Managed by UT-Battelle for the Department of Energy
Data Evaluation for AM and PMI Processes in Fusion, IAEA-NFRI, Sept. 2012, Daejeon, Korea
Predrag Krstić Joint Institute of Computational
Sciences, University of Tennessee
at Oak Ridge National Laboratory
& TheoretiK Consulting
Role of the Fusion Atomic
Databases in the Internet
Environment
OFES Review, ORNL, March 02, 2010 Managed by UT-Battelle for the Department of Energy Data Evaluation for AM and PMI Processes in Fusion, IAEA-NFRI, Sept. 2012, Daejeon, Korea
Model validation is usually defined to mean “substantiation that a computerized model within its domain of applicability possesses a satisfactory range of accuracy consistent with the intended application of the model.” (Schlesinger at al 1979). Comparison with experiment :: qualitative toward quantitative Calculation needs to mimic the experiment as close as possible
Model verification is often defined as “ensuring that the computer code of the computerized model and its implementation are correct”. Code testing against simple models; Overlap of different adjacent time and spatial scales by various methods
Uncertainty quantification science tries to determine how likely certain outcomes are if some aspects of the system are not exactly know. Here: Model parameters may vary between different instances of the same object for which predictions are sought. Example: Monte-Carlo approach to trajectories over the surface
OFES Review, ORNL, March 02, 2010 Managed by UT-Battelle for the Department of Energy Data Evaluation for AM and PMI Processes in Fusion, IAEA-NFRI, Sept. 2012, Daejeon, Korea
Guiding principle:
If Edison had a needle to find in a haystack, he would proceed at
once with the diligence of the bee to examine straw after straw
until he found the object of his search… I was a sorry witness of
such doings, knowing that a little theory and calculation would
have saved him 90% of his labor.
–Nikola Tesla, New York Times, October 19, 1931
1) The traditional trial-and-error approach to a large categories of atomic data for fusion (excited states, molecules,…) and especially PMI data by successively refitting the walls of toroidal plasma devices with different materials and component designs is becoming prohibitively slow and costly, programmatic problems.
3
LET US THINK!!!
4
Example :Astrophysical applications
CT in H+H2+ for formation of H2 in the early universe (0.1meV-10 eV);
Two-body association (hydrogen plasma) in collapse of interstellar clouds.
[a bad example from astrophysical
modeling community (can happen to
Fusion community too)
Savin et al, ApJL (2004)]
CT in H++H2(v=0)
Data “produced” as fitted the need
of a particular plasma-radiative model
These cannot be called scientific data!!!
However a critical evaluation and
recommendation can lead to the DATA!
Need for comprehensive, critically evaluated data;
Communication between various communities (theory, experimental, atomic,plasma
Also Because 2):
5
Electronically excited : *Huge increase of the cross sections (as n4 for CT)
*For a complete H/H2 CR model, Hα diagnostics,
*Fulcher-band diagnostics for H2.
Vibrationally excited: *Infrared emission plasma diagnostics.
*CR models of H2/D2 plasma.
*Lack of quantitative analysis in molecular spectr.
Rotationally : High rotational temperatures of H2 indicated?
Isotopic constitution : *D2,T2, HD, HT and DT, Sensitive on vib. energy levels.
and excitation *Wherever internal energy plays role
(“ion conversion”).
*No data for excited molecules.
*Ex.:σpex(D++H2→HD+H+) » 10 σpex(D
++HD→D2+H+).
WHAT IS NEEDED?
6
2 2 ,
2 2
2
,2 2
2 2
2
2
( ) ( ), 0 - 14
( ) (1 ) ( ), 0 - 14, 0 - 19
( ) , 0 - 14.
( ) ( ), 0 - 19
( ) ( ), 0 - 19, 0 - 14
( ) , 0 - 19.
,( )
i f i f
i f i f
i i
i f i f
i f i f
i i
f
H H H H
H H v H s H
H H v H H H
H H H H
H H v H H
H H v H H H
H H H H H
2
0 - 14.
, 0 - 19( )
f
ffH H H H H
WHAT IS NEEDED (for example)?
