pasi2012 bringa md
DESCRIPTION
MDTRANSCRIPT
COLLABORATORS: J. Rodriguez-Nieva , M. Ruda, G. Bertolino (InstitutoBalseiro), S. Ramos, E. Crespo (U. Comahue), D. Farkas, J. Monk (VaTech), A. Caro, A. Misra, M. Nastasi (LANL), R.E. Johnson, T. Cassidy (University of Virginia), C. Anders, H. Urbassek (TU Kaiserslautern), D. Schwen (UIUC), J. Wark, A. Higginbotham (Oxford), N. Park (AWE), B. Remington, J. Hawreliak(LLNL), M. Meyers, Y. Tang (UCSD), D. Bertoldi, C. Ruestes, E. Millan (UN Cuyo)
PASI 2012 - CMS4EPan-American Advanced Studies InstituteComputational Material Science for Energy Generation and Conversion
Pontificia Universidad Católica de ChileSantiago, Chile 2012
Eduardo M. Bringa
CONICET
Instituto de Ciencias Básicas, Universidad Nacional de CuyoMendoza, Argentina
https://sites.google.com/site/simafweb
Advanced Molecular Dynamics
Outline• Introduction
• Main caveats of MD simulations: potentials, time step, boundary conditions.
• Nano-thermodynamics and non-LTE conditions: temperature, pressure, melting, thermalization.
•Data analysis
• Simulation of “realistic” samples (not perfect crystals): dislocations, grains, impurities, porosity.
• Mechanical properties at high strain rate.
• Electronic excitations, taking ion tracks as case study.
•Summary, conclusions and future perspective
Given all the approximations and limitations involved …
Can we hope for quantitative agreement between MD and experiments?
OR …
Is MD just pretty movies and pictures?
Two questions …
Need to be extremely careful about applicability of classical MD
• Generally assumes Born-Oppenheimer approximation works.
•The corresponding de Broglie wavelength, proportional to (Mass Temperature)-1 has to be much smaller than the mean atomic separation, i.e. try to avoid light elements and low temperatures ☺
•One should avoid phase transitions driven by electronic effects, like metal-insulator transitions, magnetic transitions, etc.
•One should avoid regions where electronic excitations could arise and play an important role in the evolution of the system.
Pushing the limits of validity one can sometimes obtain results resembling experiments, but possibly for the wrong reasons ☺
With MD you can obtain….
“Real” time evolution of your system.
Thermodynamic properties, including T(r,t) temperature profiles that can be used in rate equations.
Mechanical properties, including elastic and plastic behavior.
Surface/bulk/cluster growth and modification.
X-ray and “IR” spectra
Etcetera …
•Can simulate only small samples (L<1 µm, up to ~109 atoms).•Can simulate only short times (t<1 µs, because ∆t~1 fs).•Computationally expensive (weeks).• Potential’s golden rule: trash in ���� trash out.• Interaction potentials for alloys, molecular solids, and excited species not well know. • Despite its limitationsMD is a very powerful tool to study nanosystems.
Limitations of MD
Atomistic simulations are extremely helpful but … still have multiple limitations
“Potential” problems: EAM potentials, phonons and elastic constants or when, even if the PV EOS is OK, other things can go wrong (Paul Erhart, LLNL)
0
100
200
300
400
500
0 20 40 60 80 100
Ela
stic
con
stan
ts (
GP
a)
Pressure (GPa)
B
c11
c12
c’
0
1
2
3
4
5
6
7
ΓPHΓN 0
5
10
15
20
25
Fre
quen
cy (
TH
z)
Fre
quen
cy (
meV
)
Guellil−Adams potential, this workexperimental, Woods (1964)DFT, Bercegeay and Bernard (2005)
−100
−50
0
50
100
150
200
250
300
350
400
−30 −20 −10 0 10 20 30
Ela
stic
con
stan
ts (
GP
a)
Pressure (GPa)
c11
c12
B
c’
0
1
2
3
4
5
6
7
ΓPHΓN 0
5
10
15
20
25
Fre
quen
cy (
TH
z)
Fre
quen
cy (
meV
)
Li−Adams potential, this workexperimental, Woods (1964)DFT, Bercegeay and Bernard (2005)
Good agreement with phonons at P=0 GPa, but discontinuities in elastic constants, due to splines in the potential, lead to multiple elastic fronts
Another example: EFS Ta Potential
• Excellent agreement with PV, equilibrium Hugoniot, melt line , etc.• Elastic constants OK up to ~1 Mbar.• BUT… BCC�HCP at ~65 GPa (Ravelo et al., SCCM-2011).
Tang, Bringa, Meyers, Acta Mat. 59 (2011) 1354
Z-L Liu, L-C Cai, Phys. Rev. B 77 (2008) 024103
X.D. Dai, Y. Kong , J. Phys. Cond. Mat. 18 (2006) 4527
Potential validity depends strongly on type of fit, which can emphasize a certain property, temperature & pressure range, structure, etc.
