EBSD.com is an educational site by Oxford Instruments

Advances in EBSD – solving the most challenging applications

As the use of EBSD becomes more widespread the number and type of applications are growing and in many cases becoming more challenging. In the Oxford Instruments AZtec system product development has delivered significant improvements in both the quality of data you can collect and the robustness of the system in order to solve these more complex applications. The engine that delivers these improvements is called Tru-I. Some examples are given here

Band Detection Improvements

Band detection is a fundamentally important aspect of EBSD analysis and is critical in terms of data quality. Improvements in band detection can increase the number of correctly indexed points, especially for materials that exhibit indistinct bands or have low quality patterns, such as those consisting of highly strained or partially recovered microstructures.

The band detection in a modern system uses a sophisticated algorithm which automatically determines which of the detected bands are best used in the indexing. This method applies a weighting function based on the band's average intensity and the position of the band within the 'area of interest '- i.e. the area of the EBSP used for band detection (Figure 1). Using this method higher priority is given to those bands at or near the centre of the area of interest, and these bands, which are more reliably detected, are used in the indexing, thus making the indexing more robust.  This weighted band detection illustrated in the figures below.

Figure 1 Weighted band detection.

A typical EBSP from a low density silicate mineral

(a) A typical EBSP from a low density silicate mineral;

Non-weighted band detection

(b) Non-weighted band detection;

Weighted band detection

(c) Weighted band detection.

Indexing Algorithm

The indexing routine is critical in achieving accurate data. The challenge for an EBSD system is to achieve a high validated hit rate without generating false positive solutions; where EBSPs are solved but with the wrong solution. Automated routines can be very sensitive to non-fitting bands, such as those occurring at grain boundaries, so the indexing algorithm must be robust enough to handle inaccuracies in band detection.

A new robust method: "Class Indexing" examines permutations of four bands, and considers band coherence (where the measured bands and reference crystal reflectors are in agreement) and non-coherence (where the measured bands and the reference crystal reflectors are not in agreement). These four-band combinations enable the indexing routine to break the solutions into smaller blocks, which are then the foundations of the indexing. Using this method, the routine is far more robust and can find the correct solution even when one or more bands are non-coherent. There are several benefits to this sophisticated method:

Grain Boundary Resolution

This Class indexing method better accommodates overlapping patterns, such as those which occur at  grain boundaries. The solution from the dominating group of bands (i.e. from one side of the boundary) generally supersedes the wrong four-band combinations including representatives from either side. As a result grain boundaries are more accurately resolved and there are fewer non solutions. This is illustrated in Figure 2.

Figure 2 Improved indexing for overlapping patterns at grain boundary in a steel sample using AZtec Class indexing method. The same data set was reprocessed

Using an older  indexing method, a high number of patterns at grain boundary are not indexed and presented as zero solution

(a) Using an older  indexing method, a high number of patterns at grain boundary are not indexed and presented as zero solution;

using AZtec Class indexing method, overlapping patterns at grain boundary are better accommodated and indexed correctly

(b) using AZtec Class indexing method, overlapping patterns at grain boundary are better accommodated and indexed correctly.

Distinguishing Similar Crystal Structures

Traditionally, it is a challenge using EBSD to distinguish phases with similar crystal structures, as this technique uses the angles between the bands to distinguish the phases. To assist in these situations where the structures are similar it is possible to measure differences in lattice parameter (resulting in differences in band width) to distinguish these phases. This is illustrated in Figure 3, in differentiating Pt from Ni in an electrode interface:

Figure 3 Pt-Ni interface from the central electrode of an automotive spark plug. Platinum and nickel have the same crystal structure with only 10% difference in lattice parameter and therefore can be difficult to differentiate on the basis of traditional EBSD. By using the band width the AZtec system is able to differentiate them.

Pt and Ni have the same crystal structure with only 10% difference in lattice parameter

(a) Pt and Ni have the same crystal structure with only 10% difference in lattice parameter.

