3D Data for Fatigue in Superalloys
3
Fatigue cracks typically initiate at the “weakest link” of the material structure.
Rigorous models for fatigue thus require knowledge of the volume of the material
that must be interrogated to capture the “rare” combinations of material structure
that result in early strain localization and subsequent crack initiation [6, 7]. This,
in turn, requires three-dimensional information on the distribution of important
structural features: precipitates, annealing twins, grains, and in some cases carbides,
nitrides, and oxides.
Nickel-base superalloys used for disks have microstructure at various length
scales from precipitate structure (10’s–100’s nm) to twin related domain structure
(10’s–100 μm) with grains containing multiple, fine twinned structures of varying
sizes that are crystallographically related to the parent material.
2 Importance of 3D Data
Many materials can be characterized using targeted 2D sections to analyze the
microstructure, especially when the microstructure is isotropic and its features can
be captured with well-known distributions [17, 18]. However, 2D inferences about
structure and crystallography will be incomplete when investigating materials with
rare features or heterogeneously distributed microstructure [17, 19, 20].
Nickel-base superalloys used for disks have microstructure at various length
scales from precipitate structure (10’s–100’s nm) to twin related domain structure
(10’s–100 μm) with grains containing multiple, fine twinned structures of varying
sizes that are crystallographically related to the parent material. Full 3D characterization is required to quantify the geometrical characteristics of the twins as well as
to capture the five grain boundary parameters (three orientation parameters and two
boundary normal parameters) [21–23]. The twin structures, which have been shown
to be critical for the localization of strain [15, 16] and eventually the initiation of
fatigue cracks [8, 11, 12], can be thin compared to the grain structure (μm thick)
and may or may not extend across the entire grain.
A range of 3D tomography techniques have emerged in recent years that utilize
femtosecond pulsed lasers [24, 25], mechanical polishing [26–29], broad ion beams
[30], focused ion beams (FIB) [31–33], plasma FIBs [21, 34], and microtomes or
serial block face SEM imaging [35, 36] to remove material in a serial sectioning
approach. If only grain information is needed, then a combination of near-field [37–
39] and far-field X-ray imaging allows for direct, nondestructive 3D characterization
[40–44]. With current data collection and reconstruction methods, the X-ray
diffraction methods have difficulty reconstructing crystallographic features that are
below 5–10 μm in size, including fine twin structure, and also with crystals with
preexisting strain gradients such as in samples that have been plastically deformed.
Here we focus on serial sectioning approaches, due to the presence of thin micronscaled annealing twins which are challenging to characterize with X-ray techniques.
Manual serial sectioning polishing techniques are effective for relatively coarse
sectioning resolutions, especially if fiducial depth markers are incorporated; how-
3
Fatigue cracks typically initiate at the “weakest link” of the material structure.
Rigorous models for fatigue thus require knowledge of the volume of the material
that must be interrogated to capture the “rare” combinations of material structure
that result in early strain localization and subsequent crack initiation [6, 7]. This,
in turn, requires three-dimensional information on the distribution of important
structural features: precipitates, annealing twins, grains, and in some cases carbides,
nitrides, and oxides.
Nickel-base superalloys used for disks have microstructure at various length
scales from precipitate structure (10’s–100’s nm) to twin related domain structure
(10’s–100 μm) with grains containing multiple, fine twinned structures of varying
sizes that are crystallographically related to the parent material.
2 Importance of 3D Data
Many materials can be characterized using targeted 2D sections to analyze the
microstructure, especially when the microstructure is isotropic and its features can
be captured with well-known distributions [17, 18]. However, 2D inferences about
structure and crystallography will be incomplete when investigating materials with
rare features or heterogeneously distributed microstructure [17, 19, 20].
Nickel-base superalloys used for disks have microstructure at various length
scales from precipitate structure (10’s–100’s nm) to twin related domain structure
(10’s–100 μm) with grains containing multiple, fine twinned structures of varying
sizes that are crystallographically related to the parent material. Full 3D characterization is required to quantify the geometrical characteristics of the twins as well as
to capture the five grain boundary parameters (three orientation parameters and two
boundary normal parameters) [21–23]. The twin structures, which have been shown
to be critical for the localization of strain [15, 16] and eventually the initiation of
fatigue cracks [8, 11, 12], can be thin compared to the grain structure (μm thick)
and may or may not extend across the entire grain.
A range of 3D tomography techniques have emerged in recent years that utilize
femtosecond pulsed lasers [24, 25], mechanical polishing [26–29], broad ion beams
[30], focused ion beams (FIB) [31–33], plasma FIBs [21, 34], and microtomes or
serial block face SEM imaging [35, 36] to remove material in a serial sectioning
approach. If only grain information is needed, then a combination of near-field [37–
39] and far-field X-ray imaging allows for direct, nondestructive 3D characterization
[40–44]. With current data collection and reconstruction methods, the X-ray
diffraction methods have difficulty reconstructing crystallographic features that are
below 5–10 μm in size, including fine twin structure, and also with crystals with
preexisting strain gradients such as in samples that have been plastically deformed.
Here we focus on serial sectioning approaches, due to the presence of thin micronscaled annealing twins which are challenging to characterize with X-ray techniques.
Manual serial sectioning polishing techniques are effective for relatively coarse
sectioning resolutions, especially if fiducial depth markers are incorporated; how-