EBSD misindexing: when the software gives a confident wrong answer

What does a misindexed EBSD map look like?

In real life: speckle, rims and empty points are easy to see and usually have a low CI. A checkerboard inside grains is the usual sign of pseudosymmetry. A grain that is wrong everywhere (for example after dictionary indexing with a wrong pattern centre) has no checkerboard, so it is the hardest to spot.

Go deeper: what cleanup does to these points · IQ, CI and MAD in one place

What is pseudosymmetry, and why does it fool the software?

• Pseudosymmetry. A crystal that is almost, but not quite, more symmetric. Its pattern looks nearly the same after a turn that is not a real symmetry of the crystal, so two or three answers fit almost equally well.
• Variants. The near-equal answers. In a nearly cubic tetragonal crystal (γ-TiAl, tetragonal ZrO₂) there are three: the c axis along any of the three cube edges.

In real life: Hough indexing uses only band positions, so it cannot separate variants whose bands sit within its measuring error. Methods that compare the whole pattern (dictionary indexing, refinement, pattern matching) also see the weak bands and the band intensities that differ between variants. For γ-TiAl, matching to simulated patterns got 96% of points right against about 70% for Hough [5].

Go deeper: the four indexing methods on a γ-TiAl map (case 6)

Which phases are known to be misindexed by EBSD?

Each card is backed by the papers or vendor notes listed under it. Filter by material or by kind of problem.

Left out: phases where we could not check a primary source, for example hcp titanium, magnesium and zirconium (a 30° turn about [0001] is sometimes mentioned), calcite, silicon and sigma phase.

What else makes EBSD give a wrong answer?

How do I catch misindexing? The misorientation histogram

In real life: a spike at one angle and axis that the material should not have is the usual first sign. The trap is that the wrong turn can equal a real twin or domain law: the quartz Dauphiné twin, γ-TiAl order domains, the ferrite Σ3. Pseudosymmetry cleanup (OIM, AZtecCrystal) merges boundaries of a given turn; a cap on the patch size keeps real twins, and a check against optical images or EDS shows whether it worked [19, 25]. This histogram is a simple model: a cubic random background, small wrong patches and large real twins.

How do I fix misindexing?

1. Load only the phases that can be there, with checked lattice parameters.
2. Detect 9 to 12 bands, and calibrate the pattern centre on many patterns.
3. For phases with the same structure, add EDS to the phase decision (TruPhase, ChI-Scan).
4. Save the raw patterns. Re-index noisy or pseudosymmetric areas with dictionary or spherical indexing, refine each variant and keep the best match (for example kikuchipy refine_orientation with pseudosymmetry operators [26]).
5. Use pseudosymmetry cleanup only with a patch-size cap, and check the result against another method.
6. Report the indexing method, the number of bands, and every cleanup step with the map.

Go deeper: indexing methods compared · cleaning EBSD data honestly · sample preparation, for sharper patterns

Questions people ask about EBSD misindexing

References

@misc{tripathy2026ebsdmisindexing,
  author = {Tripathy, Manisha},
  title  = {EBSD Misindexing and Pseudosymmetry: Which Phases Go Wrong},
  year   = {2026},
  howpublished = {\url{https://untetheredatom.com/ebsd/ebsd-misindexing-pseudosymmetry}},
  note   = {Interactive web page}
}