EBSD · Electron channeling contrast imaging
ECCI: seeing dislocations in the SEM with electron channeling contrast
Start with picture 1: drag the beam tilt across the band edge and watch the backscattered signal drop, then press Go to the channeling condition.
1. Why the backscatter signal depends on the beam angle
Inside a crystal the beam electrons form standing waves. When the beam is steeper than the Bragg angle to a set of lattice planes, the waves sit on the atom planes and many electrons bounce back. Just past the Bragg angle they sit between the planes, run deeper, and fewer come back. The backscattered electron (BSE) yield changes by a few percent.
Settings
Model: two-beam Bloch waves for the band's two edges (+g and −g), each changing the yield by ε·w/√(1+w²) with ε = 2.5%, smoothed over the beam convergence. Extinction distances from atomic structure factors (see More detail); BSE coefficient η0 from Reuter's fit (Goldstein et al. 2018). The dislocation curve is the peak contrast of a dislocation 10 nm below the surface with g·b = 1, from the same model as picture 2.
Try it: ferritic steel, 20 kV: drag the tilt from 0 to 30 mrad. The yield falls by about 3% as you cross the band edge at 21 mrad. Press "Go to the channeling condition": the tilt lands just outside the edge, where the coral curve peaks. Raise α to 15 mrad: the edge smears and the dislocation contrast falls by about a quarter.
2. An ECCI image of a deformed grain, and the g·b = 0 rule
A dislocation bends the lattice planes near its core. On one side the planes tilt toward the Bragg angle, on the other side away from it, so the yield changes along a thin line. Set near the channeling condition, the grain is dark and the dislocations are bright. Choose which band edge you sit on (the reflection g). A dislocation disappears when g·b = 0, the same rule as in the TEM.
Settings
| Defect | b or R | g·b | Seen? |
|---|
b in units of a/2, R in units of a/3. Contrast: each dislocation bends the planes by d(g·R)/dz; the local deviation w + ξg·d(g·R)/dz sets the yield, weighted by its depth D as e−D/1.7ξg. Grey levels: the background follows the BSE yield on one fixed gamma curve, and the lines are shown with one fixed gain on top of it. A qualitative model, not a full dynamical simulation.
Try it: with g = 1 1̄ 1 the blue-marked family B1 has vanished (g·b = 0). Click 1 1 1̄: B1 comes back and B2 vanishes. Click 2 0 0: B3 and the subgrain boundary vanish and the subgrain grey level matches the grain. With 0 2 2̄ the stacking fault vanishes (g·R = 0) and B3 shows a wider image (g·b = 2). Now drag w to −1: the grain turns bright and the lines become faint dark lines.
3. How deep ECCI sees, and what limits it
Channeling contrast comes only from the top few extinction distances ξg: about 5ξg, which is tens of nanometres (Yamasaki et al. 2021). ξg grows with beam voltage, so a higher kV sees deeper. A damaged surface layer hides everything below it, and a wide beam cone smears the band edge.
Settings
Detector: a BSE detector on the pole piece (solid-state, annular or split in segments) with the sample flat, or a forescatter detector with the sample tilted about 70°. Damage depths: the silicon values and the other surfaces as listed, with sources, on the EBSD sample preparation page. The model treats the damaged layer as carrying no channeling contrast. RKO: Kanaya and Okayama (1972) range.
Try it: ferritic steel 110, 20 kV, electropolished: ECCI sees to about 72 nm. Move to 30 kV: about 87 nm. Pick "FIB, Ga 30 kV": the damaged layer takes away about a third of that depth and the contrast from just under it drops to 40%. Pick "1 µm diamond": nothing is left to see.
ECCI, TEM two-beam imaging and HR-EBSD side by side
| ECCI (SEM, BSE) | TEM two-beam | HR-EBSD (SEM, cross-correlation) | |
|---|---|---|---|
| What it sees | Single dislocations, stacking faults, twins, subgrain walls | Single dislocations and faults, with the finest detail (weak beam about 1.5 nm) | Lattice rotation and elastic strain; no single dislocations |
| Burgers vector | g·b = 0 tests, with g chosen from the channeling or EBSD pattern | g·b = 0 and g·b×u tests with many g; the standard method | Only the net (geometrically necessary) content from the curvature |
| Depth it looks into | The whole foil, 50 to 200 nm thick | The top 10 to 40 nm at 20 kV | |
| Area in one session | Many grains over mm² of bulk; one g per image | A few µm² per foil | Whole maps of mm², but a pattern per point (slow) |
| Sample | Bulk, flat, no damage: electropolish or long colloidal silica | Thin foil: twin-jet electropolish or FIB lift-out | Bulk, same surface as EBSD |
| Numbers it gives | Dislocation density by counting, in the top ~5ξg | Density, b, line direction, fault type | Rotations and strains near 10−4; GND density |
What to remember
More detail: the model, the numbers and its limits
Wavelength, Bragg angle and extinction distance
Relativistic wavelength for a beam voltage V (in volts):
At 20 kV λ = 8.59 pm, so θB = λ/2d is 21 mrad (1.2°) for the steel {110} planes. The extinction distance ξg = πVccosθB/(λFg) scales with the electron speed v, because Fg grows with the relativistic mass and λ falls with the momentum. The page uses ξg(V) = ξg(100 kV) × v(V)/v(100 kV). The 100 kV values come from the atomic scattering factors in our two-beam defect simulator and agree within 5% with tabulated values (Hirsch et al.): for example Al 111: 58 nm, Fe 110: 29 nm, Au 111: 18 nm.