•Comprehensive data cross sections, if calculated then
on the “same footing”
•0.5-100 eV collision energy
EXC
CT
DISS
EXC
CT
DISS
ASSOC
ASSOC
+ENERGY&ANGULAR SPECTRA (DISS)
7
And: It is 21st century
For these kind of data (vib-rot-elec excit.,isotop.):
*Experiments difficult:
Impossible? Missing !
*Quality theoretical data: Sparse !
And because :
H++H2 is the most fundamental ion-molecule system
We should know all about it
Do not know well this only (3+2)-body system?
Electronically, rovibrationally excited processes????
H 3+
-> H + H 2+
H 3+
-> H+
+ H 2
=850
1
0
4
2
3
5
14 18
21
0
I II III V VIIV
-10.0
-8.0
-6.0
-4.0
-2.0
0.0
Wa
(eV
)
H 3+
: ADIABAT IC VIBRAT IONAL "TERMS"
0 1 2 3 4 5 6 7 8 9 10
R (a.u.)
II IV
3
=0
1
2
'=1
=5
HC HC
Demkov
=0
LZ
0.0 2.0 4.0 6.0 8.0 10.0
R (a.u.)
0
2
4
6
8
< |d
/dR
| '>
Dissociation
Charge transfer
=85 0
Physics in direct channel
Dissociative continuum discretized
Extensively rich
•We describe both electronic and nuclear motion quantum- mechanically •Solve resulting Schrödinger equation by expanding in diabatic vibrational basis •Several hundreds states to converge
Why is this so difficult?
Too many states and processes!!! Krstic, 2003
r R
U
Physics is here Reactive mechanism
H++H2 , direct channel reactive channel
Collinear configuration
Another reason:
Particle exchange
10
“Interplay” of transport and inelastic processes
Cro
ss S
ec
tio
n (
a.u
.)
10-1
100
101
102 H
++H 2(=0)
mtvib
dis
vi
=4
mt
vib
ct
dis
vi
ct
ECM (eV)
0 1 2 3 4 5 6 7 8
Cro
ss S
ec
tio
n (
a.u
.)
10-1
100
101
102
=7
mt
vib
ct
dis
vi
ECM (eV)
1 2 3 4 5 6 7 8
mt
vib
ctdis
vi
=12
a) b)
c) d)
H+H2+
ν=0
v=4
ν=7
v=12
We have this from only one QM calculation!
Example: How the same footing works! Krstic, 2003
9/4/2012 11
Verification of data in action: Calculations of various
energy scales by different methods and comparison
where they overlap
2 2( )
iH H H H
3
FQ
2
1
456
12
7
3
4
6,7
i=0
H+
+H 2( i) -> H(1s)+H 2+
i=0, rec. (Linder et al, 1985)
i=0, Holiday et al (1971)
i=0
5
TSH
(Ichihara et al, 2000)
100
101
102
CM Energy (eV)
Total charge transfer
10-17
10-16
10-15
10-14
Ch
arg
e t
ran
sfe
r c
ross s
ec
tio
n (
cm
2)
10
5,6,7
4
3
2
1
SClass (Krstic 04)
Fully QM
(Krstic 02)
10
14
b)
TSH (Ichihara et al, 2000)
Present
8
9
1011
12
13
H+
+H 2( i) -> H(1s)+H 2
+
Fig. 1b)
14
14
13
12
11
10
9
8
0 2 4 6 8 10
CM Energy (eV)
10-15
Ch
arg
e t
ran
sfe
r c
ross s
ec
tio
n (
cm
2)
Comparison with Ichihara is shown above for the charge transfer. As can be predicted, Ichihara does agree well with Krstic’s QM calculations , for quasi-resonant v>3, though it is likely overestimates toward the low energies even for v>3. However, Ichihara is not expected to be correct for lower v. Finally, when the probability of CT falls (like is a case of v=13, 14 or so, the classical Ichihara is again incorrect. Note comparison with Holiday’s experiment :Validation for v=1 only
12
10-6
10-5
10-4
10-3
10-2
10-1
100
101
10-5
10-4
10-3
10-2
10-1
100
CM
(rad)
a)
b)
c)
H ++H2
H+H 2
Herman et al. (1978)
0-1
d)
10 eV
1.26 eV
1 eV
1.58 eV
2
sin
d
el/d
(
a.u
.)
0-1
0-3
0-2
10 eV
ECM
=10 eV
10-1
100
101
102
10-1
100
101
102
10-1
100
101
102
DCS for excitation to the first three vibrationally states from the ground state.
