509 29
show up as single pixels surrounded by correctly indexed pix-
els. The software searches through the data and where there is
a single pixel of a different orientation surrounded by pixels of
a different but similar orientation, the single mis- indexed pixel
is replaced by the average of the surrounding orientations. The
next step is to deal with the pixels that are not indexed. The
same procedure is applied as with the single mis-indexed pixels
in that kernel math is used. Each individual pixel has six near-
est neighbors. If a cleaning procedure that requires six nearest
neighbors to agree then the single un-indexed pixel is assigned
the average orientation of it’s nearest neighbors. These proce-
dures allow the user to select the number of nearest neighbors
that must agree and using less than six nearest neighbors allows
larger regions of pixels with no correct indexing to be replaced.
Use of these procedures can be very dangerous and must be
disclosed when inverse pole figure maps are published.
29.2.9 Transmission Kikuchi Diffraction
in the SEM
One of the limitations of EBSD is that the resolution is com-
promised by the fact that the patterns are formed by back-
scattered electrons and that the sample is highly tilted leading
to the previously mentioned fact that the resolution perpen-
dicular to the tilt axis is much worse than that parallel to the
tilt axis. The resolution can be greatly improved if the back-
scattered volume is reduced and the geometrical factors are
reduced. One could image EBSD at reduced voltages to
reduce the interaction volume, but this process has practical
limitations related to the need for increased sample prepara-
tion quality. Also, improved EBSD cameras would be needed
to take advantage of lower voltage operation. Lower voltage
operation does nothing to reduce the geometrical effects.
Transmission Kikuchi diffraction (TKD, although some in
the literature have referred to this method as t-EBSD, which is
the acronym for transmission electron backscattered diffraction,
which is a rather non-physical description due to the inclusion
of both transmission and backscattered in the name) is the
transmission analog to EBSD is a way to achieve extremely high
spatial resolution for crystallographic orientation or phase map-
ping (Keller and Geiss 2012 ; Trimby 2012 ). TKD is realized by
using an electron-transparent sample, as in the transmission
electron microscope, that is held at normal or near- normal inci-
dence with respect to the electron beam while a standard EBSD
camera is placed at the exit surface of the sample, as shown in. Fig. 29.21. Here the electron beam accelerating voltage must
be high and the sample must be thin in order for transmission of
the electrons to occur. The maximum beam energy typically
available in modern SEMs is 30 kV, which requires that the sam-
ple must be quite thin. The use of a thin sample limits the size of
the beam interaction volume within the sample and immedi-
ately improves the spatial resolution of TKD. In addition, the
fact that the sample may be oriented normal to the incident elec-
tron beam further improves the resolution to the point that
2-nm spatial resolution has been achieved in TKD. It is incredi-
bly fortuitous that the TKD patterns are extremely similar in
appearance to EBSD patterns, as shown in. Fig. 29.22, and can
be collected with the same cameras and the same analysis soft-
ware as is used for EBSD. There are two main disadvantages to
TKD. First, it may be difficult to produce appropriate thin sam-
ples. However, most laboratories will have access to a dual-plat-
form FIB/SEM and thin samples prepared with FIB are perfectly
adequate for TKD. The second disadvantage is that when a small
pixel size is needed, it is difficult to map larger regions. For
example, if a map step size of 4 nm is selected, a 1000 × 1000 pixel
map will only cover 4 × 4 μm. However, if orientation mapping
of extremely fine grained material is needed, TKD may be the
only way to achieve the resolution needed.
A typical TKD map acquired at 30 kV of a thinned sam-
ple of polycrystalline Si layers in a semiconductor device is
shown in. Fig. 29.23. This map was acquired with a step size
ab. Fig. 29.21 Detector and sample positioning for transmission
Kikuchi diffraction. a The detector shown is an on-axis detector.
Sample and detector positioned and inserted for TKD with an on-axis
detector. Note row of solid state detectors located on the detector for
collecting transmission images of the sample. b Sample and detector
arrangement for TKD with a conventional style EBSD detector29.2 · Electron Backscatter Diffraction in the Scanning Electron Microscope