Nanomechanics: Atomistic

Modeling

Mark F. Horstemeyer, PhD

CAVS Chair Professor in Computational Solid Mechanics
Mechanical Engineering
Mississippi State University
[email protected]

Outline
• Introduction
• Energy
• Kinetics
• Kinematics
• Damage Mechanics
 METHODS USED DEPEND ON THE
ENTITY BEING MODELED
Method Entity Examples
first principles electron density functional theory
(DFT)
quantum chemistry (QC)
semi-empirical atom embedded atom method
(EAM) modified EAM (MEAM)
N-body, glue
empirical atom, grain, Lennard-Jones (LJ), Potts
dislocation dislocation dynamics (DD)
phenomena continuum Ficks Law
field elasticity/plasticity
 Energy: Embedded Atom
Method (EAM)
• Total energy E

  1
E   F    ( r )     ij (r ij )
i i ij

i  j(i )  2 ij

F i : embedding energy of atom i

i: electronic density of atom i

r ij : separation distance between atom i and j

 ij : pair potential of atom i and j

1. Molecular Dynamics (f=ma, finite temperatures)

2. Molecular Statics (rate independent, absolute zero)
3. Monte Carlo Simulations (quasi-static, finite temperatures)
 Determination of Atomic Stress Tensor
(Daw and Baskes, 1984, Phys. Rev)

Embedded Atom Method (EAM) and Modified Embedded Atom Method (MEAM) potentials
 
F    2  r 
E    1  
    
Local force determined from energy

f    E
Dipole Force Tensor (virial stress) is determined from local forces

ij   V1 m vi vj  2V1  ri f j

 

Note: the difference between EAM and MEAM is an added degree of angular rotations that
affect the electron density cloud  . For EAM, this quantity is simply a scalar, but for MEAM it
includes three terms that are physically motivated:

   free surfaces  shear  crystal asymmet ry

 EAM Represents the properties of Nickel
extremely well

Calc. Expt.
Lattice Constant (Å) 3.52 3.52
Cohesive Energy (eV) 4.45 4.45
Vacancy Formation Energy (eV) 1.59 1.6
Elastic Constants (GPa)
C11 246.4 246.5
C12 147.3 147.3
C44 124.8 124.7
Surface Energy (mJ/m2) 2060 2380
Stacking Fault Energy (mJ/m2) 85 125
 CURRENTLY DEVELOPED MEAM FUNCTIONS
COVER MOST OF THE PERIODIC TABLE

H He

Li Be B C N O

Na Mg Al Si P S
K Ca Sc Ti V Cr Mn Fe Co Ni Cu Zn Ga Ge As

Sr Y Zr Nb Mo Ru Rh Pd Ag Cd In Sn Sb

Ba La Hf Ta W Re Ir Pt Au Hg Tl Pb Bi

Pr Nd Gd Tb Dy Ho Er

Th U Pu

Impurities BCC FCC HCP DIA CUB

 Kinetics: Historical Background

Atomistic Stress
Clausius, Maxwell Viral Theorem, 1870
Maxwell Tensor form of Virial, 1874
Rayleigh, 1905
Irving-Kirkwood, 1949

Generalized Continuum Theories

Cosserat 1909
Truesdell, Toupin, Mindlin, Eringen, Green-Naghdi 1960s
 GEOMETRY FOR ATOMISTIC CALCULATIONS

• atoms follow atoms moved at constant velocity

Newton’s
second law
• temperature periodic in z
periodic,
free surface,
maintained at
or GB
300K in x
y

x
fixed atoms
z
 Deyield Deyield

5 4
4
10 atoms macroyield = 4.59 GPa 107 atoms
3.5
macroyield
4 microyield 1 = 3.79 GPa
3 microyield 1
= 4.51 GPa = 2.69 GPa
shear stress (GPa)

shear stress (GPa)

3 2.5 microyield 2
= 3.0 GPa
microyield 2 2
2 (0.2% offset)
= 4.55 GPa 1.5
1
1
0.5
0 0
0 0.02 0.04 0.06 0.08 0.1 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08
shear strain shear strain