Potentials are often non-transferable �
COLLISION DYNAMICS OF CARBON CLUSTERS, NVE (D. Bertoldi)
AIREBO potential
“dt reset”tmin = 1.E(-6) pstmax = 0,001 psXmax = 0,0001 Å
Time Evolution: be aware of possible energy drift
-1 0 1 2 3 4 5 6 7-0,016
-0,014
-0,012
-0,010
-0,008
-0,006
-0,004
-0,002
0,000
0,002
Tot
al e
nery
cha
nge
(%)
Time (ps)
•Energy drift caused by integrator and computational errors.
•Simulations far from equilibrium have to use variable time step schemes,
> 1% ���� problematic!
• PBC in (x,y), free BC with expansion (z). Langevin bath with critical damping at the sides.
• Need to re-calculate damping for each interatomic potential and bath condition.
• There are complex schemes to have impedance matching at boundaries, but none standard.
• Size has to be large enough to capture desired phenomena. Need to verify this by running simulations of different sizes: results should not change beyond certain size L, or they could be extrapolated versus 1/L.
FixLangevin, T=0.1
Mobile, NVE
Simulation details need to include info on BC
Track T=10
Simulation of hot spot
Large-scale MD links nano and microscalesin damage induced by nanoprojectiles [C. Anders et al., PRL 108, 027601 (2012)]
Rcluster =20 nm, 20 ps after impact, ~300 10 6 atoms, 15 hours using 3,840 CPU’s in Thunder (LLNL)
Only dislocations + liquid atoms are shown
Tight Binding
MD
Continuum mechanics
1 meter= 1,000,000,000 nanometers
1 meter= 1,000,000 microns
Human hair’s width ~ 200 µmEstaphilococus ~ 1 µmRhinovirus ~ 200 nm
Kaxiras’ web
Nanoscale does affect macro scale and properties of
materials.
Coupling TIME and length scales ….
• Concurrent coupling methods have many caveats ….
• Choose set of parameters from MD, save those parameters and “pass” them to a “higher” level code. Example: calculate defect concentrations as the initial configuration for a kinetic Monte Carlo code.
• Use some accelerated technique, which boost the time step, for instance “TAD” by A. Voter (LANL). Very expensive computationally, practical only for “2D” simulations or small 3D simulations.
Several people are currently working on improving this situation … Keep tuned!
• Introduction
• Main caveats of MD simulations: potentials, time step, boundary conditions.
• Nano-thermodynamics and non-LTE conditions: temperature, pressure, melting, thermalization.
•Data analysis
• Simulation of “realistic” samples (not perfect crystals): dislocations, grains, impurities, porosity.
• Mechanical properties at high strain rate.
• Electronic excitations, taking ion tracks as case study.
•Summary, conclusions and future perspective
(a) Definition of temperature in nano systems
Jellinek & Goldberg, Chem Phys. (2000)Pearson et al, PRB (1985)
Usual: (3/2) N kB T = Ekin
Nano Systems:
Correction due to non-zero flow velocity <v>:
Ekin� (m/2) (v - <v>)2
Ekin>0, but T=0
v
“Partial” T’s: T rot, Tvib, Tij
R
(b) Nanoclusters de aluminio
R =
0,3nm � 13 átomos0,5nm � 43 átomos1nm � 249 átomos2nm � 1985 átomos
Potencial interatómico: átomo embebido (EAM) de Mishin[2]
Dinámica Molecular -LAMMPS[1] -
[1] www.lammps.sandia.gov[2] Y. Mishin et al., “Embedded-atom potential for B2-NiAl” PRB 65 (2002)
Ensamble Microcanónio (N,V,E)Temperatura ~ 300K
D. Bertoldi (ICB)
10 100 1000240
260
280
300
320
340
360
380 simulation value theoretical value
Tim
e av
erag
e te
mpe
ratu
re (
K)
N° atoms
bulk
Cluster Temperature
---Calculated with
Td (Jellinek et al.)
---MD
Averaged during the last 0,1 ps of simulation
Velocity Distribution in clustersAproximate with a Maxwell-Boltzmann distribution
When there are not enough atoms, distribution have to be averaged over time, but M-B still works reasonably well.
0 300 600 900 1200 1500 1800
f
bulk 1985 atoms249 atoms
Velocity (m/s)
25
20
15
10
5
0
0 200 400 600 800 1000
43 atoms25
0
5
10
15
f
20
Velocity (m/s)
(c) Can we define an atomic stress tensor? Only with caveats
PdH nanoclusters . Using Voronoi or mean volume gives roughly the same results. Work with
G. Bertolino, M. Ruda (Centro Atomico Bariloche), S. Ramos, E. Crespo (UN Comahue, Neuquen)
Int. J. Hydrogen Energy (2012), in press .
a, b= x,y,z . Includes thermal, pair, bond, angle, dihedral, improper, and “fix” Be careful with NkBT term …it should discount flow velocity in calculation of TS = σσσσ V � how do we define “atomic” volume to calculate momentum flux?