| Material, band | d (nm) | ξg 100 kV | ξg 20 kV | 5ξg 20 kV |
|---|
The rocking curve
In the two-beam picture, a deviation s from the Bragg angle gives w = sξg, with s = Δθ/d. The Bloch wave with its maxima on the atom planes carries a share ½(1 − w/√(1+w²)) of the electrons. Backscattering happens at the atom cores, so the yield goes as
for the two edges of a band. ε = 2.5% sets the size; real channeling contrast is a few percent (Joy et al. 1982). This form (Reimer 1998, chapter 9; Wilkinson and Hirsch 1997) gives a bright band with a step at each edge. It leaves out higher orders and absorption, which add the dark lines seen just outside real bands. The convergence α is applied as an average over a disc of directions.
η0: η = −0.0254 + 0.016Z − 1.86×10−4Z² + 8.3×10−7Z³ (Reuter's fit, in Goldstein et al.), for example 0.28 for iron and 0.49 for gold.
Dislocation contrast
A dislocation displaces the lattice by R. Near the line the planes are tilted by d(g·R)/dz, which for a column at a distance x from the line is taken as (g·b / 2π) · x / (x² + c²), with c = ξg/4 standing in for the depth over which the channeling wave feels the bend. The local deviation is w + ξg d(g·R)/dz, and the column's yield is the rocking curve at that local deviation. This gives image widths of 10 to 20 nm, as in real ECCI images. A dislocation at depth D contributes with a weight e−D/Λ, Λ = 1.7ξg. That Λ makes the contrast of a dislocation at 5ξg about 5% of one near the surface, matching the measured visibility depth of 5 to 6ξg (Yamasaki et al. 2021). The model gives what Zaefferer and Elhami and Kriaa et al. report: at s slightly above 0 thin bright lines on a dark background, at s = 0 bright and dark lines side by side, inside the band (s < 0) weak contrast on a bright background.
Stacking fault: contrast ∝ sin(2π g·R) with fringes of period ξg/√(1+w²) along the depth. Subgrain: a rotation ω about axis a tilts plane g by ω a·(ĝ × k), which shifts w by ξg/d times that angle.
Numbers in picture 2
Surface normal k = [011]. The four reflections lie in the surface plane. B1: b = a/2[110], B2: a/2[1 1̄ 0], B3 and the boundary: a/2[0 1 1̄]. Fault R = a/3[111]. Boundary: ω = 0.3° about [100], so the dislocation spacing is b/ω = 49 nm. Contrast is shown as grey level after a fixed contrast stretch, as in a real ECCI image.
Limits
A simplified bending profile for every dislocation, no surface relaxation, no residual g·b×u contrast of edge dislocations, and only one beam plus g. Real images need dynamical Bloch-wave calculations (Kriaa et al. 2017, Zaefferer and Elhami 2014). The model is right on the trends: where the contrast is strongest, when defects vanish, and how depth and convergence reduce it.
On this site: the same g·b rule in the TEM with the two-beam defect image simulator and the g·b invisibility planner; Burgers vectors from scratch in dislocations and Burgers vectors; how bands form in Kikuchi bands to orientation; dislocation density from EBSD in KAM, GNDs and step size; making the surface in EBSD sample preparation; sharper TEM images in weak-beam dark field.
More in this chapter: TKD at the nanoscale · pattern quality · EBSD pattern simulator · misorientation and boundaries.
Questions people ask
What is electron channeling contrast imaging (ECCI)?
An SEM method that images crystal defects in a bulk sample with backscattered electrons. The crystal is tilted so the beam sits near a Bragg condition for one set of planes. The perfect crystal then gives a low, even signal, and the bent planes around a dislocation or fault give a thin bright line.
What is the channeling condition in ECCI?
The beam direction just outside the edge of a band in the channeling (or Kikuchi) pattern, with a small positive deviation s from the Bragg angle. There the background is dark and dislocations are thin and bright. You find it from an electron channeling pattern or from the EBSD orientation of the grain.