(a) (b)

Shear stress-strain curve of material blocks at an applied strain rate of 2.4e8/sec with (a) ten thousand atoms and
(b) ten million atoms showing microyield 1 at the proportional limit, microyield 2 at 0.2% offset strain, and macroyield.
EVOLUTION OF DISLOCATION SUBSTRUCTURE

5
shear stress (GPa)

2 Strain rate 108/sec

1
T = 300K
0
0 0.05 0.1 0.15 0.2 0.25 0.3
shear strain
 LARGER SAMPLES HAVE CONSIDERABLY
LESS STRUCTURE IN THE
STRESS/STRAIN CURVES
14 14
shear stress (GPa)

12 12
10 10
8 8

shear stress (GPa)

6 6 2.4e8/sec
1.53e10/sec
4 2.4e8/sec 4
6.58e9/sec
2 1.53e10/sec 2
5.26e10/sec
0 0
0 0.05 0.1 0.15 0.2 0.25 0 0.05 0.1 0.15 0.2
shear strain shear strain

106 atoms 107 atoms

0.16 mm x 0.08 mm 0.5 mm x 0.25 mm
 Schmid type plasticity observed at nanoscale
Horstemeyer, M.F. Baskes, M.I., Hughes, D.A., and Godfrey, A.
"Orientation Effects on the Stress State of Molecular Dynamics
Large Deformation Simulations," Int. J. Plasticity, Vol. 18, pp.
203-209, 2002.
5

resolved shear stress (GPa)

12
single slip 4
shear stress (GPa)

10 double slip
quadruple slip
8 octal slip 3
pseudopolycrystal
6
2
4
1
2

0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.1 0.2 0.3 0.4 0.5 0.6
shear strain shear strain

Strain rate 1010/sec

T = 300K
 Yield Stress Depends on Specimen Size
Because of Dislocation Nucleation Dominance

196 atoms
1384 atoms 100,000 atoms
20000 atoms 1,000,000 atoms
12 4
3.5
10
3

shear stress (GPa)

shear stress (GPa)

8
2.5
6 2

4 1.5
1
2
0.5
0 0
0 0.05 0.1 0.15 0.2 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07
shear strain shear strain

Horstemeyer, M.F., Plimpton, S.J., and Baskes, M.I."Size Scale and Strain Rate Effects
on Yield and Plasticity of Metals," Acta Mater., Vol. 49, pp. 4363-4374, 2001.

Strain rate 2.4 x 108/sec

T= 300K
 A SIMPLE MODEL TO EXPLAIN STRAIN
RATE DEPENDENCE

T Dislocation is nucleated at a distance x0

T from the free surface at a critical stress *

d mv
  eff b  
3kT  
v
c Dislocation accelerates
dt 1 v
   
10b 2 1  v
2 2
according to classical law
c c

 0 M    I    0  0 Effective stress depends on
 eff  
M   I    0 M    I    0  0
 image stress, Peierls barrier,
and orientation
x0
I   * At nucleation there is no net
x stress on the dislocation

Yield is defined when b x = 0.2%

Experimental data examining yield stress versus
applied strain rate from Follansbee (1988) and
Edington (1969) for copper.

0.1
Follansbee (1988)
L=0.1 mm
0.01
yield/elastic modulus

Follansbee (1988)
L=1 mm

0.001

Edington (1969) L=2 mm

0.0001

Edington (1969) L=1.25 cm

10-5
10-5 0.001 0.1 10 1000 105 107
strain rate
PREDICTED STRAIN RATE DEPENDENCE
OF YIELD STRENGTH IS CONSISTENT
WITH EXPERIMENTS

0.15
yield stress/elastic modulus

Calculations 1332 atoms

2e4 atoms
0.1 1e5 atoms
106 atoms
107 atoms
Experiment 108 atoms (1.6 microns)
0.05 exp (Maloy et al. 1995)

