501 29
where θ is the sample tilt with respect to the horizontal, Lperp
is the resolution perpendicular to the tilt axis, and Lpara is the
resolution parallel to the tilt axis. The resolution perpendicu-
lar to the tilt axis is roughly three times the resolution parallel
to the tilt axis for a tilt angle of 70° and increases to 5.75 times
for a tilt angle of 80°. Thus, it is best to work at the lowest
sample tilt angles possible consistent with obtaining good
EBSD patterns.
29.2.4 How Does a Modern EBSD System
Index Patterns
Modern EBSD systems all use the same basic steps to go from
a collected EBSD pattern to indexing and to the determina-
tion of the crystal orientation with respect to a reference
frame. Once the pattern is collected, the necessary informa-
tion for indexing of the pattern (indexing refers to assigning
a consistent set of crystallographic directions to the pattern
with respect to a given or a set of given candidate structures)
must be derived from the diffraction pattern. Once the pat-
tern has been indexed correctly (various vendors use differ-
ent measures of what constitutes correct indexing), it is a
simple matter to determine the crystallographic orientation
represented by the EBSD pattern.
The most important factor in obtaining a quality orienta-
tion is that the system have good quality patterns for index-
ing. The quality of the detected patterns needed for a given
experimental measurement can be influenced by choices
made over how the camera is operated. EBSD patterns are of
low inherent contrast mainly due to electrons that have lost
significant amounts of energy contributing to the overall
intensity of the pattern background. To compensate for this,
methods of removing the background are utilized. Some sys-
tems use a software-generated background to compensate for
the background and increase the pattern contrast. A more
established approach uses a background image obtained by
scanning over a large number of grains that produce the
background signal. This background is then used to normal-
ize the raw EBSD pattern to produce a pattern with high con-
trast (Michael and Goehner 1996 ).
EBSD cameras are generally able to collect patterns at
higher angular resolution per pixel than is needed for most
orientation mapping. However, if high accuracy is needed, it
is possible to use the full resolution of the camera to produce
EBSD patterns. The disadvantage of using the full resolution
of the EBSD camera is that longer exposure times are needed
to produce usable EBSD patterns. This longer exposure can
really slow down map acquisition. It is possible to bin the
camera resolution, usually by factors of 2, to produce lower
resolution patterns but with the ability to collect the patterns
at a much higher collection rate. Thus, if the high angular
resolution is not required, the camera should be used at a
reduced resolution, (2 × 2, 4 × 4, or 8 × 8 binning) to speed up
the acquisition. The tradeoff is between orientation accuracy
and speed of the measurement.
Once the camera is set correctly for the given experiment,
the patterns should be of high quality with good contrast.
The next step is for the software to extract the line positions
from the collected pattern. In all modern systems this is
accomplished with an algorithm called the Hough Transform,
which takes straight lines and transfers them into points05ab10 15 20# Backscattered electronsEnergy (keV)200 100 0 –100 –200 –300 –400 –500
Distance (nm)# BackscatteredelectronsAll electrons>18 kV>19.8 kV. Fig. 29.14 a Monte Carlo electron trajectory simulations of the back-
scattered electron distributions for Ni at 20 kV at a tilt of 70° (red) and for a
sample at 0° tilt (blue) that is normal to the electron beam. b Backscattered
electron distributions from Ni at 20 kV and a sample tilt of 70° for different
levels of energy loss as a function of distance from the beam impact point29.2 · Electron Backscatter Diffraction in the Scanning Electron Microscope