Radiation damage in fission and fusion reactor materials

1
Radiation damage in fission and fusion reactor
materials

• Kai Nordlund
• Accelerator laboratory
• University of Helsinki

2
The guilty parties

• University of Helsinki
• Kai Nordlund, Emppu Salonen, Krister Henriksson,
  Petra Träskelin, Niklas Juslin, Juhani Keinonen
• University of Illinois
• Mai Ghaly, Bob Averback, Pascal Bellon
• MPI für Plasmaphysik, Garching
• Chung Wu
• University of Liverpool
• Fei Gao

3
Contents

• Overview of types of primary state of reactor
  damage
• Fission reactors
• Fusion reactors
• Our approach to studying it MD
• What is MD?
• Capabilities and limitations of MD today
• Brief history of MD for irradiation effects
• Direct production of dislocation loops in bulk
  cascades
• Surface effects during heavy ion bombardment

4
Radiation damage in fission reactors

• Neutron-induced damage in structural materials
• Practically purely bulk damage
• Primary state of damage caused by a few MeV
  neutrons which give a recoil energy of 50 100
  keV to sample atoms
• These recoil atoms then cause damage
• The primary damage becomes on long time scales
  dislocation loops and voids, which degrade the
  mechanical properties of the material

5
Radiation damage in fusion reactors

• Neutron-induced damage in structural materials
• As in fission reactors, but recoil atoms are
  produced by 14 MeV neutrons and have a broader
  spectrum
• He produced in reactions
• Damage at the reactor first wall
• Ions escape from the hot plasma and erode the
  wall of the reactor

6
Our activities

• In my group we study
• 1. Irradiation effects in semiconductors
• 2. Ion, neutron and plasma modification of
  metals
• 3. Plasma-wall interactions in fusion reactors
• 4. Nanocluster interactions with surfaces
• 5. Carbon nanotubes
• Main working tool Molecular dynamics computer
  simulations
• Also now DFT and KMC

7
What is molecular dynamics?

• Central idea simulate atom motion
• Solve Newtons equations of motion numerically
• Forces derived from classical or quantum
  mechanical models

8
The MD algorithm, slightly simplified
9
Force calculation

• All other steps except the force calculation can
  usually be made arbitrarily accurate
• Forces between atoms can be obtained from
• quantum mechanical models
• classical interatomic potentials
• Quantum mechanical models
• Very slow, number of atoms usually limited to a
  few hundred
• Good for studying small defects
• But mostly too slow to study irradiation effects

10
Capabilities classical models

• Number of atoms 5 million
• System size 40 x 40 x 40 nm
• Time scale 1 ns
• Materials that can be modelled (for irradiation
  effects)
• Almost all commonly available pure elements
• Some common binary compounds and alloys
• In my group we have simulated irradiation effects
  in
• FCC Al, Cu, Ag, Au, Ni, Pd, Pt, Pb, CuAu, CoCu,
  NiCu,
• BCC Mo, W, Fe, FeCr
• HCP Co
• Other C, Si, Ge, C nanotubes, Ga, a-CH, SiGe,
  GaAs, GaN,

11
Limitations

• Number of atoms 5 million
• System size 40 x 40 x 40 nm
• Time scale 1 ns
• But these are actually not so serious for
  irradiation effects
• Main limitation interatomic potential
  availability
• If there is no suitable potential available for
  your material of interest, what can you do?
• Create a potential - but needed effort 0.3 - 3
  person-years
• Use some model systems expected to behave
  qualitatively like your real system

12
Collision cascade simulations

• Principle initiate ion movement in or outside
  simulation cell with enough atoms, follow what
  happens until system has cooled down
• Heat removed at cell borders
• Electronic stopping included

13
Example surface event

• 30 keV Xe -gt Au

14
Example bulk event

• 10 keV recoil in Cu3Au

15
History first MD of heat spike

• In 1987 Diaz de la Rubia, Averback, Benedek and
  King showed using MD that heat spikes in metals
  behave much like the prediction in 1954 by
  Brinkman

16
History direct dislocation production

• In 1991 Diaz de la Rubia and Guinan showed that
  small dislocations can be directly produced in
  collision cascades

17
History phase changes

• In 1995 Diaz de la Rubia and Gilmer showed that
  irradiation in semiconductors can directly induce
  amorphization

18
Historysurface effects

• Between 1994 and 1999 we have shown how a nearby
  surface can alter the damage production picture
  from that predicted by bulk models

19
Contents revisited

• Overview of types of primary state of reactor
  damage
• Fission reactors
• Fusion reactors
• Our apprach to studying it MD
• What is MD?
• Capabilities and limitations of MD today
• Brief history of MD for irradiation effects
• Direct production of dislocation loops in bulk
  cascades
• Surface effects during heavy ion bombardment

20
Vacancy cluster production in metals

• Vacancy clusters are typically produced in the
  center of the cascade
• The vacancies are actually pushed inwards by the
  advancing resolidification front and cluster there

21
Interstitial cluster production in metals

• Interstitial clusters can also be produced
  directly in a cascade due to the
  recrystallization
• "Liquid isolation mechanism"

Nordlund et al, PRB 57 (1998) 7556
22
Vacancy cluster shape?