0
0.001 0.1 10 1000 105 107 10 9 1011
strain rate (1/sec)
Strain Rate and Size Scale Effects

nanoscale
microscale
macroscale
0.1 g h i

0.08 c
yield/elastic modulus

0.06 b
e

0.04

a d
0.02

0
0.001 0.1 10 1000 105 107 109 101 1
strain rate
Size Scale Dependence Observed in Copper Also

8
7 332 atoms
23628 atoms
6
shear stress (GPa)

5
4
3
2
1
0
0 0.05 0.1 0.15 0.2
shear strain
Size Scale Effect Observed in Tension Also

100 atoms
324 atoms
4356 atoms
20

15
true stress (GPa)

10

0
0 0.05 0.1 0.15 0.2 0.25 0.3
true strain
 Volume averaged stress is a function of
volume per surface area
Horstemeyer, M.F. and Baskes, M.I., “Atomistic Finite Deformation Simulations: A Discussion on Length Scale
Effects in Relation to Mechanical Stresses,” J. Eng.Matls. Techn. Trans. ASME, Vol. 121, pp. 114-119, 1998.

1 interfacial force
microscopy experiments  Conventional theory
0.1 EAM calculations
predicts that yield stress is
independent of sample size
yield stress/elastic modulus

0.01  Because of their small

size, properties of materials
0.001 indentation and torsion to be used in nano-devices
experiments are predicted to be vastly
0.0001 different than the properties
of materials used in
10-5 large scale experiments conventional devices

10-6
10-10 10-8 10-6 0.0001 0.01 1
size (m)

Horstemeyer, M.F., Plimpton, S.J., and Baskes, M.I."Size Scale and Strain Rate Effects on Yield and
Plasticity of Metals," Acta Mater., Vol. 49, pp. 4363-4374, 2001.
 Size Scale is related to Dislocation Nucleation
(volume per surface area) and strain gradients

• EAM Ni
normalized resolved yield

• Expt.
various fcc
 x 0.38  1
x0 metals
stress

• Free
not converged
x
surface BC
rst strain rate • T = 300 K

Gerberich, W.W., Tymak, N.I., Grunlan, J.C., Horstemeyer, M.F., and Baskes, M.I., “Interpretations
of Indentation Size Effects,” J. Applied Mechanics, Vol. 69, No. 4, pp. 443-452, 2002
Schematic showing the stress-strain responses
at four different size scales.

length scale 1: nanoscale

stress

length scale 2: submicron scale

length scale 3: microscale

length scale 4: macroscale

strain
 Determination of Stress

classical continuum
Edge dislocation nonlocal continuum
MEAM s12
MEAM s21
EAM s12
10

Dislocations are fundamental defects stress (MPa)

related to plasticity and damage
0

-5

-10

-30 -20 -10 0 10 20 30

distance (Angstroms)
 Comparison of stress asymmetry in a simple
shear simulation at 300K with one dislocation
2

1.5 sig12
sig21

average shear stress (GPa)

1

0.5
 21
0

-0.5

-1

-1.5
0 10 20 30 40 50
length (anstroms)

sig21
 12
distance for
distance for 21
12 stress component
stress component
sig12
Distance range is 4-48 Angstroms
from center of dislocation
 Comparison of stress asymmetry in a simple
shear simulation at 300K with four dislocations

1.5 1.5
sig12
1 sig21

average shear stress (GPa)

1

0.5 0.5

0 0

-0.5 -0.5

-1 -1
10 15 20 25 30 35 40 45 50
length (anstroms)

distance for 12
sig21
distance for 21
stress component
stress component

Distance range is 12-44 Angstroms sig12

from center of three dislocations
 Comparison of stress asymmetry in a simple
shear simulation at 10K with one dislocation

2.5

2 sig12
sig21

average shear stress (GPa)

1.5

0.5

-0.5
0 10 20 30 40 50
length (angstroms)

sig21
distance for
distance for 21
12 stress component
stress component

Distance range is 4-48 Angstroms sig12

from center of dislocation
Comparison of stress asymmetry in a simple
shear simulation at 10K with four dislocations

sig12
1.5 sig21

average shear stress (GPa)