Phase map overlaid onto the electron image. This map is processed using the traditional routine indexing algorithm to solve the patterns. This map shows no differentiation of the Pt or Ni, but an arbitrary solving of the pixels

(b) Phase map overlaid onto the electron image. This map is processed using the traditional routine indexing algorithm to solve the patterns. This map shows no differentiation of the Pt or Ni, but an arbitrary solving of the pixels.

A phase map of the same area with the two phases clearly differentiated; this is achieved by grouping the two phases and sorting the solutions using the pattern band width

(c) A phase map of the same area with the two phases clearly differentiated; this is achieved by grouping the two phases and sorting the solutions using the pattern band width.

The corresponding X-ray maps collected simultaneously with the EBSD data, illustrate the concentration of these elements in the two phases

(d) The corresponding X-ray maps collected simultaneously with the EBSD data, illustrate the concentration of these elements in the two phases.

Advanced Phase Differentiation

An alternative method is to combine the EBSD data with chemical data, by simultaneously collecting EDS (Energy Dispersive Spectrometry) and EBSD data. This technique is possible with AZtecTruPhase. This requires EDS and EBSD to be at a comparable geometry on the SEM (shown in Figure 4). A reference spectrum is collected from those phases with similar structure to assist in differentiating the phases. 

The EBSP and EDS spectrum are acquired simultaneously. When the EBSP, from a point, offers more than one viable phase, the simultaneously acquired EDS data is correlated against the reference spectrum in order to rank the solutions.  This weights the possible EBSD solutions and the result is an accurate differentiation of these similar phases. This is illustrated in Figure 5. In this example a Ni superalloy is examined. The second phases are the result of precipitation hardening, and it is important to be able to identify these correctly.

Figure 4 Geometry for combined EDS and EBSD analysis of a  sample.

Geometry for combined EDS and EBSD analysis of a  sample

Figure 5 A Ni-based superalloy contains a number of secondary phases within the nickel matrix including pervasive needle-like, Nb-rich precipitates and rounded Nb /Ti carbonitrides precipitates. The carbonitrides and Nihave very similar crystal structures and therefore can be difficult to differentiate on the basis of traditional EBSD. These phases have different chemistries, therefore they can be differentiated using AZtec Truphase.

An EBSD band contrast map, illustrating the nickel matrix coupled with the needle-like Ni-Nb precipitates and the rounded carbonitrides

(a) An EBSD band contrast map, illustrating the nickel matrix coupled with the needle-like Ni-Nb precipitates and the rounded carbonitrides.

Crystallographic properties of the phases in the superalloy. </em><em>The carbonitrides are crystallographically similar to the Ni; they have an FCC structure, belong to the same space group (225) and have similar unit cell dimensions. This makes it difficult to distinguish them, both from each other and from the matrix, using EBSD alone

(b) Crystallographic properties of the phases in the superalloy. The carbonitrides are crystallographically similar to the Ni; they have an FCC structure, belong to the same space group (225) and have similar unit cell dimensions. This makes it difficult to distinguish them, both from each other and from the matrix, using EBSD alone.

These phases have different chemistries. The X-ray maps show the needle-like precipitates have a higher concentration of Ni and Nb, whereas the rounded carbonitrides give higher Nb and Ti and markedly lower Ni concentration.  This indicates these phases can be differentiated using AZtec Truphase>

(c) These phases have different chemistries. The X-ray maps show the needle-like precipitates have a higher concentration of Ni and Nb, whereas the rounded carbonitrides give higher Nb and Ti and markedly lower Ni concentration.  This indicates these phases can be differentiated using AZtec Truphase.

A standard EBSD phase map solved against the two phases Ni and Ni3Nb. Although the matrix and Ni3Nb phases are readily distinguished, many of the carbonitrides index as Ni

(d) A standard EBSD phase map solved against the two phases Ni and Ni3Nb. Although the matrix and Ni3Nb phases are readily distinguished, many of the carbonitrides index as Ni.