• The vacancy clusters typically are disordered,
  but already directly close to the shape of an
  dislocation loop
• With some annealing (even room temperature
  enough) these become well-ordered dislocation
  loops

23
Vacancy clusters SFT's

• We have also observed the formation of perfect
  stacking fault tetrahedra in collision cascades,
  at 0 K ambient T!

24
Vacancy clusters SFT's

• More commonly a cascade produces an SFT-like
  shape which after minor annealing becomes a
  perfect SFT

Initial cluster shape
After 1 ns at 800 K
Nordlund and Gao, APL 74 (1999) 2720
25
Interstitial cluster shape?

• Also interstitial clusters can be directly in the
  shape of interstitial dislocation loops
• These will then subsequently affect the
  mechanical properties of the wall material

26
Irradiation of metal surfaces

• But what happens when a surface is present?

27
Results crater formation

• Formation of a crater when a 50 keV Xe ion hits a
  Au surface

Nordlund, Physics World 14, 3 (2001) 22
28
Comparison to experiment

• Final crater shape

Simulated TEM image
Experimental TEM image
Donnelly, PRB 56 (1997) 13599
29
Comparison to experiment

• Crater radius as a function of energy

30
Damage inside material

• The liquid formed by the incoming ion can
  extend very deep into thematerial

31
Damage inside material

• Thus there can be large defective crystal zones
  not only at the surface but also very deep in the
  sample

Nordlund et al, Nature 398 (1999) 6722
32
Difference in damage production

• Comparison of damage production in bulk and
  surface events

33
Conclusions on metal irradiation

• The surface can have a huge effect on damage
  production even at energies of the order of 100
  keV!

34
How about lower energies?

• We saw that heavy ions or neutrals with energies
  of the order of 50 keV produce really massive
  erosion of metals
• But you plasma physics people will take care that
  such ions will never hit the fusion reactor first
  wall at such high energies
• The limit for possible run-off erosion is when Y
  gt 1
• For realistic divertor materials, what is the
  energy limit for that?
• And could atom clusters enhance the effect?

35
Cluster sputtering in W?

• We studied both self-ion and self-cluster
  sputtering in Mo or W
• Our result
• Cluster sputtering enhancements can occur in Mo
  and W
• But if the incoming energies are lt 2 keV the
  yields are lt 1 gt no run-off effect

J. Nucl. Mater. 15 (2003) 5845
36
Lighter ions?

• But the main erosion issue for the plasma-wall
  community is of course the possible erosion by
  low-energy light ions (H, He) at the divertor
• There are several important ways in which these
  can cause erosion
• Physical sputtering of Be, C and W
• Chemical sputtering of Be and C
• Chemical sputtering of mixed materials (e.g. WC)
• Bubble formation and blistering
• Only the first of these is well understood!
• The following three talks will discuss the latter
  three issues!

37
Hydrocarbon erosion from divertors

• Consider a fusion reactor

• Zoom in on divertor

• And finally on the atomic level

38
Problem why does carbon erode?

• According to physical sputtering theory the
  erosion should co down to exactly zero when the H
  energy is about 40 eV
• Experiments show something else
• Why??

Theory prediction
39
Method to study question

• We have used MD simulations to study this issue
• We manufacture and amorphous CH cell which
  corresponds to the materials in the reactor
• We bombard it with hydrogen with fusion
  reactor-relevant energies

a-CH (H/C 0.4) 1000 atoms
40
Solution to problem

• New erosion mechanism
• swift chemical sputtering)

41

• The H ion hits the middle of a C-C chemical bond.
  This raises the energy enough to break the bond

Salonen et al, Europhys. Lett. 52 (2000) 504
42
Comparison to experiments

• Explains the observed erosion at energies gt 10 eV
• Predicts that erosion goes to zero at about 2 eV.

43
Analysis of erosion products

• We can predict the relative abundance of the
  erosion products

44
Temperature dependence

• Erosion maximum at about 900 eV
• Same observed in experiments
• Explanation
• Rise decrease in number of C-C bonds
• Decrease dynamical rebonding

45
Good news (1)

• What happens after prolonged bombardment

Erosion yield 0.01
46
Good news (2)

• By doping the a-CH with Si one can also reduce
  the sputtering yield

47
Conclusions

• Molecular dynamics simulations can be useful for
  understanding lots of different radiation damage
  issues in reactor materials