How deep can ECCI see dislocations?
About 5 to 6 extinction distances. That is roughly 70 to 90 nm in steel at 20 to 30 kV, and about 150 nm measured in a Ti alloy (Yamasaki et al. 2021). Lighter materials and higher voltages see deeper.
Is ECCI as good as TEM for dislocations?
For counting dislocations and seeing their arrangement over many grains, often yes, and with no thin foil. TEM still gives finer images (weak beam about 1.5 nm), sees the full foil thickness, and makes Burgers vector analysis with many reflections routine.
Does the g·b = 0 invisibility criterion work in ECCI?
Yes. A dislocation loses its contrast when its Burgers vector lies in the diffracting planes, g·b = 0, as in the TEM. Edge dislocations can keep a weak residual contrast from g·b×u. Two reflections with g·b = 0 give the direction of b.
How do you prepare a sample for ECCI?
The surface must be flat and free of deformation in the top few tens of nanometres. Electropolishing or a long colloidal silica vibratory polish are the usual last steps. Low-kV ion polishing also works. Diamond polishing alone leaves a deformed layer far thicker than ECCI can see through.
What voltage and detector are used for ECCI?
Usually 20 to 30 kV, a field emission gun, a beam current of a few hundred pA to a few nA, a short working distance, and a BSE detector on the pole piece with the sample flat. A small beam convergence keeps the band edge sharp.
What is the difference between an electron channeling pattern and an EBSD pattern?
Both show bands from the same lattice planes. An EBSD pattern maps the directions in which electrons leave the sample, recorded on a screen. A channeling pattern maps the BSE yield as the incoming beam direction is rocked. ECCI uses the channeling pattern to set the beam at a band edge.
References
Show the 11 references
- S. Zaefferer and N.-N. Elhami, Theory and application of electron channelling contrast imaging under controlled diffraction conditions, Acta Materialia 75, 20 to 50 (2014). doi:10.1016/j.actamat.2014.04.018
- D. C. Joy, D. E. Newbury and D. L. Davidson, Electron channeling patterns in the scanning electron microscope, Journal of Applied Physics 53, R81 (1982). doi:10.1063/1.331668
- A. J. Wilkinson and P. B. Hirsch, Electron diffraction based techniques in scanning electron microscopy of bulk materials, Micron 28, 279 to 308 (1997). doi:10.1016/S0968-4328(97)00032-2
- I. Gutierrez-Urrutia, S. Zaefferer and D. Raabe, Coupling of electron channeling with EBSD: toward the quantitative characterization of deformation structures in the SEM, JOM 65, 1229 to 1236 (2013). doi:10.1007/s11837-013-0678-0
- M. A. Crimp, Scanning electron microscopy imaging of dislocations in bulk materials, using electron channeling contrast, Microscopy Research and Technique 69, 374 to 381 (2006). doi:10.1002/jemt.20293
- H. Kriaa, A. Guitton and N. Maloufi, Fundamental and experimental aspects of diffraction for characterizing dislocations by electron channeling contrast imaging in scanning electron microscope, Scientific Reports 7, 9742 (2017). doi:10.1038/s41598-017-09756-3
- S. Yamasaki, M. Mitsuhara and H. Nakashima, Evaluation of depth of dislocation visibility in SEM electron channeling contrast imaging in Ti-6Al-4V alloy using serial sectioning method, Microscopy 70, 265 to 277 (2021). doi:10.1093/jmicro/dfaa060
- L. Reimer, Scanning Electron Microscopy: Physics of Image Formation and Microanalysis, 2nd ed., Springer (1998). Chapter 9, electron channeling and the two-beam rocking curve.
- J. I. Goldstein, D. E. Newbury, J. R. Michael, N. W. M. Ritchie, J. H. J. Scott and D. C. Joy, Scanning Electron Microscopy and X-Ray Microanalysis, 4th ed., Springer (2018). Backscatter coefficient (Reuter's fit) and the Kanaya-Okayama range.
- P. B. Hirsch, A. Howie, R. B. Nicholson, D. W. Pashley and M. J. Whelan, Electron Microscopy of Thin Crystals, Butterworths (1965). Extinction distances, the g·b = 0 criterion and the column approximation.
- A. J. Wilkinson and T. B. Britton, Strains, planes, and EBSD in materials science, Materials Today 15, 366 to 376 (2012). doi:10.1016/S1369-7021(12)70163-3. HR-EBSD sensitivity and GND densities.
BibTeX
@misc{tripathy2026ecci,
author = {Tripathy, Manisha},
title = {ECCI: Electron Channeling Contrast Imaging},
year = {2026},
howpublished = {\url{https://untetheredatom.com/ebsd/ecci-electron-channeling-contrast}},
note = {Interactive web tool}
}