Multiphase analysis - combining EDS and EBSD · Automated phase recognition (PCA) 1 - The spectra...
Transcript of Multiphase analysis - combining EDS and EBSD · Automated phase recognition (PCA) 1 - The spectra...
Multiphase analysis - combining EDS and
EBSD
Review:
Conventional materials characterisation:
Analysing polyphase materials using EDS
Analysing polyphase materials using EBSD
Combined EDS and EBSD measurements
Phase identification
Phase discrimination during mapping
Simultaneous EBSD and EDS scanning
Chemistry assisted Indexing ChI scan
Outline
Plot showing drift path during map acquisition
EDS (live) spectral mapping
EDS area analysis: Spectral mapping
(live) Spectral mapping procedure:
• With a short dwell time a spectrum is collected for each pixel
in the entire image. This is repeated while adding all additional
counts to the spectrum for each pixel.
• Drift correction is often needed to ensure continuous
alignment of subsequent maps. At fixed intervals a reference
image is collected and matched to the original location.
Sulfide ore - example
Application example:
Spectral mapping, phase identification, and phase
differentiation on a polyphase sulfide ore rock
Spot mode in NiAs grain.
Acquisition parameters:
HT = 25 kV
Mag = 190 x
Matrix = 512 x 400 pixels
Scan area = 695 x 542 micrometer
Nr frames = 183
Acquisition time = 2.8 hrs
Spectral mapping data review
Spot mode in FeNiS grain.
Spectral mapping data review
Area mode in NiAs grain.
Spectral mapping data review
Area mode in NiFeS grain.
Spectral mapping data review
Freedraw mode in biotite grain.
Spectral mapping data review
Freedraw mode in (NiCoFe)AsS grain.
Spectral mapping data review
Full area spectrum.
Spectral mapping data review
As Ni
Cu
Sb SSi
Co
Fe
X-Ray maps are
constructed for all
elements as identified by
the auto peak ID routines
How many phases are
present ?
Collected X-Ray maps
A method of statistical analysis of spectral mapping data:
Phase cluster analysis
Allows the user to automatically find phases in the
recorded data without prior knowledge
Clusters spectra at each pixel based on similarity in the
channels counts
With PCA the user can automatically
• Find Clusters
• Review the phases
• Build Maps
• Match Spectra
Step 1 - The spectra from the first 4x4 block
of measurement points is assumed to be
from the first phase.
An automated routine has been
developed to automatically determine the
chemical fingerprint of each phase –
phase cluster analysis (PCA).
Phase 1
Phase cluster analysis
Step 2 - The spectra from the second 4x4
block is compared to the first phase.
Phase 1 Phase 2
Phase cluster analysis
Step 3 - If the second block matches the
first phase within a specified tolerance,
this second spectrum is added to the
first to refine the phase spectrum.
+
Phase 1 Phase 1
Phase cluster analysis
Step 4 - If the second block does not
match, this second spectra then defines
a second phase.
Phase 1 Phase 2
Phase cluster analysis
Step 5 - This comparative process is
continues until each 4x4 block has been
matched to a phase.
Phase 1 Phase 2 Phase 3
Phase cluster analysis
Step 6 - The spectrum of each individual
measurement pixel is then compared
with and matched to one of the
determined phases.
Phase 1 Phase 2 Phase 3
Phase cluster analysis
Automatic phase map of sulfide ore
PCA – phase distribution
Using the stored spectral data, a maximum
intensity spectrum can be extracted using the
maximum intensity found in a map data cube
for each individual channel in the spectrum.
Max Channel spectrum
As Ni
Cu
Sb SSi
C
o
FeZn
Reconstructed X-Ray maps
Add the Zn component to the
phase cluster analysis
Corrected phase distribution
AsNiFe and
NiZnS
composite RGB
maps
AsCoFe X-Ray
overlay
Combined X-Ray maps
Combined X-Ray maps
Phase identification
The chemistry was determined using EDS analysis
Combining EDS with EBSD crystal information provides the mineral phase
Example: grain containing Zn, Fe, and S
Zn Fe S phase EBSD pattern
(Zn,Fe) S Sphalerite
Cubic (Oh) [m3m]
a = 5.41 Å
EDS - EBSD phase identification
sphaleritepentlandite pyrrhotite
chalcopyrite NickelineCobaltitebiotite
EBSD patterns of all phases
OIM scan area on SE
image
EBSD mapping results
EBSD Image Quality map
Grain boundaries and polishing
scratches are clearly visible in
the darker phases.