More on comparisons
with sparse experiments.
Validation in action: Comparison with experiment
Vibrational excitation: Differential cross sections
Excellent agreement! Partially validated! Krstic, 1998
13
Ar ion on supersaturated a:C (100 eV)
Courtesy of S. Stuart (CMD)
How are we going to pretend
that we can ab initio simulate
collision dynamics in a mezo-size
system if we are not acquainted
with H3+???
BTW:Trivia question -
What will be one of the greatest
achievements of the 21st century
theoretical physics? An answer:
Excited-state MD! (prerequisite: Excited-state
computational chemistry)
HERE WE ARE: Here we need to be (at least):
Going to bigger systems:
We cannot! Validation and UQ is essential
OFES Review, ORNL, March 02, 2010 Managed by UT-Battelle for the Department of Energy Data Evaluation for AM and PMI Processes in Fusion, IAEA-NFRI, Sept. 2012, Daejeon, Korea
All energy from D-T fusion reactions passes through first wall
Vac.
Supercon–
ducting
magnet
Shield Blanket
Turbine
generator
Plasma
a
Plasma heating
(rf, microwave, . . .)
Schematic magnetic fusion reactor
14
• Flux of (particles + heat + 14 MeV neutrons) ~10 MW/m2
A FUSION REACTOR IMPLIES MANY INTERFACES BETWEEN THE PLASMA AND MATERIALS
Particles and surfaces
Unlike nuclear fission where energy is volume-distributed
Key role of PMI in fusion research well recognized in US and internationally! A difficult interfacial physics !
Why lithium? Carbon?
Why is PMI important?
OFES Review, ORNL, March 02, 2010 Managed by UT-Battelle for the Department of Energy
Why defining the data for PMI is so difficult
problem?
2) It mixes material of the two worlds,
creating in between a new entity :
P-M DYNAMICAL SURFACE
which communicates between the two!
1) Interfacial physics, “when the two
worlds meet” : traditionally the most
challenging areas of science
3) Plasma is source of synergistic effects of many energies, angles, particles…
These are the main causes of the simulation difficulties
OFES Review, ORNL, March 02, 2010 Managed by UT-Battelle for the Department of Energy 16
PMI has many fundamental processes & synergies
elastic reflection
implantation
re-emission &
sputtering &
chemistry
trapping/detrapping
retention
Plasma Material
diffusion, permeation
Give rise to synergistic effects
Damage Effects: Vacancies, bubbles, blisters, dislocations, voids, neutrons?
Drivers: Multi -T, -n, -species, plasma irradiation, neutrons sheath acceleration
Erosion
Ablation
Melting (metals)
Re-deposition
Co-deposition
When an ion or neutral arrives at a surface it undergoes a series of elastic and inelastic collisions
with the atoms of the solid.
What surface :Chemistry, Mixture, Morphology?
OFES Review, ORNL, March 02, 2010 Managed by UT-Battelle for the Department of Energy Data Evaluation for AM and PMI Processes in Fusion, IAEA-NFRI, Sept. 2012, Daejeon, Korea
Model validation is usually defined to mean “substantiation that a computerized model within its domain of applicability possesses a satisfactory range of accuracy consistent with the intended application of the model.” (Schlesinger at al 1979). Comparison with experiment :: qualitative
Model verification is often defined as “ensuring that the computer code of the computerized model and its implementation are correct”. Code testing against simple models
Uncertainty quantification science tries to determine how likely certain outcomes are if some aspects of the system are not exactly know. Here: Model parameters may vary between different instances of the same object for which predictions are sought. Monte-Carlo approach to trajectories over the surface
OFES Review, ORNL, March 02, 2010 Managed by UT-Battelle for the Department of Energy Data Evaluation for AM and PMI Processes in Fusion, IAEA-NFRI, Sept. 2012, Daejeon, Korea
What does flux of 1025 particles/m2s mean (ITER) for a typical atomistic (MD) simulation?
At a box of surface of 3 nm lateral dim? a few thousands atoms (carbon) The flux is 0.01 particle/nm2ns 1) 1 particle at the interface surface of the cell each 10 ns. But for deuterium with impact energy less then 100 eV: Penetration is less than 2 nm, typical sputtering process takes up to 50 ps Is each impact independent, uncorrelated?
