atomistic simulations

26
Atomistic Simulations Ju Li, Libor Kovarik

Upload: braeden-lane

Post on 01-Jan-2016

52 views

Category:

Documents


1 download

DESCRIPTION

Atomistic Simulations. Ju Li, Libor Kovarik. 8 nm. Ardell & Ozolins, Nature Mater . 4 (2005) 309. Mishin, Acta Mater . 52 (2004) 1451. (NT t ) ensemble with two vacancies. 2D activation. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Atomistic Simulations

Atomistic Simulations

Ju Li, Libor Kovarik

Page 2: Atomistic Simulations

8 nm

Mishin, Acta Mater. 52 (2004) 1451 Ardell & Ozolins, Nature Mater. 4 (2005) 309

Page 3: Atomistic Simulations
Page 4: Atomistic Simulations
Page 5: Atomistic Simulations
Page 6: Atomistic Simulations

(NT) ensemble with two vacancies

Page 7: Atomistic Simulations
Page 8: Atomistic Simulations

Transition pathwaysobtained using

Nudged Elastic Band (NEB) method.

Henkelman & Jonsson, J. Chem. Phys. 113 (2000) 9901;

ibid 113 (2000) 9978.

I

I IC

0.44MPa m0.75

KG G

2D activation

3D activation

sorta too long

Page 9: Atomistic Simulations

fN

kN

A new NEB methodconnecting to unstable final state

“Free-end” algorithm:

last nodeconstrained

to move only along

energy contour

Page 10: Atomistic Simulations

T. Zhu, J. Li, A. Samanta, H.G. Kim, S. Suresh, “Interfacial plasticity governs strain rate sensitivity and ductility in nanostructured metals,” PNAS 104 (2007) 3031.

Lu et al., Acta Mater.

53 (2005)2169.

dislocationtransmission?

Lu et al., Science 287 (2000) 1463; 304 (2004) 422.

Page 11: Atomistic Simulations

*

activationvolume v

strain-ratesensitivity m

Uniaxial tension[Lu04]

Nanoindentation[Lu05]

Atomistic calculation

Diffusion-controlled processes

Bulk forest hardening

Comparison of yield stress, activation volume and strain-rate sensitivity between experimental measurements and atomistic calculation

Nano-twinnedCopper

~ 1 GPa

*700 MPa

780 MPa

32212 b

34424 b

036.0025.0

023.0013.0

31.0~ b ~ 1

31000100 b 005.00 bulk~ b

yield stress

* extracted from measured hardness as .3

HH

*

activationvolume v

strain-ratesensitivity m

Uniaxial tension[Lu04]

Nanoindentation[Lu05]

Atomistic calculation

Diffusion-controlled processes

Bulk forest hardening

Comparison of yield stress, activation volume and strain-rate sensitivity between experimental measurements and atomistic calculation

Nano-twinnedCopper

~ 1 GPa

*700 MPa

780 MPa

32212 b

34424 b

036.0025.0

023.0013.0

31.0~ b ~ 1

31000100 b 005.00 bulk~ b

yield stress

* extracted from measured hardness as .3

HH

First time atomistic calculation provides strain-rate sensitivity information, at experimentally realistic strain rate of ~10-4/s.

Page 12: Atomistic Simulations

avg. shear stress = 750 MPa

Page 13: Atomistic Simulations

initial equilibrium

free-end node

node 2

node 3

node 4

Page 14: Atomistic Simulations

3initial saddlesupercellActivation volume is estimated by 100

Qb

G

constant supercellcalculation

saddle-point configuration

Page 15: Atomistic Simulations
Page 16: Atomistic Simulations

[112]/6[112]/6

pseudo-twinlayer

true-twinlayer

pure Ni column

half Ni column

push inpop out

slightly tilted viewred: Al black: Ni

Libor vacancy reorderingmechanism

Page 17: Atomistic Simulations

Vacancy-aided reordering in 2-layer pseudo-twin long behind dislocations

For comparison,VNi migration barrier

in perfect Ni3Al is 1.24 eV.

Page 18: Atomistic Simulations

shear stress = 900 MPa

Page 19: Atomistic Simulations
Page 20: Atomistic Simulations

Peter Sarosi

Page 21: Atomistic Simulations

High Tensile Strength and Ductility of Cu with Nano-Sized Twins

Lu et al., Science 287 (2000) 1463; 304 (2004) 422.

Page 22: Atomistic Simulations

Lu et al., Acta Mater.

53 (2005)2169.

dislocationtransmission?

Page 23: Atomistic Simulations

B

,

log 3strain-rate sensitivity activation volume *

logT

k Tm v

m

Lu et al., Acta Mater. 53 (2005) 2169.

Like other nanocrystals, nanotwinned Cu shows increased strain-rate sensitivity (~0.03) and small activation volume (~12b3)

Can atomistic calculation provide strain-rate sensitivity (m) and activation volume (v*) information of experimental relevance?

Page 24: Atomistic Simulations

stress

Act

ivat

ion

ener

gy Q

()

athermal threshold

ath

( )dQd

activation volume

0.7eV

0eV very likely to happen in 1s

very unlikely to happen in 1s

1 2

large 2

small thermal uncertainty

small 1

large thermal uncertainty

process 1

process 2

Stress-driven activated process

Larger meansthe activation is

more “collective”,less thermal

uncertainty & the process

more “athermal”.

point defect diffusion: ~0.02-0.1b3 forest dislocation cutting: ~103b3

J. Li, “The Mechanics and Physics of Defect Nucleation,” MRS Bulletin 32 (2007) 151-159.

Page 25: Atomistic Simulations

= 252MPaQtms=0.67eV

Qabs=0.49eV

Qdes~5eV

Page 26: Atomistic Simulations

In experiment, stress applied is uniaxial tension, not pure shear → Taylor factor M ≈ 3.1 to convert

shear stress to uniaxial stress : = M

int( , )True activation volume:

Q

We’ve computed tms≈79b3, abs≈des≈43b3

at = 252MPa.

* * *tms abs des

3 3

,

3

44 24

A conversion factor M/ between experimentally measured * and :

b b

vv v v