Energy dispersive X-ray spectroscopy principle
When the primary electron beam of an SEM strikes a sample, some electrons knock out an inner-shell electron from an atom. An electron from a higher shell drops into the vacancy, and the energy difference can be released as an X-ray photon. Because shell energies are fixed for each element, these characteristic X-rays act as a fingerprint: iron, silicon and oxygen each produce lines at their own energies.
Lines are named by the shell that was ionised. K lines come from K-shell vacancies, L and M lines from outer shells, so a heavy element such as lead shows L and M lines in the usual energy range, while light elements show only K lines. Alongside the peaks, electrons slowing down in the sample produce a continuous background called bremsstrahlung, which sets a floor for detecting small amounts.
If you are new to how the beam interacts with the sample and which signals form SEM images, start with our overview of how a scanning electron microscope works.
The EDS detector: why silicon drift detectors took over
Each X-ray absorbed in the detector creates a number of electron-hole pairs proportional to its energy. The electronics measure that charge, sort the event by energy and build a spectrum: counts on the vertical axis, energy in keV on the horizontal axis.
Modern systems use a silicon drift detector (SDD). According to JEOL, an SDD routinely handles count rates above 100,000 counts per second, offers energy resolution of about 129-133 eV at Mn Kα, and is cooled by a Peltier element rather than liquid nitrogen, which older Si(Li) detectors needed. At accelerating voltages below 5 kV, SDDs make it practical to detect light elements such as boron, carbon and nitrogen. Energy resolution is quoted as the width of the Mn Kα peak (FWHM), the convention defined in ISO 15632; systems that claim detection below 1 keV must also state resolution at C Kα and F Kα, and every figure must name the count rate it applies to.
| Detector property | Why it matters |
|---|---|
| Energy resolution (eV at Mn Kα) | Narrower peaks separate neighbouring lines and improve detection of minor elements |
| Solid angle and take-off angle | Determine how many X-rays are collected; larger solid angle helps at low beam current |
| Maximum count rate | Allows fast mapping without excessive dead time |
| Window type | Affects transmission of low-energy X-rays and light-element sensitivity |
Energy dispersive X-ray spectroscopy analysis modes
EDS software offers three main ways to collect data, and choosing the right one saves time.
- Point (spot) analysis: the beam is held on one spot or a small area and a full spectrum is collected. Best for identifying an inclusion, particle or phase.
- Line scan: the beam steps along a line and element signals are plotted against distance. Useful for diffusion profiles, coatings and interfaces.
- Elemental mapping: a spectrum is recorded at every pixel to build colour maps of each element. Ideal for showing phase distribution, segregation or contamination.
Maps based on simple energy windows can show an apparent signal that really comes from background or a neighbouring peak. Deconvolved or quantitative maps are slower to process but more reliable when peaks overlap.
How to run a reliable EDS analysis in the SEM
- Prepare the right sample. For quantitative work the surface should be flat and polished, homogeneous and thick compared with the beam interaction volume. A carbon coat is usually preferred for non-conductive samples, since a metal coat adds its own peaks; see our guide to SEM sample preparation.
- Choose the accelerating voltage. JEOL recommends a beam energy 1.5 to 2 times the energy of the highest X-ray line of interest, and 15-20 kV for an unknown sample.
- Set the beam current and working distance. Use enough current for a statistically significant count rate and the working distance specified for the detector geometry.
- Check the line of sight. Only X-rays that reach the detector are counted, so rough topography can shadow the signal. Rotate or tilt the sample so the region of interest faces the detector.
- Identify every peak. Confirm each element by its full family of lines (for example Kα and Kβ, or the L series), not by a single peak and an automatic label.
- Quantify with care. Average several areas on inhomogeneous samples and note that results normalised to 100% can hide missing elements or errors.
Key tip: an automatic peak label is a suggestion, not a result. Always check that the suggested element shows all of its expected lines at the right relative heights.
Detection limits, accuracy and spatial resolution
EDS is a fast, broad technique rather than a trace-element method. JEOL gives typical detection limits of about 0.1 wt% (1000 ppm) for heavier elements and about 1 wt% for light elements from fluorine down to beryllium, with elements from Be to U detectable. For major elements in a flat, polished, homogeneous sample, standardless analysis can reach about ±2-5% relative, but errors grow significantly for particles and rough surfaces. Standards-based quantification, where the sample is compared with measured reference materials under the same conditions, is more accurate than standardless routines.
Spatial resolution is limited by the interaction volume, not the beam diameter. In bulk samples at typical SEM voltages, X-rays are generated from a volume of the order of a micrometre, so a feature smaller than that will include signal from its surroundings. Lowering the voltage shrinks the volume, and thin samples in a TEM or STEM give much finer EDS resolution.
| Feature | EDS (EDX) | WDS |
|---|---|---|
| Speed | All elements at once, seconds to minutes | One element per spectrometer at a time, slower |
| Energy resolution | About 120-130 eV class | Much finer, separates most overlaps |
| Detection limits | About 0.1 wt% for most elements | Lower, suited to minor and trace elements |
| Typical use | Phase identification, mapping, routine analysis | High-accuracy quantification, overlaps, trace elements |
Common EDS pitfalls and peak overlaps
Most mistakes in EDS analysis in SEM come from a handful of predictable causes:
- Peak overlaps: S Kα, Mo Lα and Pb Mα all sit near 2.3 keV, and Ba L lines overlap Ti K lines near 4.5 keV. Higher-energy lines or WDS help resolve them.
- Artefact peaks: sum peaks at high count rates and escape peaks from the silicon detector can be mislabelled as real elements.
- Coating and substrate signal: carbon tape, coating and the stub can appear in spectra of small or thin features.
- Low overvoltage: if the beam energy is too low for a line, that element will be under-reported or missed.
- Beam damage and charging: mobile elements such as sodium can migrate under the beam in glasses and minerals, and charging shifts the beam position during mapping.
Energy dispersive X-ray spectroscopy applications
EDS is used wherever composition at the micrometre scale matters: identifying inclusions and corrosion products in metals, checking solder joints and contamination in electronics, mapping minerals in geology, analysing particles in environmental and forensic samples, and confirming the composition of nanoparticles and coatings in materials research. In life sciences it is used to locate heavy-metal stains, mineral deposits and labels. A common workflow is to survey a sample first with optical microscopy, as described in our guide to compound light microscope parts, then move to the SEM for imaging and EDS on selected areas.
Conclusion
Energy dispersive X-ray spectroscopy is the quickest way to answer "what elements are here?" in an electron microscope. Silicon drift detectors made it fast enough for routine mapping, but good results still depend on a suitable sample, the right accelerating voltage, careful peak identification and realistic expectations of detection limits and spatial resolution.