imaging
Spectrum imaging (SI) is the preferred method for X-ray mapping.
It works by collecting a full spectrum at every pixel in the display image.
The result of the SI process is a 3D data cube containing electron images and XEDS spectra.
The best part of this method is that once a data cube is produced you can always recheck your data or do a completely different analysis.
• The biggest problem of this method is the time it takes to obtain a SI and the enormous amount of information we obtain, so it is important to search this database efficiently and extract as much meaningful data as possible.
• In the image we can see a schematic of the spectrum-image data cube (A) and a series of X-ray maps of a Ni super-alloy taken at specific energies from an SI data cube.

22. Post-tagged spectrometry
Post-tagged spectrometry (PTS) is a specific commercial version of SI which eliminates the conflict between the time it takes to obtain spectrum images while still having a full spectrum saved at each pixel.
In PTS, the beam is scanned rapidly across the area of interest while the X-rays are counted in the analysis computer, preserving both special and spectral information.
PTS also enables monitoring of phenomena such as specimen drift, contamination or damage during analysis.

23. Summary
X-ray spectrums help us determine the presence of specific elements within our sample.
Certain artifacts within the TEM produce false X-ray spectrums that make it harder for us to analyze our sample. These can come directly from the sample, the detection system or from the TEM itself.
The quality of the X-ray analysis is based on the peak-to-background ratio and detector efficiency.
We can construct maps of our sample using X-ray spectra based on the location of different elements.