Each particle will functionalize the material, change the surface for the subsequent impact! Processes essentially discrete Atomistic approach!!! But with memory!!!
18
Why atomistic approach?
OFES Review, ORNL, March 02, 2010 Managed by UT-Battelle for the Department of Energy 19
Classical MD is only as good as the interatomic potential model used
Most advanced: hydro-carbon potential developed for chemistry
• Brenner, 1990 , 2002 : REBO, short range, 0.2nm
• more sophisticated AIREBO (Stuart, 2000, 2004, 1.1 nm)
• > 400 semi-empirical parameters, “bond order”, chemistry
EX: MD calc. of reflection coeff.
• Significant sensitivity to changes
in potential model for some
processes
• Experimental validation essential to
establish credible MD simulation.
• Interatomic potentials for W, Be, C
exist (talk of Nordlund)
• Experimental validation?
Adaptive Intermolecular Reactive Bond Order (AIREBO) potential : torsion, dispersion, Van der Waals,
Improvements to CH potentials done (Kent et al, 2010) New Li-C-H-O potentials being developed (Dadras et al, 2010)
Reinhold et al, Nuc. Instr. Meth. B 267, 691 (2009).
Notice the problem with Eirene database!
OFES Review, ORNL, March 02, 2010 Managed by UT-Battelle for the Department of Energy 20
Beam-surface exp’t: precision control of projectiles & targets . . .
. . . enabled development & validation of MD approach
Meyer et al, Physica Scripta T128, 50 (2007).
Remarkable agreement of theory & exp’t when
simulation mimics exp’t. No fitting
parameters! Key: simulation prepares surface by bombardment!
+ Monte-Carlo 30,000 random trajectories- see the
error bars at the theory!!!
This od an example of UQ (Uncertainty
Quantification)
•Fluence important (not flux) learned from experiment
• Type, internal state, energy, angle as in exp’t exp with D 2
+
exp with D+
CD 3+CD 4
MD with D 2
(*)
MD with D
total C
MD with D 2(g)
hydrocarbon
Impact energy (eV/D)
7 8 9 10 20 30 40
Sp
utt
eri
ng
yie
ld (
/D)
10-3
10-2
10-1
exp with D 3+
What have we learned from the “next door”
beam-surface experiments? Vadidation a must!
Reaching “steady state”
21 Managed by UT-Battelle for the U.S. Department of Energy
Ion-surface scattering experiments: Ions on high-dpa, high temp W
MD and MC
with plasma
synergy
FNSF,
DEMO
Molecular Dynamics (MD)
& Monte Carlo (MC)
simulation
PMI
Design &
Validation
High-flux linear PMI experiment: Plasma on high-dpa, high temp W
QuickTime™ and a
decompressor
are needed to see this picture.
Toroidal confinement experiments
Potential models
Quantum-classical MD
Increases in
computational power
Integrated experimental and theoretical PMI research: Only way toward trusted data
Material
Science
Predictive science!!!
When available
Simulation of additional fast n & heat load, damage
OFES Review, ORNL, March 02, 2010 Managed by UT-Battelle for the Department of Energy
CONCLUSIONS:
• PMI extremely difficult interfacial problem (Material mixing create
SURFACE entity; scale depends on impact energy; What data?)
• PMI data can be built from bottom-up recognizing its multiscale
character and building from shortest time/spatial scales (fs/Angstrom)
up many orders of magnitude
• Theory&modeling of PMI MUST be validated by experiment (and v.v.),
(at least to explain phenomenology, is this then the data??)
• Irradiation create dynamical surface, changing interface, data must
come form the steady state, cumulative bombardment!!!!
• Surface responds to synergy in plasma irradiation (angles, energies,
particles), data do not NOT follow linear superposition principle; NEED
plasma irradiation modeling and experiments; dedicated plasma
devices a must for data validation
• Role of the national and international datacenters must be shifted
form collection and dissemination of atomic and PMI data for
fusion to the data evaluation and recommendation!!!!!
22
Categorization of PMI data with respect to surface and projectile states, to time-space scales, to environment, to type of users and to format – EXTREMELY DIFFICULT
WHAT ACCURACY? Need to be learned from fusion modelers!