Dislocation Nucleation Rate at Finite Temperature in Copper Nanorod Seunghwa Ryu1 and Wei Cai2 1 Physics, Stanford University 2 Mechanical Engineering, Stanford University 2 Dec 2009 MRS Fall Meeting, Boston
p. 1/13
Sampling Dislocation Nucleation at finite temperature -
Copper Nano Rod under Deformation of nanomaterial is controlled by - Nucleation is a rare event, i.e. Uniaxial Compresson dislocation nucleation.
(waiting time >> reaction time )
- Molecular dynamics => Limited time scale - Transition State Theory with Fb from Nudged Elastic Band method ( Zhu (2008) ) => Zero temperature Need to directly sample calculation T. Zhu, J. Li et al, PRL 100 025502 (2008) dislocation nucleation Embedded Atom Method – Mishin Potential at finite temperature ! 15 x 15 x 20 100 | 010 |100 Cu Nano Rod 2 Dec 2009 MRS Fall Meeting, Boston
p. 2/13
Outline 1. Limitation of Nudged Elastic Band (NEB)
2. Forward Flux Sampling (FFS) for Nucleation Rate and Free Energy calculation
3. Results
2 Dec 2009 MRS Fall Meeting, Boston
p. 3/13
Limitation of NEB Dislocation Nucleation Rate from Transition State I TST Number of Nucleation Site
Theory (TST) Fb (σ , T ) = Nν exp(− ) k BT
E( F(λ)
Trial Frequency
State A
Free Energy Barrier
E (σ,T=0) Fb(σ,T) State B λ Size of Partial Slip.
- NEB computes Energy Profile along nucleation path at 0K. Fb (σ , T ) = Eb (σ , T ) − TS (σ , T ) - FromEb (σ , T = 0) , How can we get ~ - Zhu et al (2008) assumes Fb (σ , T ) = (1 − T / Tm ) Eb (σ , T = 0) 2 Dec 2009 MRS Fall Meeting, Boston
p. 4/13
?
E(λ)
Free Energy Barrier
Eb(σ,T=0)
State A
State B λ Size of Partial Slip.
- What is physical reason behind ~ Fb (σ , T ) = (1 − T / Tm ) Eb (σ , T = 0) ?
Free energy decreases linearly with temperature, where Tm is surface melting temperature, approximated to be 1 T bulk ≈ 700 K m
2
Temperature dependence has the correct limit. At T=0, it recovers to NEB results, at T=Tm it becomes zero. This is a reasonable approximation. 2 Dec 2009 MRS Fall Meeting, Boston
p. 5/13
Sample Dislocation Nucleation at finite temperature Define reaction coordinate
λ from state A to state B
λ : number of atoms inside State B slip 0 0 0 0 r = | r r = | r − r | xi = max{| rij − rij |} ij i − ri | ij i j For each atom i, define State A
j
perfect crystal j : over all value A. H. W.neighbor Ngan, Proc.of R.iSoc. A 462, 1662 (2006) 400K 1.7GPa
xi > xc → slip λ : size of the largest xi > xc cluster of atoms with 2 Dec 2009 MRS Fall Meeting, Boston
Red – Surface Yellow – Atoms with
xi>xc Purple – Largest Cluster p. 6/13
Compute Dislocation Nucleation Rate from FFS λ : number of atoms in the largest cluster λ < λA : no slip,
λ > λB : with slip
I FFS = I λ0 P (λ B | λ0 ) Success Probability of small cluster with λ0 evolving into λB
Nucleation rate of 18 16 14
small clusters with λ0
λ0
10
n
λ
max
12
8 6 4 2
0
50
100
150
200
250
300
350
400
450
500
MCS
Time
P (λ B | λ 0 ) =
i = m −1
∏ P (λ
i +1
| λi ) Sample
success
i=0 FFS works for non-Markovian and non-equilibrium configurations process ! Allen et al, JCP (2006).
Crystallization T.:Li et al, Nat. Mat. (2009) Sanz et al, J. Phys(2008).
With Backward Switching, F (λ )
2 Dec 2009 MRS Fall Meeting, Boston
can be obtained!
