growth of helium bubbles in tungsten under realistic rates
TRANSCRIPT
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
U N C L A S S I F I E DLA-UR-14-27989
Growth of Helium Bubbles in Tungsten underRealistic Rates
November 6, 2014
Luis Sandoval, Danny Perez, Blas P. Uberuagaand Arthur F. Voter
1
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
U N C L A S S I F I E DLA-UR-14-27989
Motivation: Helium induced morphology in tungsten
300 s 2000 s 4300 s 9000 s 22000 s
Cross-sectional SEM images of W targets exposed to He plasma. T = 1120 K,ΓHe+ ∼ 5× 1022m2s−1, 〈Eions〉 ∼ 60 eV. 1
1Baldwin, M. J. and Doerner, R. P. Nucl. Fusion 48, 035001 (2008).
2
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
U N C L A S S I F I E DLA-UR-14-27989
Motivation: time limitations of direct MD simulations
t = 5 ns t = 10 ns
t = 15 ns
t = 25 ns t = 30 ns
t = 20 ns
• Impact of He atoms on W at a rate of0.2 He ps−1. The kinetic energy per He atomis 60 eV. The interactions are determined bya short-range-modified Ackland-Thetfordpotential (Juslin and Wirth, 2013).
• For the simulation box used, this impact ratecorresponds to a flux of 5× 1027 He m s−1
(∼ 4 orders of magnitude higher than theone expected in the ITER divertor).
• Only ∼ 2.5 % of the incoming ionscontribute to the growth process of the deepbubble, that is, its growth rate is∼ 5 He ns−1.
• A study at slower impact (and growth)rates, comparable to experiments, is needed.
3
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
U N C L A S S I F I E DLA-UR-14-27989
Single He Bubble Growth in a perfect W lattice
Langevin thermostatat 1000 K
Initial configuration:1 tungsten vacancywith 8 helium atoms
d = 1.9 nm
NVT
NVE
At a given rate anew helium atomis inserted insidethe buble
Simulation setup. Pressure in the helium bubble. a
aSefta, F. et al. Nuclear Fusion 53, 073015 (2013).
4
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
U N C L A S S I F I E DLA-UR-14-27989
Parallel Replica Dynamics2 (ParRep)
True infrequent events have an exponential first-passage time distribution:
p(t) = k exp(−kt) . (1)
We can exploit properties of exponential to parallelize time, by having manyprocessors seek the first escape event:
dephasing correlated events
parallel time
Arbitrary accurate dynamics if implemented carefully.2Voter, A. F. Phys. Rev. B 57, 13985 (1998).
5
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
U N C L A S S I F I E DLA-UR-14-27989
Standard Parallel Replica Dynamics
• Standard Parallel Replica Dynamics relies on the separation of timescalebetween vibrations and transitions between basins.
• A state is taken to be the ensemble of points of configuration space thatconverge to the same fixed point under a local minimization of the energyof the system.
• This definition limits the range of possible applications to systems wherethe basins are deep enough and well separated from each other.
6
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
U N C L A S S I F I E DLA-UR-14-27989
Superstate Parallel Replica Dynamics
• A superstate is defined as all the points of configuration space that sharethe same values/range of suitably-defined slow variables of the system, sothat equilibration within the state is much faster then escape from thestate.
The definition of states can be optimized by lumping individual shallowstates into superstates.
7
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
U N C L A S S I F I E DLA-UR-14-27989
Superstate Parallel Replica Dynamics
• In our study, ParRep transitions are defined as changes in atomicpositions where at least one tungsten atom has moved a distance greaterthan 0.25 nm, slightly lower than 〈111〉/2, the Burgers vector ofprismatic 〈111〉 dislocation loops. Concerning these transitions, themotion of He atoms is ignored.
Frenkel pairnucleation
Interstitial diffusion Vacancy diffusion Adatom diffusion
8
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
U N C L A S S I F I E DLA-UR-14-27989
Growth rate effects
Point obtained by using 10.000 replicas(160.000 cores ~ 53% of Titan)
ParRep MD
Average He content in the bubble at thetime of the first detected event.
10 15 20 25 30 35He atoms
45
50
55
60
Pre
ssure
(G
Pa)
1012 He s−1
1011 He s−1
1010 He s−1
109 He s−1
108 He s−1
b)
Average pressure in the He bubble vs Hecontent and growth rate.
9
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
U N C L A S S I F I E DLA-UR-14-27989
Growth rate effects
0 20 40 60 80 100 120He atoms
20.019.519.018.518.017.517.016.516.0
Avera
ge p
osi
tion
of
cente
r of
mass
()
1012 He s−1
1011 He s−1
1010 He s−1
109 He s−1
108 He s−1c)
Average position of the center of mass ofthe He bubble.
