Post on 27-Mar-2015
Molecular Dynamics modelling of mixed layer formation
K. Nordlund, C. Björkas, N. Juslin, K. Vörtler
Accelerator Laboratory, University of Helsinki
P. Erhart
Lawrence Livermore National Laboratory, Livermore, USA
K. Henriksson
Kungliga Tekniska Högskolan, Stockholm, Sweden
Carolina Björkas and Kai Nordlund 20072
Outline
Motivation for MD simulation of mixed materials What is needed to model mixed materials by
molecular dynamics? Current status of BeWCH potential development
Simulations of mixed material formation and erosion WC formation from melt
WC structural formation and erosion by D ions
Outlook/workplan WC cobombardment
BeC
BeW
Carolina Björkas and Kai Nordlund 20073
Motivation
Central phenomena for mixed materials in ITER are erosion,
sputtering and redeposition All of these originate in atomic level processes
To attempt to understand these theoretically it is natural to use
atomic-level methods
In this context 4 main levels of methods can be considered
relevant: Binary collision approximation (BCA):
- Ideal for purely ballistic effects, e.g. linear cascade sputtering
Quantum chemistry (in practice mainly DFT methods):
- Individual chemical reactions for a few tens of atoms
Molecular dynamics (MD):
- Long time scale processes for thousands of atoms
Kinetic Monte Carlo (KMC)
- Diffusion of impurities in surfaces and bulk
Carolina Björkas and Kai Nordlund 20074
What is needed for MD
We do all of these levels of simulations, but in the
remainder of this talk I will concentrate on MD MD simulations always require the existence of an
interatomic potential for the system to be modelled A potential must describe the relevant interactions at
least qualitatively correctly For fusion reactor issues the potentials must in addition
be fully reactive in the sense that they should be able to
describe all coordination states and chemical bond
breaking => molecular mechanics models are out
Until recently, reactive MD potentials existed only for
elemental and binary systems Clearly not enough for the ITER BeWCH materials mix…
Carolina Björkas and Kai Nordlund 20075
Examples of past MD plasma-wall work
“Swift chemical sputtering”
mechanism which explains
the low-temperature
chemical erosion of C in
divertors Now much additional work
by LLNL, ORNL, FOM on
this; good agreement on
main points
H and He interactions with
W surfaces: big difference
between H and He
clustering explained
[Salonen, EPL 52 (2000) 50; PRB 63 (2000) 195415 ]
[Henriksson et al, Nucl. Instr. Meth. B 244 (2005) 377]
Carolina Björkas and Kai Nordlund 20076
Status of availability of divertor-relevant potentials: overview
H He C W Be
H
He
C
W
Be
2002, done by others
2006, CarWMod
[Juslin et al, J. Appl. Phys. 98, 123520 (2005)]
BeTunCMod aim 2008
Carolina Björkas and Kai Nordlund 20077
What is needed for a new potential
We use a 3-stage fitting procedure, which has been
proven to work well in a wide range of complex systems e.g. Si, C, SiC, WCH, Pt, Zn, ZnO, GaAs, GaN, Fe, ...
Our fitting stages are, for each interaction pair:
1. Obtain data to be fit: obtained from
experiments and DFT calculations,
relevant to problems of interest
2. Fit a Tersoff-Brenner-like potential
into the data
3. Test important non-fitted properties, iterate
back to stage 2 as long as necessary…
Expected effort needed for success/interaction pair: 2-12
person-months depending on complexity
Carolina Björkas and Kai Nordlund 20078
Status of availability of divertor-relevant potentials: status of today
H He C W Be
H
He
C
W
Be
2002
2006, CarWMod
1. DFT
2. Pot. fit
Almost all DFT data for fitting calculated, some more
for Be-H and Be-He coming from Alain Allouche Be-Be has good fit, Be-C fit still to be improved
Carolina Björkas and Kai Nordlund 20079
WC formation from melt
One way to form WC mixed layers is by cooling from a
melt, ‘quenching’ Not necessarily same structure as that formed during
irradiation
But good as ‘ground state’
comparison point for
irradiated layers
- E.g. in Si same comparison
shows similar densities between
irradiated and quenched cells
We obtained reasonable WC amorphous layers WxC1-x, x = 0.5 – 0.9
considering e.g. density compared to crystals and melt
Carolina Björkas and Kai Nordlund 200710
D irradiation of WC
D irradiation of initially crystalline showed a clear
amorphization effect in the cells What is surprising is that the amorphization correlates
with D implantation depth and not deposited energy Not only ballistic collisions, but also chemistry needed
20 eV 50 eV
100 eV 200 eV
Original 10 eV
1000 eV 2000 eV
Structure of WC after 2000 D impacts
Carolina Björkas and Kai Nordlund 200711
D irradiation of WC
We observed (as expected) preferential sputtering of C →
increasing W surface concentration
Results for change of W surface concentration compared to
experiment:
Experiment at 300 eV D on WC:
63 ± 10 at.% /(1018 ions/cm2)
Simulation at 300 eV D on WC:
crystalline:
57 ± 7 at.% /(1018 ions/cm2)
amorphous:
110 ± 10 at.% /(1018 ions/cm2)
Excellent agreement considering
there are no adjusted parameters!
[Träskelin et al, Phys. Rev. B 75 (2007) 174113]
Carolina Björkas and Kai Nordlund 200712
Outlook
Potential construction: Complete Be-C-H system potential, test it
Make Be-W system potential, test total Be-W-C-H
potential
Make He-Be potential
WC mixed layer erosion/formation: Response of WC to mixed D + 10% He/Ne/Ar/C/W
bombardment
- Especially C/W interesting: deposition vs. erosion
- Details still to be determined
Carolina Björkas and Kai Nordlund 200713
Outlook
Formation and erosion of BeC and BeW mixed layers: Construction of quenched a-BeC and a-BeW layers and
simulation of their erosion by D
- Relatively fast
Simulation of formation of mixed layers by Be/C/W
deposition on Be or C or W
- Very slow simulations
- Comparable to results of Doerner/Linsmeier/Krug
presented earlier during this meeting
- Time scale a challenge, but high T might work
Details still to be determined
- Will be fixed based on discussion with EFDA, you,
and technical simulation limitations
Carolina Björkas and Kai Nordlund 200714
Conclusions
It is now possible to simulate on an atomic level mixed
layer formation and erosion in the ternary WCH system Plus any noble gas (He/Ne/Ar/…) with a pair potential
Results until now show no obvious major shortcomings =>
results likely to be qualitatively reliable
Potentials now under development for the quaternary
BeWCH system We are hopeful this will do at least as well as the WCH one
But this is one of the first fully reactive potentials ever made
for a quaternary system for any application area. Complexity
is high and I am sure there will be shortcomings
somewhere.
- But some potential is still better than no potential…
Carolina Björkas and Kai Nordlund 200715
Thank you for your attention!Thank you for your attention!