Valeriani, JCP (2007)
p. 7/13
Nucleation Rate from FFS I
FFS
~ Fb (σ , T ) ~ I = Nν exp(− ) k BT
λ0
= I P (λ B | λ 0 )
10
10
700K
5
10
500K
5
10
600K 400K
I
500K
700K 600K
-1
(σ,T) (s )
10
TST
IFFS(σ,T) (s-1)
10
0
400K
0
10
10 1
1.2
1.4 1.6 Stress(GPa)
1.8
2
1
1.2
1.4 1.6 Stress(GPa)
1.8
2
FFS results predict much larger stress dependence (i.e. stiffer slope w.r.t stress change) on the nucleation rate! Where does discrepancy come from? 2 Dec 2009 MRS Fall Meeting, Boston
p. 8/13
Comparison of the Prefactor PrefactorA
FFS
? or Free energy Fbarrier b (σ , T ) FFS
FFS
TST Now, Prefactor of TSTAis = Nν
0.02
AFFS/ATST
0.015
0.01 400K 500K 600K 700K
0.005
0 1
1.2
1.4 1.6 Stress(GPa)
Fb (σ , T ) ) k BT ~ F (σ , T ) ~ because I = Nν exp(− b ) k BT The prefactor is almost independent of T A FFS ≈ 0.01⋅ ATST and σ. Recrossing event would reduce the The prefactor! nucleation rate
I that =A such
A Define Prefactor of FFS
1.8
2 Dec 2009 MRS Fall Meeting, Boston
FFS ?
FFS
exp(−
discrepancy is Mainly from free energy barrier! 2 p. 9/13
2
Temperature and Stress Dependence of Free Energy Barrier 2
1.5
Ω(σ , T ) 0K NEB
1
0.5
0 1
Fb (eV)
Fb (eV)
1.5
FFS 700K
1.2
1
0K NEB
Ω(σ , T )
0.5 FFS 600K
FFS 500K
1.4 1.6 Stress(GPa)
FFS 400K
500K 600K
1.8
2
Circles : our data, Line : fit to smooth curve What does this imply? ∂Fb (σ , T ) Ω(σ , T ) = −
400K
0 1
700K 1.2
1.4 1.6 Stress(GPa)
1.8
~ Fb (σ , T ) = (1 − T / Tm ) Eb (σ , T = 0)
Eb (σ , T = 0) = 4.8(1 − σ / σ ath ) 4.1
∂σ
Activation Volume Is constant or increasing function of 2T.Dec 2009 MRS Fall Meeting, Boston
p. 10/13
2
Entropy Contribution in Fb 1
1.5
1
1.2 GPa
Fb (eV)
Fb (eV)
0.8
1.2 GPa
1.7 GPa
0.5
0
0.6 0.4
1.7 GPa
0.2
S (σ , T )
S (σ , T ) -0.5 300
400
500 Temperature
Entropy S (σ , T ) = −
600
∂Fb (σ , T ) ∂T
700
0 300
400
500 Temperature
Entropy S (σ , T ) = −
600
700
∂Fb (σ , T ) ∂T
is (slightly) increasing function of is decreasing function of stress. stress. and constant over temperature. 2 Dec 2009 MRS Fall Meeting, Boston p. 11/13 and increases over temperature.
Nucleation Stress(GPa)
Strain Rate Dependence of Nucleation Stress Zhu(2008) ThisStudy
4 3
ε = 108 2
The discrepancy becomes larger at lower strain rate!
ε = 10 −3
1 0 0
200 400 Temperature(K)
2 Dec 2009 MRS Fall Meeting, Boston
Our results suggest much higher nucleation stress (i.e. yield stress) at high temperature!
600
Experimental Test is needed!
p. 12/13
Conclusion • Free energy barrier and Nucleation Rate for dislocation nucleation at finite temperature is obtained from FFS. • We predict (1) activation volume : constant (or increasing) function of T (2) entropy : increasing function of T constant (or increasing) function of σ • Our results suggest higher nucleation stress at high temperature. 2 Dec 2009 MRS Fall Meeting, Boston
p. 13/13
Free Energy from FFS Stationary Distribution
Backward Flux Sampling Required!
PA IAB = PB IBA
PA : probability being in A ФA : Flux from A boundary Τ+(q;λ0) : AVG time spent on q π+(q;λi) : histogram of q from a trajectory from λi
04/16/09 MRS March Meeting, San Francisco
p. 14/11
Configuration from FFS
700K A lot of fluctuations
2 Dec 2009 MRS Fall Meeting, Boston
400K Small fluctuations
p. 12/12
Anderson Thermostat κ ⋅ ∆T ⋅ V 1/ 3
Heat Gain Per Unit Time ~
Rate of Energy Gain by collision with frequency ν ~
2 Dec 2009 MRS Fall Meeting, Boston
3 Nνk B ∆T 2
p. 12/12