1

0.5

-0.5
10 15 20 25 30 35 40 45 50
length (angstroms)

distance for 12
sig21
distance for 21
stress component
stress component

Distance range is 12-44 Angstroms sig12

from center of three dislocations
 Comparison of stress asymmetry in a pure
shear simulation at 10K with four dislocations

0.012
-12
0.01 -21

0.008

0.006

Stre ss
0.004

0.002

-0.002

-0.004
10 15 20 25 30 35 40 45 50
Le ngth (an gstro ms)

distance for 12 distance for 21

stress component stress component

Distance range is 12-44 Angstroms -sig12

from center of three dislocations

-sig21
 Comparison of stress asymmetry in a pure
shear simulation at 10K with one dislocation

0.02
-12
-21
0.01

Stre ss
-0.01

-0.02

-0.03
0 10 20 30 40 50
Le ngth (an gstro ms)

distance for 12 distance for 21

stress component stress component

Distance range is 12-44 Angstroms -sig12

from center of three dislocations

-sig21
 Comparison of stress asymmetry in a pure
shear simulation at 10K with four dislocations

0.014

-12
0.012 -21

0.01

0.008

Stre ss
0.006

0.004

0.002

-0.002
10 15 20 25 30 35 40 45 50
Le ngth (ang strom s)

distance for 12 distance for 21

stress component stress component

Distance range is 12-44 Angstroms -sig12

from center of three dislocations

-sig21
 Comparison of stress asymmetry in a pure
shear simulation at 10K with one dislocation

0.02
-12
-21
0.01

Stre ss
-0.01

-0.02

-0.03

-0.04
0 10 20 30 40 50
Le ngth (ang strom s)

distance for 12 distance for 21

stress component stress component

Distance range is 12-44 Angstroms -sig12

from center of three dislocations

-sig21
 Local Continuum Theory
Equations
Local Continuum Equations
 
 rr  A 1r  Br sin
   A1r  3Brsin 
 r  A1r  Brcos

Where constants A, B incorporate elastic constants

and boundary conditions
 Nonlocal Continuum Theory
Equations
Nonlocal Continuum, Eringen 1977
  2 
 rr  2 m1bk a  1  B   31 e sin 
  
  2 


 
    2 m1bk a  1  3B    1 2   2 1 er sin
 
  2 
 r  2 m1bk a  1  B   31 e cos
  

where,
 2  P 2  2 a: internal characteristic length
B  1 P 1 e P
   k: constant
a ,
P   kR kr
a
b: relative radial displacement
EVOLUTION OF MICROSTRUCTURE SHOWS
TWIN FORMATION

4.3% 5% 2.3% 4.8%

compression tension
slice through sample
 Simulation Setup
140Å
70Å
Single crystal

35Å 5800 atoms

[0 1 1] [0 1 1]
[1 0 0] [1 0 0]

Low angle grain boundary

35Å 5860 atoms

[0 1 1]
[2 10 10] [1 0 0]
[10 1 1]
High angle grain boundary
35Å 5840 atoms

[1 0 0] [0 1 1]
[0 1 1] [1 0 0]
 Boundary Conditions
Fixed
Prescribed velocity
V=0.035 Å/ps

Fixed
300 K
ends Periodic Periodic
Forward loading
Fixed
Fixed

Flexible
Prescribed velocity
V=0.035 Å/ps

Flexible
300 K
ends Periodic Periodic
Forward loading
Fixed
Fixed
Single Crystal (Fixed-end
Low Angle Grain Boundary
(fixed-end)
High Angle Grain Boundary
(fixed-end)
 Bauschinger Effect Analysis
• Bauschinger Stress Parameter / Bauschinger Effect Parameter

 f
r  f
r
BSP  BEP 
 f
 f
y

 f: stress in the forward load path at reverse point

 r: yield stress in the reverse load path
 y: initial yield stress in the forward load path

• Bauschinger effects indicated by BSP and BEP

The larger the BSP and BEP are, the stronger the Bauschinger effect
 Bauschinger Effect Results
• BSP and BEP for fixed ends, reverse load at 9% strain

 f (GPa)  r (GPa)  y (GPa) BSP BEP

Single Crystal 6.22 3.27 6.92 0.475 2.105
Low Angle GB 6.79 2.22 7.61 0.672 2.783
High Angle GB 6.89 0.21 7.47 0.969 5.734