The application of AZtecTruPhase enables an accurate identification of the phases in the Ni superalloy

(e) The application of AZtecTruPhase enables an accurate identification of the phases in the Ni superalloy.

Highest Orientation Precision

The precision and accuracy of routine EBSD measurements is constrained by many factors, and the angular resolution of standard EBSD systems is often quoted as between 0.5 – 0.7 degrees (e.g. Humphreys 2001).  Fundamental to the EBSD technique is the method of identifying and locating the Kikuchi bands, and the method chosen is one of the main factors determining the measurement accuracy and precision.

To achieve the highest precision misorientation measurements (for example to characterise subtle sub-grain structures), the Refined Accuracy (K. Thomsen 2013) mode in Oxford Instruments AZtec offers a solution. This new band refinement method improves significantly upon the angular resolution and accuracy of conventional EBSD without impacting speed and practicality of analysis. This method recognises that Kikuchi bands are hyperbolic and refines the fit accordingly. This method follows three stages:

1. Primary band detection
Use fast, low resolution, 2-D Hough to detect a set of bands (approx. ρ, ϕ determined)
2. Indexing
Determine phase by matching measured bands to a data base; (Bragg angles are now determined)
3. Secondary band refinement
Given the Bragg angle, the expected hyperbola is simulated; a relatively fast optimisation can now be performed, constrained to (ρ, ϕ) only, fitting the simulated band to the real, recorded band (in the high-resolution EBSP space). (Exact ρ, ϕ now determined)

Figure 6 Sketch to illustrate optimising the fit of a single, simulated Kikuchi band to the real, measured Kikuchi band (in the high-resolution EBSP space) – fit parameters are (ρ, ϕ).

Sketch to illustrate optimising the fit of a single, simulated Kikuchi band to the real, measured Kikuchi band (in the high-resolution EBSP space) – fit parameters are (</em><em>ρ, </em><em>ϕ</em>

The result of applying AZtec Refined Accuracy is illustrated in the histogram below. 

A single crystal silicon is used to characterize orientation precision (K. Thomsen 2013); this removes complications resulting from the sample and simplifies analysis of different band detection and calibration methods. With a single crystal, the orientation should be constant throughout the field of view so differences in orientation from point to point are due to variability in the measurement.

As examples of the improvement offered by the new band detection, the same set of EBSPs from a single map have been processed using both the conventional Hough-based method (in blue); and the new refinement method (in yellow). Figure 7 is a plot of KAM distribution (kernel average misorientation is a calculation of the average misorientation between each pixel and its nearest neighbours).

It is clear that by applying the secondary refinement the kernel average misorientation measurement is significantly improved. These improvements indicate a more accurate fit between pattern and solution which should yield benefits in ability to distinguish between phases and in misorientation precision.

Figure 7 KAM distribution, data collected from single crystal Si. It shows the distribution and average KAM value decreases with a mean KAM<0.05 degrees when using Refined Accuracy.

KAM distribution, data collected from single crystal Si. It shows the distribution and average KAM value decreases with a mean KAM<0.05 degrees when using Refined Accuracy

This increase in orientation accuracy offers the capability to reveal subtle subgrain structures in materials due to strain. This is illustrated in Figure 8.

Figure 8 The sample is a crack tip in a Ni superalloy. Data is collected by AZtec and reprocessed using (a) traditional indexing and (b) Refined Accuracy. A kernel average misorientation (or Local Average Misorientation) map generated from AZtec The higher level of misorientation, and therefore strain, is seen in the brighter green color in the grains around the crack tip with a complex pattern of deformation and the formation of sub cells of 2 μm in size. In Refined Accuracy data there is significantly less noise in the map and subtle subgrains are clear.

Traditional Indexing

(a) Traditional indexing.

 Refined Accuracy

(b) Refined Accuracy. A kernel average misorientation (or Local Average Misorientation) map generated from AZtec The higher level of misorientation, and therefore strain, is seen in the brighter green color in the grains around the crack tip with a complex pattern of deformation and the formation of sub cells of 2 μm in size. In Refined Accuracy data there is significantly less noise in the map and subtle subgrains are clear.

