Introduction to Plasma-Surface Interactions

1
Introduction to Plasma-Surface Interactions

• Lecture 5
• Sputtering

2
Topics

• Physical sputtering
• Sputter yields
• Energy distribution of sputtered atoms
• Chemical sputtering
• Yields
• Flux dependence of yields
• Global Model
• Comparing effect of different materials

3
Physical sputtering

• This is the most common mechanism for bulk
  impurity atoms getting into the plasma.
• Sputtering occurs as a result of momentum
  transfer from an energetic incident ion to solid
  surfaces
• It is a well understood physical process and
  results agree well with calculations
• Calculations are normally made with the TRIM
  Monte Carlo code. Tabulation of data for a wide
  range of ions and targets energies are available

4
Energy threshold

• Because an atom leaving the surface has to
  overcome the surface binding energy Es there is a
  threshold energy ET for sputtering. This is
  given by
• Where

5
Sputter yields for Be, C and W by D and self ions
Note the increasing threshold energy with target
mass. Using D ions yield is same for Be, C and
W
W Eckstein PMI, Garching, Report PP9/8 (1993)
6
High energy sputtering

• The maximum in the yield and the decreas at high
  energies is due to the collision cascade
  occurring deeper and deeper in the solid.
• The surface atoms have less chance of receiving
  sufficient energy to be sputtered

7
Uncertainty in yields

• There is an variation in yields measured
  experimentally 2.
• This is not experimental error but genuine
  variations which depend on surface conditions
  which can affect the binding energy
• Examples are variation in the structure, surface
  roughness or impurity levels

8
Effect of incident angle

• The sputter yield increases as the angle (q)
  increase from normal (q 0)
• This is due to the increased probability of the
  incident ion being backscattered
• At energies lt300 eV the variation of yield with
  angle is small.
• This is the region of most interest in plasma
  physics (the sheath potential tends to make ions
  arrive at normal incidence)

9
Energy distribution of sputtered atoms

• The energy of the sputtered atoms is important
  because it determines how far they penetrate into
  the plasma
• It too has been well studied and is understood
  theoretically
• The most probable energy is 0.5 Es (2 to 5 eV)
• At higher energies the energy distribution has a
  tail going as E-2 with a cut-off at the incident
  ion energy

10
Sputtered atom energy distribution for C
Measured spectroscopically using doppler shift
and compared with Thomson model using B.E. 9.3
eV Bogen and Ruesbueldt JNM 179 (1992) 196
11
Chemical sputtering

• This only applies to C but because C is widely
  used it has received a lot of attention
• A typical reaction is
• 4 H C CH4
• Methane is the most common product but higher
  hydrocarbons are also produced eg C2H4, C3H6
• The details of the reactions are not well
  understood and there is no reliable theory

12
Chemical Sputtering of CIon energy dependence
10-1 10-2 10-3
Yield almost independent of energy
10 20 50
100 200 eV
Mech et al JNM 255 (1998) 153
13
Chemical sputtering of CSurface temperature
dependence
The behaviour is complex and not understood (to
my knowledge)
10-2 10-3 10-4
CD4 CH4
C2H4 C2D4
400 600 800
1000 K
Mech et al JNM 255 (1998) 153
14
Extrapolating to high flux conditions

• Over the last few years there has been much
  discussion about how the chem. sputt. yield
  varies with incident ion flux.
• Results from 7 different devices have been
  correlated and analyzed to obtain a consensus
  view (Roth et NF 44 (2004) L21)
• Results from this study are presented in the next
  slide

15
Extrapolating to high flux conditionsresults
from many experiments
Roth Nuclear Fusion 44 (2004) L21
16
Modelling global behaviout

• The operation of a plasma physics device is
  complicated because there are so many interacting
  processes.
• These are generally studied using large computer
  programs, often using fluid codes. It is
  difficult to see the importance of different
  processes.
• An attempt has been made to present a simpler
  analytical model. It is not expected to give
  accurate description of the systems but to try
  and see the relative importance of different
  processes

17
Carbon as a target material

• The reduction of the sputter yield at high flux
  compensates for the higher flux
• The concern over chem. sputt. is not as serious
  as originally thought.
• However the major concern in using carbon in a DT
  machine is th ehigh inventory built up in the
  deposited layers

18
Global model of sputtering
From particle balance of confinement times
and sputtering yields we can
get Where is a screening coefficient We
can get a measure of the edge Te from energy
balance Where PH,PR, and PC are the input,
radiated and conducted power. Although crude
this model allows us to see the difference in
behaviour of low and high Z materials
19
Calculation of radiated power and Zeff vs ne
Comparison of Be and W
Calculations based on the global model At low ne
Te is high and W is sputtered fast, resulting
in High Pr and Zeff Be has reached or is over the
maximum in the sputter yield. Becaause of low Z
and low density Pr and Zeff are low. At high ne
Te is low and W sputter rate is low or zero,
resulting in low Pr and Zeff Be sputter rate is
still high and ne is high so Pr is high G
McCracken and G Matthews JNM 176-177 (1990) 312
20
Choice of materials

• A figure of merit M was proposed by Lazlo and
  Eckstein (1991)
• Where fi is the maximum impurity concentration
  allowed in the plasma.
• The larger M the less power will be radiated
• Both the sputter yields and fi are functions of
  edge Te . M can be plotted against edge Te

21
Figure of merit for materialsas a function of
edge Te
At low enough Te all materials are good At high
Te Mo and W are useless For high edge Te only
very low Z materials are tolerable
High M is good low M is bad!
22
Health warning!

• Dont take these models too seriously, but they
  are worth thinking about, particularly in terms
  of comparing high and low Z materials
• The thresholds even for hig Z materials like W
  are not very high, especially when you take into
  account multiply charged ions

23
Summary - 1

• Physical sputtering is a real threat. Only by
  keeping the edge Te low can it be avoided
• At low density i.e. higher Te, only low Z
  materials stand any chance

24
Summary - 2

• Chemical sputtering is only a problem with
  carbon.
• Unlike phys. sputt. there is no good theoretical
  model and so it is difficult to include it
  general plasma codes
• Recent data of lower yields at high fluxes look
  helpful

25
Schematic of arc tracks
Because the arc tracks are driven by an JxB
force, for a fixed field, on a curved surface the
current changes direction and henc the force
changes This results in curved tracks Typical
patterns seen in tokamaks are shown Thhe tracks
go in the opposite direction to the JxB force.
There are at least 20 explanations for this
effect but none are very convincing!