Title: The investigation of the constraints imposed by the distributions of orientations and boundary norma
1The investigation of the constraints imposed by
the distributions of orientations and boundary
normals in sample space on the grain boundary
character distribution
Thesis Overview
- By Lisa Chan
- Advisor Dr. Anthony Rollett and Dr. Gregory
Rohrer
2Outline
2
3Motivation
- Grain Boundary Engineering (GBE) consists of
repeated cycles of deformation annealing, to
generate large fractions of special boundaries
and avoid strong recrystallization textures.
Palumbo et al., JOM., 50 2 40-43 (1998).
3
4Motivation
- Several ?3 boundaries with larger deviations
from ideal misorientation cracked.
Gertsman et al., Acta Mater., 49 (9) 1589-1598
(2001).
4
5Motivation
- Cracked ?3 boundaries were found to deviate more
than 5from the trace of the 111 plane.
Lin et al., Acta Metall. Mater., 41 (2) 553-562
(1993).
5
6Scientific Questions
- What are the limits on generating realistic
microstructures and grain boundary character
distribution based on given characteristics? - If fixed with grain shape and texture, can we
still optimize the distribution of CSL
boundaries?
6
7Hypothesis
Hypothesis The grain boundary character
distribution is constrained by the grain
orientations and grain boundary plane
orientations in sample space (grain shapes).
For example, it is not possible to obtain
populations of a specific boundary type with a
three-dimensional equiaxed polycrystalline
microstructure, and texture strength greater than
5.
7
8Microstructure Generation
2
- Microstructure Generation
- Synthetic Microstructures
- Microstructure Builder
- Plank Generator
- Voronoi Tessellation
- FIB-Reconstructed Structures
- Twin Insertion
2
8
9Microstructure Builder
9
10Microstructure Builder
11Plank Generator
11
12Voronoi Tessellation
Step 2. Segment volume such that each cell
contains only 1 point
Tetrakaidecahedron-shaped grains
12
13Tetrakaidecahedrons
14FIB-reconstructed Structures
Zirconia obtained from Dr. Shen Dillon, and
Inconel 100 obtained from Dr. Michael Groeber
15Twin Insertion
- Pick random grain
- Identify orientation of grain
- Rotate 111 vector from crystal to sample
reference frame - Use equation of plane
- D distance from center (range 0-1)
16Twin Insertion
- Tolerance controls thickness of twin
Tolerance (twin width) 0.5 Tolerance (
twin width) 3
17Crystallography Generation
3
- Crystallography Generation
- Fiber Texture
- Random Texture
- Rolling Texture
- CSL Misorientations
3
17
18Fiber Texture
Probability density distribution for normal
distribution
where ? mean ? standard deviation
Garbacz et al., Acta Metall. Mater., 412
475-483 (1993).
19Fiber Texture
20Random Texture
21Rolling Texture Rolled Copper
Rolled copper data obtained from Dr. Samuel Lim
22OD and MD of Rolled Copper
23CSL Misorientations
?3
?1, 3, 7
24Orientation Assignment
4
- Orientation Assignment
- Simulated Annealing Algorithm
- Validation of Algorithm
4
24
25Simulated Annealing
(1) Assign an orientation to each grain
(2) Simulated annealing minimization1
Approach a configuration that minimizes error
1Miodownik et al., Acta Mater., 47 (9) 2661-2668
(1999).
Saylors TMS 2002 slide
26Simulated Annealing
Weighting Values
or
where
R v - 1
2
Probability of accepting evolution step
T annealing temperature
27Weighting Values
Inputs
odf_weighting 0mdf_weighting 2
odf_weighting 0.67mdf_weighting 0.22
odf_weighting 1.82mdf_weighting 0.031
28Weighting Values
RMS ODF difference RMS MDF difference
best MD fit
Log scale
best OD fit
29Weighting Values
best OD and MD fit
RMS ODF difference RMS MDF difference
constant total error
30Weighting Values
best linear fit
31Validation of Algorithm
Garbacz et al., Scr. Mater., 23 (8) 1369-1374
(1989). Gertsman et al., Acta Metall. Mater., 42
(6) 1785-1804 (1994). Morawiec et. al., Acta
Metall. Mater., 41 (10) 2825-2832 (1993). Pan et
al., Scr. Metall. Mater., 30 (8) 1055-1060
(1994).
32Results
- Results
- Geometrical Effects on MD
- Number of Grains
- Grain Size Distribution
- Grain Shapes
- Effects of OD on MD
5
32
33Number of Grains
34Grain Size Distribution
35Grain Size Distribution
36Grain Shapes
37Effects of OD on MD
Most compatible OD and MD
37
38Summary
38
39Research Plan
39
40Research Plan
- Twin Insertion Algorithm
- maintain texture of material
- match twin cluster statistics observed
experimentally - match distribution of twins per parent grain
- match twin width distributions
- Grain Shape Analysis
- quantify grain shapes in current structures with
mean width - create structures with specific mean widths
- Improve GBCD Algorithm
- use performance analysis tools such as Shark and
Tau - parallelization of the algorithm such that more
CPUs are used to calculate GBCD - obtain additional GBCDs for exploration on the
limitations OD and grain shape has - on the GBCD
40
41Questions
41
42Supplemental Slides
42
43Simulated Annealing
ODF Error
Error
1/Temperature
Possible ODF
44Simulated Annealing
- Randomly populate Representative Volume
Element (RVE) OD in homochoric space - Sample RVE OD from step (1) to select
2000xN representative orientations
(N number of specific 2000 grain
RVEs used) - Randomly assign orientations to each grain of
each RVE - Perform internal simulated annealing loop to
shuffle orientations so that the MD of each RVE
matches the specified MD - Compute error as sum of error from all models,
automatically accept if error is lower or compute
probability of acceptance if error is higher - Accept if delt0 or random numberltexp(-de/t)
- Randomly perturb OD bins, reduce temp of
simulation and repeat
45Simulated Annealing Temperature
45
46Simulated Annealing Temperature
46
47Simulated Annealing Temperature
47
48Twin Cluster
48
49Random Texture Components
Values in volume percent ()
50Pole Figure for Random Texture
51Experimental GBCD
52Title
52