ICTP/CAS/IAEA School and Workshop
Plasma-Material Interaction in Fusion Devices
Hefei, China
Ion Radiation Albedo Effect: Influence of
Surface Roughness on Ion Retention and
Sputtering of Materials
Yong-gang Li and Zhi Zeng
Research Laboratory for Computational Materials Sciences,
Institute of Solid State Physics, Chinese Academy of Sciences
June 23rd, 2016
1
Outline
• Background
• Monte Carlo simulation of primary radiation damage
• Influence of surface roughness on ion retention and
sputtering of materials
• Summary
2
to years) scales be addressed simultaneously, and that extensive
physical processes across the plasma–surface-bulk materials interfaces be integrated. Figs. 1 and 2 illustrate phenomena that govern
the response of the materials surface to plasma exposure [9], and
the computational models that must be accurately integrated.
While vastly different length scales characterize the surface
($nm) and plasma processes ($mm) as indicated in Fig. 1, the
plasma and the material’s surface are strongly coupled to each
other, mediated by an electrostatic and magnetic sheath, through
the nearly continuous exchange and recycling of incident ion and
neutral species and the re-deposition of eroded particles. These
interactions are more explicitly shown in Fig. 2, along with the corresponding time scales upon which they occur. These physical processes occur over a disparate range of time scales, which poses a
challenge both to modeling, and experimental characterization of
both the individual and coupled processes. As one example, the high
probability (>90%) of prompt local ionization and re-deposition of
sputtered material atoms means that the surface material that is
in contact with the plasma is itself a plasma-deposited surface, as
opposed to the original well-ordered surface of the material that
existed at the beginning of operation [9]. Likewise, the recycling
of hydrogen plasma (fuel) is self-regulated through processes
involving near-surface diffusion, trapping, and gas bubble formation, coupled to the ionization that results from interactions with
the plasma. The multitude of time and length scales controlling
material evolution and device performance requires the development not only of detailed physics models and computational
Background
the implantation depth is generally only a few nanome
more implanted particles accumulate within the surfac
eventually a steady-state condition can result, in which t
of species implanted into the materials is balanced by that r
from the material. The extent to which both surface morp
and sub-surface defect creation and evolution processes
by neutron-induced damage influence the diffusion, trapp
precipitation of hydrogen and helium species into gas bu
an outstanding question that impacts the tritium perm
retention and near-surface saturation levels.
Tungsten has recently been selected as the sole diverto
rial in ITER [10,11], and is the leading candidate material fo
and future fusion reactors. Laboratory experiments perfor
linear plasma devices indicate the possibility of substantial
modification in tungsten exposed to low-energy, helium pla
mixed helium–hydrogen plasma, although the observed
response is strongly temperature-dependent and likely dep
on the ion energy and flux. Pitted surfaces are observed
%1000 K [12], whereas a ‘‘nanostructured,’’ low-density ‘‘f
‘‘coral’’ surface morphology is observed between approx
1000 and 2000 K [13–16], while micron-sized holes, or p
observed to form above about 2000 K [17,18]. The nanostr
‘‘fuzz’’ has recently been observed in the divertor regions of
mak device operating with a helium plasma as well [19
surface features could lead to changes in heat transf
(deuterium/tritium) retention [20], increased rates of
through both sputtering and dust formation [21
• Plasma-material interactions (PMI) in nuclear fusion devices:
Cause the surface reconstruction of plasma-facing materials (PFMs, W) to
roughness or even more complex nanostructures (mounds, fuzz, bubbles, pores
and blisters) & ion (D/T/He) retention and sputtering of PFMs & degradation of
structural materials.
ITER – PFMS (Be, W)
D+, T+, He+, n
150 000 000 oC
Fig. 1. Schematic illustration of the synergistic plasma surface interaction processes that dictate material evolution and performance in the magnetic fusio
environment, as reproduced from [9].
B.D. Wirth et al., J. Nucl. Mater. 463 (2015) 30.
Please cite this article in press as: B.D. Wirth et al., J. Nucl. Mater. (2014), />
3
• Surface morphology of W under ion irradiation
• Loop punching and bubble rupture causing
surface roughening: mounds, fuzz, bubbles,
pores and blisters.
