This paper describes objective technical results and analysis.

Any subjective views or opinions that might be expressed

in the paper do not necessarily represent the views of the U.S. Department of Energy or the United States Government.

SAND2018-12622C

Dynamics of Cottrell
Atmospheres and Dislocations

PRESENTED BY

Ethan Epperly and Ryan Sills

Wan
Sandia National Laboratories is a multirnission
Laboratory managed and operated by National
Technology Et Engineering Solutions of Sandia,
LLC, a wholly owned subsidiary of Honeywell
international inc., for the U.S. Department of
Energy's National Nuclear Security
Adrninistration under contract DE-NA0003525.
2 Outline

Introduction: Dislocations, Solutes, and Dynamic Strain Aging

Mathematical Model and Computational Considerations
• Results and Comparison with Existing Models
3 Outline

Introduction: Dislocations, Solutes, and Dynamic Strain Aging

Mathematical Model and Computational Considerations
• Results and Comparison with Existing Models

•••• •••••
•••• •010•
•0•• ••• •
•••• •• ••
4 Dislocations, the Mechanism for Plastic Deformation

• At the atomic level, metals are crystals—their atoms exist in regular, repeating
patterns. This pattern is the "crystal lattice".
• Dislocations are local distortions in the crystal lattice, in which an additional half-
plane of atoms have been introduced.
• When a metal is subjected to loading, its dislocations move, creating permanent
deformations in the metal which persist after loading is removed. This deformation
is called plastic strain.

•• •• 0 000 0 •• • • •
O 000 0 010 0 O 0 0 01_,
O 000 O 0 00 O 000
O 000 O 000 O 000
Crystal with Motion of Dislocation
Perfect Crystal
Dislocation Creates Plastic Strain
5 Dislocations, the Mechanism for Plastic Deformation

• Dislocations are line objects, and can curve and bend to form intricate structures.
• The strength of metals is determined by the collective action of ensembles of
dislocations and their interactions with other defects.
• Understanding the behavior of dislocations through theory, experiment, and
simulation has been an important area of research in materials science.

•
In 3D,Dislocations 3D Discrete Dislocation
are Line Objects Dynamics Simulations (LLNL)
6 Solutes, Single Atoms Different from the Surrounding Lattice

• Solutes are single atoms which differ from the surrounding lattice. Two types:
• Substitutional Solutes: an atom in the lattice has been replaced (e.g. alloys like AlMg)
Interstitial Solutes: an atom which lies in between lattice sites (e.g. carbon in steel)

Solutes expand (or contract) the surrounding lattice, causing a local distortion in
the lattice.
• Solutes strengthen a metal by restricting the motion of dislocations.

•••e O 000 O 0 00
cf o ff e O 000 O 0 00
•
oooo 0000 O 0• •
O 000 •• •• O 0••
Crystal with Motion of Dislocation
Perfect Crystal
Dislocation Creates Plastic Strain
7 Dynamic Strain Aging

Dynamic Strain Aging

• Below the dislocation, atoms are more spread out, leading bulky •••••
solute atoms to congregate underneath the dislocations •Or•
forming Cottrell atmospheres.
••••
••• •• ••••
•Oln )
••• • •••••
• ••• •el.•

• These atmospheres restrict dislocation motion, holding them in •CO•

place. ••••
This causes an increase in strength as dislocations arrested by
their solute atmospheres are unable to move, and thus unable •••••
to produce plastic strain.
•01*•
• This process is called dynamic strain aging(DSA)
• "Dynamic" because solute atoms must move to form these
•CO•
atmospheres. This motion takes time, making DSA dependent on the ••••
amount of time solutes have to move.
8 The Problem

• Many theoretical/computational models of DSA exist in the literature.

• Still a gap between predictions from theory and results of experiment.

• Two questions central to understanding DSA.

1. How quickly do solute atmospheres form?
2. How strongly do they restrict dislocation motion?

We seek to understand solute-dislocations and DSA via a continuum model.

• Continuum model gives a general understanding of many materials at the cost of perfect
accuracy for any specific material.

Computational Models in Materials Science

More Accurate Less Accurate
More Computation Less Computation
Smaller Systems Bigger Systems
Material-Specific More General

Density
Molecular Kinetic
Functional
Dynamics Monte Carlo
Theory
9 Outline

Introduction: Dislocations, Solutes, and Dynamic Strain Aging

Mathematical Model and Computational Considerations
• Results and Comparison with Existing Models

100

80

60

40

ax(t,x) 20

X(t)x)Vpt(x))
at kBT -20

-40

-60

-84
-100
100 -50 so 100
x0
/6
10 The Mathematical Model

• Solute distribution is modelled as a continuum concentration field x(t,x). (a)

• x is the fraction of solute sites filled.
• Motion of solutes is governed by the diffusion equation (b)
Chemical potential (c) contains enthalpic contribution 017 due to solute-dislocation
interaction and entropic contribution kgT ln x/(1 — x)
Solute concentration considered in a 2D plane perpendicular to dislocation line.

Diffusivity
AV Chemical Potential

aX(t,x) _ D AV
J.
V• k X(t)x)V
[1(x)
•
••
• •
0
.0
••
at BT
*--- Temperature
• • Diffusion Equation (b)
•• •
Solute Misfit Volume
s
( X
• p.(x) = p(x)AV + kgT ln
,a. 1 X
Dislocation Pressure Field
Continuum Approximation (a) Chemical Potential (c)
11 Computational and Numerical Considerations

• The diffusion equation (b) is a second-order,

Diffusion Equation (b)
nonlinear, parabolic partial differential equation.
dx(t,x) D
• Discretized using a second-order accurate finite = v (—
• x(t,x)vt,t(x)) I
difference scheme on a nonuniform mesh (d) at kBT
• Mesh has higher density near dislocation core, where
higher solute gradients are more difficult and more Mesh (d) I
important to resolve. 100

so
so
• Time stepping is performed using the implicit 40

trapezoidal scheme. 20

I 0
• Time step sizes is controlled adaptively to bound the error -20

-40
per step. .6.
.80
• A sparse direct solver, UMFPACK,is used to solve .1.0
100 -50 so 100

sparse linear systems.