0 20 40 60 80 100 120He atoms
22.5
22.0
21.5
21.0
20.5
20.0
19.5
Avera
ge p
osi
tion o
f t
he low
est
He a
tom
()
1012 He s−1
1011 He s−1
1010 He s−1
109 He s−1
108 He s−1
d)
Average position of the lowest/deepest Heatom.
10
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
U N C L A S S I F I E DLA-UR-14-27989
Growth rate effects
108 109 1010 1011 1012
Growth rate (He s−1 )
200
220
240
260
280
300
Bubble
conte
nt
(He a
tom
s)
at
the b
urs
ting p
oin
tParRep ParRep
ParRep
ParRep
MDMD
MD
e)
Mean value of the He content in the bubble at thebursting point as a function of growth rate.
11
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
U N C L A S S I F I E DLA-UR-14-27989
Interstitial diffusion and adatom formation
a)
Diffusion hopping time as a function ofthe number of He atoms per W vacancyfor two cases: one W interstitial on thesurface of a 2-vacancy bubble, and one Winterstitial on a 15-vacancy bubble.
b)
Snapshots showing the diffusion of a W in-terstitial to the top of the bubble, the sub-sequent nucleation of additional Frenkelpairs, and the tearing off process ofadatom nucleation. Blue: W vacancies;red: W interstitials; green: adatoms.
12
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
U N C L A S S I F I E DLA-UR-14-27989
Interstitial diffusion and bubble growth
Initial
Final vacancy configuration c)
1 2,3 4 5 6 7 8
Final vacancy configuration for the 15-vacancy bubble (when adatoms are formed)from 8 independent ParRep simulations. The orange spheres denote the vacanciesnucleated in the tearing off process.
13
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
U N C L A S S I F I E DLA-UR-14-27989
Interstitial diffusion and bubble growth
107 108 109 1010 1011 1012
Growth rate (He s−1 )
0.00.51.01.52.02.53.03.54.0
Next
Frenke
l pair
a)
rc =3.2
rc =4.8
Average number of interstitial clustersaround the He bubble after a new Frenkelpair is nucleated, when at least one inter-stitial is already present, as a function ofthe growth rate.
107 108 109 1010 1011 1012
Growth rate (He s−1 )
0.00.51.01.52.02.53.03.54.0
Frenke
l pair
per
event
b)
Average number of Frenkel pairs per eventas detected by ParRep as a function of thegrowth rate.
14
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
U N C L A S S I F I E DLA-UR-14-27989
Interstitial diffusion and bubble growth
108 109 1010 1011 1012
Growth rate (He s−1 )
0
20
40
60
80
100
Loca
tion o
f v
aca
nci
es
(%) top
side
bottom
c)
Spatial probability for the nucleation ofnew vacancies with respect to the centerof the current vacancy cluster (see d)), asa function of the growth rate.
Norm
aliz
ed
his
tog
ram
dy
new vacancy
top
side
bottom
current vacancycluster (bubble)
d)
Histogram of the location of new vacan-cies in the direction perpendicular to thesurface for all the simulations.
15
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
U N C L A S S I F I E DLA-UR-14-27989
Conclusions
• For the slowest growth rates we considered, the system is able toefficiently explore the accessible state space, facilitating the occurrence oftransitions involving fewer W atoms.
• Significant differences across time scales are observed, which include thepressure experienced by the He bubble, the number of W vacancies andHe atoms in the bubble at bursting point, and the dynamics of Frenkelpair nucleation.
• Our main finding is the existence of two growth regimes, depending onwhether the growth of the bubbles occur slower or faster than thediffusion of interstitials around it.
• These findings highlight the importance of simulating materials underrealistic conditions and the potential pitfalls of extrapolating from shorttimescale simulations alone.
16
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
U N C L A S S I F I E DLA-UR-14-27989
Acknowledgements
The authors would like to thank Brian Wirth and Chun-Yaung (Albert) Lu forthe useful discussion. L.S., D.P., and B.P.U. ackowledge support by theU.S.DOE, Office of Science, Office of Fusion Energy Sciences and Office ofAdvanced Scientific Computing Research through the Scientific Discoverythrough Advanced Computing (SciDAC) project on Plasma-SurfaceInteractions. A.F.V. was supported by the U.S. U.S.DOE, Office of BasicEnergy Sciences, Materials Sciences and Engineering Division. This researchused resources of the National Energy Research Scientific Computing Center,which is supported by the Office of Science of the U.S.DOE under ContractNo. DE-AC02-05CH11231, and resources of the Oak Ridge LeadershipComputing Facility at Oak Ridge National Laboratory, which is supported bythe Office of Science of the U.S.DOE under Contract DE-AC05-00OR22725.Los Alamos National Laboratory is operated by Los Alamos National Security,LLC, for the National Nuclear Security Administration of the U.S. DOE, undercontract DE-AC52-O6NA25396.
17