• What is the influence of surface roughness
on ion retention and sputtering of materials
under energetic ion irradiation?
~ t
Fuzz
S. Kajita et al., Nucl. Fusion 49 (2009) 095005.
He-W
Loop punching & bubble rupture
O. El-Atwani et al, Nucl. Fusion 54 (2014) 083013.
F. Sefta et al, Nucl. Fusion 53 (2013) 073015.
4
• How does surface roughness enhance ion retention and
reduce ion sputtering?
W
D. Nishijima et al. J. Nucl.
Mater. 415 (2011) S96.
Roughness
W
fuzz
Ar
RNRA
He
Surface roughness: polishing process
C. González et al., Nucl. Fusion 55 (2015) 113009.
Around one order of magnitude
below the expected sputtering yields.
I. Tanyeli et al., Sci. Rep. 5 (2015) 9779.
5
Monte Carlo simulation of primary radiation damage
• Primary radiation damage: Ballistic phase, in the range of ~ nm and the timescale of ~ sub-ps; two types of collision – binary & cascade/spike collision.
• Until now radiation damage simulation codes (like SRIM) have been limited
in ability to describe 3D geometry, computational efficiency, or both.
Advantages: MC v.s. MD
BCA
Binary collision
• Simple and high efficiency;
• Arbitrary 1D/3D structures;
• Accounting of electronic energy loss and
multiple- and plural-scattering;
Cascade/spike collision
• No limitations in nanostructure sizes, ion
energies, or availability of empirical interatomic potentials.
6
• IM3D: Primary radiation damage under ion irradiation
IM3D: A 3D Parallel MC Code for Efficient Simulation of Primary Radiation Damage
(0, 0, 0)
y
x
Ions
z = zin
z
z = zsub
Standard SRIM
database
Substrate
Constructive Solid(x , y , z )
Geometry (CSG)
t0
(0, 0, 0)
x
y
t0
t0
Ions
z = zin
z
+
Fast database
indexing technique
(xb1, yb1, zb1)
z = zsub
dpa
MPI parallel
(xb0, yb0, zb0)
nm
Substrate
50 keV Si ! GaAs
(xt0, yt0, zt0)
Finite Element Triangular
Mesh (FETM)
Arbitrarily complex
3D structures
C & MPI
Efficiency ~ at least 2 orders higher
Y.G. Li et al., Sci. Rep. 5 (2015) 18130.
As accurate as SRIM
More efficient and universal
MIT
7
• IM3D: Arbitrarily complex targets based on CSG/FETM methods
500nm
H
-> W 1 MeV He -> Ni
0
(a)
10 keV He ions, Si
(b)
dpa
200
Fe
z (nm)
400
600
Cu
800
CSG -dpa
0
He ion
100 nm
Bulk - Spatial correlation
1000
(a)
FETM - ion
(c)
100
CSG - dpa
(b)
Ga ions
FETM - Vs
NV - N
200
z (nm)
300
He ion
400
500
600
700
He ions, 100 keV
Total
50 keV, x 1
100 nm
100 keV, x 3.2
150 keV, x 4
200 keV, x 8
Ga -> W
NiP
130 nm
FETM - dpa
W
8
• Random rough surface model
• FETM - Gaussian distribution
f ( Z ) ∝ exp ( − Z 2 2σ 2 ) , Z ∈[ −3σ , 3σ ]
• Square mesh – a (50 nm)
100 eV D -> W
9
Influence of surface roughness on ion retention
& sputtering of materials
• Two Key factors:
Smooth surface – Incident angle
Roughness 3σ & Incident angle θ
Rough surface
– Incident angle
10
• The ion radiation albedo effect
• Both primary ion backscattering (1retention) and sputtering yields decrease
with increasing roughness, and increase
with more oblique irradiation angles.
• It is mainly dominated by the direct, lineof-sight deposition of a fraction of emitted
atoms onto neighboring asperities.
W
Nishijima et al., J. Nucl.
Mater. 415 (2011) S96.