Note that these are not visible in
the SEM image.
EBSD mapping results
EBSD maps:
Left: Image Quality map
Middle: Phase map 3 phases with similar
crystal structure
Right: Inverse Pole Figure crystal direction map
EBSD mapping results: conventional
Sphalerite EBSD pattern
Cu
bic
(O
h)
[m3
m]
sphalerite a=5.41 Å chalcopyrite a=5.60 Å cobaltite a=5.66 Å
ZnS pattern indexed as …
cobalt iron zinc
sulphur antimony arsenic copper
nickel
X-ray maps collected together with the EBSD map indicate the phase distribution.
Chemical information is used to select the proper EBSD crystal structure file.
Corresponding EBSD patterns are forced to be indexed with the correct phase.
EBSD mapping results: ChI-Scan
EBSD maps:
Left: Image Quality map
Middle: Phase map
Right: Inverse Pole Figure crystal direction map
EBSD mapping results: ChI-Scan
Issues with phase selection based on chemistry:
• Variation in EDS intensity over the scan area
-1- Because of the high-tilts required for EBSD,
there is often a change in the EDS signal with WD
-2- Beam instabilites may cause variations in
countrates during long scans
• Difference in spatial resolution of EDS and EBSD
The spatial resolutions of the two techniques are approximately 50 nm and
1 micron for EBSD and EDS respectively.
Thus, there will be some “smearing” at the boundaries where the EBSD
must be used exclusively for the phase differentiation.
These issues can be minimised with Automated Phase Recognition
Automated phase recognition (PCA)
1 - The spectra from the first 2x2 block of measurement points is
assumed to be from the first phase.
2 - The spectra from the second 2x2 block is compared to the
first phase.
3 - If the second block matches the first phase within a specified
tolerance, this second spectrum is added to the first to refine
the phase spectrum.
4 - If the second block does not match, this second spectra then
defines a second phase.
5 - This comparative process continues until each 2x2 block has
been matched to a phase.
Phase 1 Phase 2 Phase 3
PCA - Phase Cluster Analysis
PCA - Phase Cluster Analysis
Phase 1 Phase 2 Phase 3
6 - The spectrum of each individual measurement pixel is then
compared with and matched to one of the determined phases.
SE image BSE image
The dark phase is dolomite (Ca,Mg)CO3
The lighter phase is calcite CaCO3
Example: marble
CaCO3 (Ca,Mg)CO3
Possible phases are:
Calcite: CaCO3 Trigonal (D3d) [-3m] a = 4.94Å c = 16.85Å
Mg-rich Calcite: (Mg,Ca)CO3 Trigonal (S6, C3i) [-3] a = 4.99Å c = 17.06Å
Dolomite: (Mg,Ca)CO3 Trigonal (D3d) [-3m] a = 4.83Å c = 15.94Å
Marble: chemistry
calcite
calcite
dolomite
Marble: EBSD patterns
dolomite
calcite
dolomite
Marble: EBSD patterns
Red is calcite
Green is dolomite
Conventionally indexed
X-ray maps acquired during scanning - 16 points/second - 63 msec dwell time
X-Ray maps
The automated component analysis
found three components:
Green is dolomite, red is calcite, and
blue is a (Ca) Al K Si O phase
(probably feldspar).
Results from automated analysis
IPF on IQ map, yellow
boundaries are calcite
twins
Phase map, orange is
dolomite, yellow is
calcite
Calcite IPF map Dolomite IPF map
ChI-scan results
2 Phases3 Phases
4 Phases
5 Phases 7 Phases
9 Phases
With ChiScan it no longer matters how many phases
are in your material.
ChI-scan = multi-phase EBSD mapping
Thank you for your attention