A kernel average misorientation (or Local Average Misorientation) map generated from AZtec The higher level of misorientation, and therefore strain, is seen in the brighter green color in the grains around the crack tip with a complex pattern of deformation and the formation of sub cells of 2 μm in size. In Refined Accuracy data there is significantly less noise in the map and subtle subgrains are clear.

PseudoSymmetry

Refined Accuracy has advantages in the discrimination of fine inter-planar angle differences. This helps in specific phases that generate extremely similar Kikuchi patterns for different crystallographic orientations. In these cases the indexing engine may not clearly identify the correct orientation solution. This phenomenon is called "pseudo symmetry", in these cases, only very slight differences in inter-band angle separate candidate solutions, and only robust and accurate band detection, as offered by Refined Accuracy, may identify the correct one among them. Here is an example of AZtec Refined Accuracy in solving pseudo symmetry problems in γ-TiAl.

Pseudo symmetric indexing problems in γ-TiAl arises from its close tetragonal c : a unit cell parameter ratio of 1.018, giving the generated Kikuchi patterns a pseudo-cubic configuration. This results in indexing inaccuracies, commonly with 60° orientation solution errors about the primary prismatic axes. These errors show the same mis-orientations as boundaries between real γ-TiAl lamellaes, causing additional problems in revealing the true microstructure. There are three variants of γ/ γinterfaces:

Pseudo Symmetry

The definition of the crystal structure including the <111> 3-fold pseudo-symmetry element, will ensure that AZtec generates the 3 pseudo-symmetry related solutions. Due to the application of Refined Accuracy, the band detection has a high accuracy, which in turn leads to low MAD (misfit) values. The solution with the lowest MAD value is chosen as the correct solution. This is illustrated in Figure 9.

Figure 9 Analysis of γ-TiAl with AZtec pseudo-symmetric processing.

A typical γ-TiAl pattern with the 3 pseudo-symmetry related solutions overlaid.Due to the application of Refined Accuracy, the band detection has a high accuracy, which in turn leads to low MAD (misfit) values. The solution with the lowest MAD=0.11<sup>o</sup> is chosen as the correct solution

(a) A typical γ-TiAl pattern with the 3 pseudo-symmetry related solutions overlaid.Due to the application of Refined Accuracy, the band detection has a high accuracy, which in turn leads to low MAD (misfit) values. The solution with the lowest MAD=0.11o is chosen as the correct solution.

IPF-z of the γphase and in addition the α<sub>2</sub> phase collected using traditional indexing method. The speckled appearance indicates difficulty in resolving the pseudo-symmetry

(b) IPF-z of the γphase and in addition the α2 phase collected using traditional indexing method. The speckled appearance indicates difficulty in resolving the pseudo-symmetry.

IPF-z of the γphase and in addition the α<sub>2</sub> phase collected using AZtec Refined Accuracy with pseudo-symmetry definitions. The data clearly shows the accurate grain structure without pseudo-symmetric mis-indexing

(c) IPF-z of the γphase and in addition the α2 phase collected using AZtec Refined Accuracy with pseudo-symmetry definitions. The data clearly shows the accurate grain structure without pseudo-symmetric mis-indexing.

Transmission Kikuchi Diffraction (TKD)

There is an increasing requirement to characterise materials on the nanoscale and despite significant technological developments in recent years, EBSD is still limited by the pattern source volume to resolutions in the order of 25-100nm. This is insufficient to accurately measure truly nanostructured materials (with mean grain sizes below 100 nm).

A new approach to SEM-based diffraction has received a lot of interest; it applies conventional EBSD hardware to an electron-transparent sample. The technique, referred to as transmission EBSD (t-EBSD: Keller and Geiss, 2012) or transmission Kikuchi diffraction (TKD: Trimby, 2012) as been proven to enable spatial resolutions better than 10 nm. This technique is ideal for routine EBSD characterisation of both nanostructured and highly deformed samples.