W
Rough peaks
11
• Ion retention and sputtering of W with roughness surface
• Primary ion retention rate:
R2 = 1 − η (α ) + R20 ⋅ η (α ) ⋅ Ps
Backsca+ering Nano-geometric
/Shadingeffect
effect
• Sputtering yield:
Y = A (α ) ⋅Y0 (α ) ⋅ (1− Ps )
Nano-geometric/Shadingeffect
12
Summary
• A new, sophisticated 3D Monte Carlo code (IM3D) and a random rough
surface model have been developed.
• Both primary ion backscattering and sputtering yields decrease with
increasing roughness, which is mainly dominated by the direct, line-of-sight
deposition of a fraction of emitted atoms onto neighboring asperities.
• Backscattering and sputtering increase with more oblique irradiation angles.
• A simple analytical model is proposed to relate rough-surface and smoothsurface results.
• There could be an additional positive feedback mechanism to promote the
dendritic growth of surface asperities besides the loop-punching and bubble
rupture mechanisms.
Ions
Outgoing atoms
Atom re-deposition
13
Acknowledgements
• Institute of Solid State Physics, CAS, China
Cuan-guo Zhang, Liang Hu, Zhe Zhao, Gu-yue Pan, Pan-fei Tang
• Massachusetts Institute of Technology, USA
Ju Li, Michal P. Short, Yang Yang
• University of Sciences and Technology of China, China
Ze-jun Ding, Shi-feng Mao
Thanks for your attention!
14
• Comparison of IM3D and SRIM
106
Serial: 2-3 orders higher
70
FC
KP
103
102
40
30
20
Parallel: ~ 80%
10
101
0
100
Order-N scaling
IM3D scaling
50
Slow
Fast
104
105,2 MeV,Au->ZrO2/Si
60
Speepup
CPU Time (s)
105
Li et al., Sci. Rep. 5 (2015) 18130.
SRIM
Iradina
0
8
16
24
32
40
48
56
64
IM3D
Processors
So9ware
SRIM
IM3D />
Scattering angles
MAGIC approximation
Fast database indexing
Geometries
1D bulk or multi-layers
Arbitrarily 3D geometries
Computational
Efficiency
Serial, low
2-3 orders faster for serial
version, MPI parallel (> 80%)
Defect distributions 1D depth-distributions
> 700 citations per year
3D space-distributions, spatial
correlation
More efficient and general
15
• Validation and Verification of IM3D
• IM3D vs. SRIM for bulk
Borschel et al., Nucl. Inst. Meth. Phys. Res. B 269 (2011) 2133.
Stoller et al., Nucl. Inst. Meth. Phys. Res. B 310 (2013) 75.
• Ion depth-distributions under ion implantation with different energies
• V depth-distribution predicted by full-cascade and Kinchin-Pease models
FC : KP ~ 2
16
Z
L ∝a
W
Zp
Profile
element
α
Z=0
Zv
ΔZ
17
• Nano-energetic and nano-geometric effects
Ren et al., Phys. Rev. B 86 (2012) 104114.
Ar ion
~ 50 %
Modified NRT model:
⎞
⎛ 2Sb
⎞
N D ( R ) Ed ( ∞ ) N d ( R ) ⎛
1
1
−R t
⎡
⎤⎦
=
⋅
=
1+
exp
⋅
1−
1+
R
t
⋅
e
(
)
0
0
⎣
⎜⎝
⎜⎝ 3R 4R h − 1 ⎟⎠
ND
Ed ( R ) N d
4R h − 2 ⎟⎠
c
Nano-energetic effect ~ 20 nm Nano-geometric effect
-- Volume
Vanithakumari et al., Phys. Lett. A 372 (2008) 6930; Ouyang et al., Nanotech. 19 (2008) 045709.
18
et al. sputtering induced the bending of W nanowire
• IonCui
beam
213112-2
Appl. Phys
W
bending
W
fuzz
Cui et al., Appl. Phys. Lett. 102 (2013) 213112.
(a)
10 * 10 * 10 nm3
(b)
FETM
(c)
Ga ions
2402
NV - NI
2400
100 nm
W
-110
-111.5
19
Y. Ueda et al. J. Nucl. Mater. 442 (2013) S267;
S. Kajita et al., Nucl. Fusion 49 (2009) 095005.
20