• BSP and BEP for flexible ends, reverse load at 8% strain

 f (GPa)  r (GPa)  y (GPa) BSP BEP

Single Crystal 4.22 2.04 3.34 0.517 1.244
Low Angle GB 3.92 -0.88 4.9 0.776 1.559
High Angle GB 3.65 0.74 3.2 0.797 3.222
Bauschinger Effects Summary
• Less constrained materials exhibit reduced
yield stresses compared to highly constrained
materials

• Single crystal material has the smallest

Bauschinger effects while high angle GB
material has the strongest for both fixed-end
and flexible-end BC’s

• Dislocation nucleation occurs earlier in low

angle GB than single crystal and high angle
GB cases for both BC’s
 Kinematics/Strain
• No size scale effects observed
– Macroscale torsion experiments, crystal
plasticity, and atomistics show the same
plastic spin independent of size scale
– Macroscale internal state variable theory,
crystal plasticity, and atomistics show same
strain contours independent of size scale
– High rate plastic collapse of same geometry in
macroscale experiments show identical result
at atomistic results
 Crystal plasticity/finite element simulation
of torsion single crystal Cu

max CCW rotation

Plastic
Spin

max CW rotation

peaks

oscillation troughs
Displ.
 peaks
troughs

Plastic Spin is same

throughout length
scales

Condition/Method wave amplitude ratio

Torsion
experiment (cm) 0.02
finite elements (cm) 0.05
molecular dynamics (solid 0.06 wave amplitude ratio=A/L
cylinder)

molecular dynamics (hollow 0.025

cylinder)
A

Simple Shear
finite elements (cm) 0.25 L

molecular dynamics (nm) 0.23

 STRAIN CONTOURS
AT 30% STRAIN - 8x1 Aspect Ratio

ISV

atomistics

crystal plasticity

orientation angle - crystal plasticity

 Average yield stress for a block of material of
varying aspect ratios computed with different
modeling methods: a) constant y, varying x; b)
constant x, varying y.
1
0.1 max crystalline strength
max crystalline strength
0.1 atomistics (8 unit cells)
atomistics (16 unit cells)

0.01 atomistics (8 unit cells)

0.01 atomistics (24 unit cells)
shear stress/G

shear stress/G
atomistics (16 unit cells)
0.001 atomistics (24 unit cells)
0.001
macroscale internal
0.0001 crystal plasticity state variable theory

0.0001
-5 crystal plasticity
10
macroscale internal
state variable theory

10-6 10-5
0 2 4 6 8 10 0 2 4 6 8 10
x/y ratio y/x ratio

(a) (b)
 EAM Model, Single-crystal Copper (1 3 4),
Periodic in z (4 unit cells)
constant 30 Å

velocity and
strain rate of
109 s-1 y

x
z

60 Å 13 Å
 Copper (1 3 4)
Partial Collapse

Crack initiation
point

112.2 ps
Experiment
al

40 ms FEA
EAM Copper (1 3 4) FEA

Fully Collapse
anisotropic
flow localization
inelastic behavior
point
8.328ms (removed at 5.5ms)
219.8 ps

experimental
 Copper (1 3 4)
Fully Collapse
FEA central curved lines
inner boundary

8.328ms (removed at 5.5ms)

EAM
 Size Scale Effect (30Å/18mm)

EAM
Experimental

EAM
Experimental

outer boundary inner boundary

Copper (1 3 4)
172.8 ps
Copper (0 0 1) 230.0 ps

flow localization

FEA FEA FEA

10.0 ms 12.5 ms 15.0 ms

Macroscale Modeling of Size Scale
Effects
• Observation 1: we observe a size scale
effect on stress state
• Observation 2: we do not observe a size
scale effect on the geometric, strain, or
plastic spin states
• Question: where should the macroscale
repository be for the size scale
dependence?
• Answer: constitutive relation in dislocation
density, not the strain tensor