TKD samples are prepared in the standard way for transmission electron microscopy. The sample thickness is critical: best results are achieved using relatively thin samples, in the range of 50 nm to 150 nm.

The samples are typically mounted horizontally in the SEM chamber, at a level above the top of the EBSD detector's phosphor screen. This is illustrated in Figure 10. 

Figure 10 The geometry of a system set up for TKD. Electron transparent samples are mounted horizontally in the SEM chamber and posi­tioned towards the top of the EBSD detector's phosphor screen.

The geometry of a system set up for TKD. Electron transparent samples are mounted horizontally in the SEM chamber and posi­tioned towards the top of the EBSD detector's phosphor screen.

This is usually at a short working distance (e.g. 5-10 mm), depending on the position of the EBSD detector. This geometry maximises the opportunity to achieve the best spatial resolution when mapping.

Although the geometry when using TKD is very different from the geometry when using reflective EBSD the AZtecsystem will compensate for this change in geometry without the need to refine or change the calibration.

However, there are additional challenges with TKD patterns resulting from the TKD geometry. The most notable effects are:

  • Wider than normal bands at the base of TKD patterns
  • A non symmetric intensity across these wide bands

This is illustrated in Figure 11.

Figure 11 Challenges in solving TKD patterns using conventional band detection.

A typical TKD pattern from Al. Broad bands can be seen at the base of the pattern. The intensity across these broad bands is non symmetric resulting in the band edges being very bright or very dark

(a) A typical TKD pattern from Al. Broad bands can be seen at the base of the pattern. The intensity across these broad bands is non symmetric resulting in the band edges being very bright or very dark.

The same pattern with the bands detected using conventional band detection routine. There are several errors in the band detection resulting from the artefacts mentioned above. The most obvious is the incorrect band detection of the broadest lower band – where the non symmetric intensity has resulted in the top edge of the band only being detected. The poorer fit between pattern and solution is readily seen in the higher than acceptable MAD number

(b) The same pattern with the bands detected using conventional band detection routine. There are several errors in the band detection resulting from the artefacts mentioned above. The most obvious is the incorrect band detection of the broadest lower band – where the non symmetric intensity has resulted in the top edge of the band only being detected. The poorer fit between pattern and solution is readily seen in the higher than acceptable MAD number.

These artefacts can result in inaccuracies in band detection, with an associated drop in hit rate and reduced orientation accuracy. In order to address this, a new innovative TKD optimised mode has been implemented in the AZtecsoftware. This mode is designed to accurately solve these TKD patterns, to deliver the highest hit rate and most accurate data; this is illustrated in Figure 12.

Figure 12 Accurate solve TKD patterns using TKD optimised mode in AZtec. The same pattern is now processed using the TKD optimised step in AZtec. The bands, shown in red, are correctly detected. The black bands are those reflectors that were present in the match-unit (and within field of view). The MAD value is reduced to 0.13 which is very acceptable.

Accurate solve TKD patterns using TKD optimised mode in AZtec. The same pattern is now processed using the TKD optimised step in AZtec. The bands, shown in red, are correctly detected. The black bands are those reflectors that were present in the match-unit (and within field of view). The MAD value is reduced to 0.13 which is very acceptable.

 In TKD mode AZtec calculates the minimum Bragg angle expected from the list of candidate phases. Thus the minimum band widths relative to the Pattern Centre in the TKD geometry are also known so that false, narrow bands can be ignored. Band centres are correctly identified resulting in hit rates comparable to those of EBSD in reflection mode.

References
F.J. Humphreys, J. Mater. Sci. 36 (2001), 3833–3854.
K. Thomsen, Microscopy and Microsnalysis 19 (2013), 724-725
R.R. Keller, R.H. Geiss, Transmission EBSD from 10 nm domains in a scanning electron microscope, Journal of Microsopy 245 (2012) 245-251.
P.W. Trimby, Ultramicroscopy 120 (2012), 16-24

Go To Top