• Implemented in a C++ code.
12 Outline

Introduction: Dislocations, Solutes, and Dynamic Strain Aging

Mathematical Model and Computational Considerations
• Results and Comparison with Existing Models

I la, 20
to,
S = 4.0
102 S = 10.0 15
101 S = 20.0
S = 40.0 10
10°
S = 60.0
S = 100.0
S = 150.0 5
10-2
S = 200.0
A 10-3
10-4 102 103 104
I 10— 10-310-2 10-1 100 101 102 103 104 to, (sec-1)
t/ttrans
13 Condensation

0.45
0.5
0.4
0.4
0.35

0.3 0.3
0.25

0.2 0.2
0.15
0.1 = max x(x)
0.1

0.05
o
o 50 100 150 200
tD/1)2

Computational simulation of atmosphere formation allows us to predict the rate of

formation and strength of atmospheres
14 Competing Models

• Two existing models for atmosphere formation:

1. Classical Continuum (Cottrell and Bilby 1949; Louat 1981): atmosphere condensation
results from solute atoms from outside the dislocation core entering the core—so called
"bulk diffusion".
2. Cross-Core (Curtin, Olmstead,and Hector 2006): atmosphere condensation results
from solute atoms crossing from one side of the dislocation to the other.

• Both diffusion types are important: can we capture both?

••,_,L,•10 •o_aA•
•(1
•Id<1* • (wit*
/a
•\,• ,/* •\,• ,.•
•i-e• •i-i•
Classical Continuum Cross-core
15 Bulk and Cross-Core Diffusion

• Our continuum model captures both the cross-core and bulk diffusion types.
• Cross-core diffusion was believed to be beyond the scope of the continuum model

•.
= 4.0
()2
, ••
S = 10.0
101 S = 20.0
S = 40.0 High S
i0°
S = 60.0
10-1 S = 100.0 • •
S = 150.0
•.
l 10-2 .
S = 200.0 • •
10-3
S = atmosphere strength
I' 10-140_4 101 Low S
10-3 10-2 10-1 100 1o2 1o4 10')
t/ttrans

• Our accurate continuum computations show the classical continuum theory is

based on bad assumptions and must be corrected.
16 Elevated Core Diffusivity •
Elevated Core
• It was observed by Picu and Zhang (2004) that the ••e G44 Diffusivity (e)
diffusivity D of solutes in the core is much higher •••L., 3 ••* Slow
than in the bulk. (e) ••,_•.,f•
• This can be incorporated in our model by using a •41 2
. 1",• Fast
diffusivity function (f), rather than a diffusivity
constant. •` -441
!'•
•41• "Diffusivity Radius"

20
Diffusion Function (f)
D(x) = Db +(Dc — Db)e -1142/27-A
Increasing rD
Bulk Diffusivity Core Diffusivity
Close agreement for
small strain rates • By choosing a sufficiently large diffusivity
radius, we are able to get reasonably close
Cross-Core
lation agreement between simulation (blues) and
Mode cross-core model (green). See (g)
901 102 103 104
(sec-1)

Strengthening versus Strain Rate (g)

17 Comparison of Models

• Two diffusion types for atmosphere formation:

1. Bulk
2. Cross-Core

Both diffusion types are important: can we capture both? Are both diffusion types
important?
Diffusion Types Experimental Agreement

Model Bulk Core Timescale Strength

Classical ./ X X X
Cross-Core X ./ ./ si
Us ./ ./ ./ X

• Bulk diffusion occurs in simulations of all types (molecular dynamics, kinetic

Monte Carlo, continuum), yet the best agreement with experiments is achieved by
ignoring it.
• The Puzzle: How can we develop of a theoretical model of DSA which agrees
with both simulation and experiment which includes bulk and core diffusion?
18 Conclusions

• Summary
• Developed a continuum model of dislocation-solute interactions
• Model is able to accurately capture both cross-core and bulk diffusion types
• By accounting for the elevated core diffusivity, the model correctly predicts the time scale
of DSA effects, but still over-predicts the strengthening

Draft manuscript under preparation

Acknowledgements
Ryan Sills, my mentor at Sandia and collaborator on this work
• Jonathan Hu,who mentored me on a different project this summer at Sandia
• Sandia's Student Intern Programs and my manager, Sean Stieper, who have both been
immensely helpful to me performing this work year-round

• References
• A. H. Cottrell and B. A. Bilby. Dislocation Theory of Yielding and Strain Ageing of Iron. Proceedings of the Physical
Society. Section A,62(1):49-62, 1949.
• W. A. Curtin, D. L. Olmsted, and L. G. Hector. A predictive mechanism for dynamic strain ageing in aluminium-
magnesium alloys. Nature materials, 5(11):875-880, 2006.
• N. Louat. On the theory of the Portevin-Le Chatelier effect. Script Metallurgica, 15(11):1167-1170, 1981.
• R. C. Picu and D. Zhang. Atomistic study of pipe diffusion in Al-Mg alloys. Acta Materialia, 52(1):161-171, 2004.
• R. B. Sills and W. Cai. Solute drag on perfect and extended dislocations. Philosophical Magazine, 96(10):895-921, 2016.