Possible solution: use Voronoi polyhedra
Virial stress for atom I(lammps)
(1700 MPa, 0 K) (12000 MPa, 5000 K)
P = 1700 + 2.06T
P= 4 γ /dEc. Laplace-Young :d = 4 γ /(1700 + 2.06T )
in equilibrium
γ= (γD + γG )/2
0 500 1000 1500 2000 2500
2
4
6
8
diam
etro
(nm
)
Temperatura (K)
Diamante
Grafito
Transitions diamond �graphite and graphite � diamond , have different cinetics
Zhao, D. et al, “Size and temperature dependence of nanodiamond-nanographite transition related with surface stress”, Diamond and Related Materials 11 (2002) 234.
(d) Phase Diagram for carbon clusters
Diagram for bulk sample Diagram for a spherical cluster
Pext =0 GPa
Temperatura (K)
Pre
sión
(G
Pa)
COLLISION OF CARBON NANOGRAINS
Dalia Bertoldi (ICB)
(f) Melting of Au nanoclusters
Lin, Leveugle, Bringa, Zhigilei, J. Phys. Chem. (2010)Melting temperature changes only ~15%
����difficult to detect experimentally
R=2.4 nm
(g) Heating a solid without a surface (no transport)
∆ΚΕ∆ΚΕ∆ΚΕ∆ΚΕ
Equilibration
•Is thermodynamics still valid at small times and length scales?•How long does it take for a “perturbed” system to equilibrate?
LJ solidBinding energy U
Eexc/atom
Thermodynamic Equilibrium?
EQUILIBRATION ISFASTER THAN
EXPECTED (compared with hard
spheres).Each atom interacts with ~80 neighbors
Equilibration time in homogeneous system
Bringa & Johnson, Nuclear Instruments and
Methods in Physics Research B 143(1998)
513-535
H(t)=
Rough way to obtain equilibration time
•Large gradients (103 K/nm) ….•How long does it take for LTE?
Local equilibration time is longer than equilibration time for homogeneous heating
(h) Heating a solid without a surface (only radial transport)heating of cylinder with radius rcyl
“Local” Boltzmann H-Function constant �thermal equilibrium
“Local” M-B distribution
Temperature profiles show axial energy transportLarge difference between radial and axial temperatu re
Temperature profiles (with a surface): lack of LTE and deviations from Fourier equation
Problems:•Lack of LTE more serious for molecular solids. Bringa and Johnson, Surf. Sci. (1997)
•Parabolic ���� HyperbolicD.D. Joshep, L. Preziosi, Rev. Mod. Phys. 61 (1988) 41.S. Volz, J. Saulnier, M. Lallemand, B. Perrin, P. Depondt, M. Mareschal, Phys. Rev. B 54 (1996) 340.•Stefan problem (moving boundary). Caro, Tarzia, etc.
Outline• Introduction
• Main caveats of MD simulations: potentials, time step, boundary conditions.
• Nano-thermodynamics and non-LTE conditions: temperature, pressure, melting, thermalization.
•Data analysis
• Simulation of “realistic” samples (not perfect crystals): dislocations, grains, impurities, porosity.
• Mechanical properties at high strain rate.
• Electronic excitations, taking ion tracks as case study.
•Summary, conclusions and future perspective
•On the fly and post-processing of data takes considerable time …•Need to choose appropriate analysis tools to avoid artificial results.• Whenever possible, carry out the analysis in parallel with domain decomposition and neighbor lists.• Care must be taken with time averaging.
Data analysis
Centro-Symmetry Parameter (CSP)*� Centro-symmetry parameter (CSP): a parameter to measure the
local disorder, particularly useful to study cubic structures.
* Kelchner, Plimpton, Hamilton, Phys Rev B, 58, 11085 (1998)
f.c.c structure
CSP expression for a f.c.c. unit cell
Kelchner et al, FIG. 2, partial view. Defect structure at the first plastic yield point during indentation on Au (111), (a) view along [112], (b) rotated 45° about [111]. The colors indicate defect types as determined by the centrosymmetry parameter: partial dislocation (red), stacking fault (yellow), and surface atoms (white). Only atoms with P>0.5 are shown.
This is done for every atom in the
sample ���� high computational cost
Common Neighbor Analysis• CNA: a parameter to measure the local disorder
• Sensitive to cutoff radius
• 12 nearest neighbor for perfect FCC and HCP crystals, 14 nearest neighbors for perfect BCC crystals
• Faken, Jonsson, Comput Mater Sci, 2, 279 (1994).• Tsuzuki, Branicio, Rino, Comput Phys Comm, 177, 518 (2007).
This is done for every atom in the
sample ���� high computational cost
DXA (Dislocation eXtraction Algorithm)
Stepwise conversion of atomistic dislocation cores into a geometric linerepresentation.
(a) Atomistic input data.
(b) Bonds between disordered atoms.
(c) Interface mesh.
(d) Smoothed output.
A. Stukowski and K. Albe, Modelling Simul. Mater. Sci. Eng. 18 (2010) 085001.
Changes in DXA parameters can have large effect on results
Outline• Introduction
• Main caveats of MD simulations: potentials, time step, boundary conditions.
• Nano-thermodynamics and non-LTE conditions: temperature, pressure, melting, thermalization.
•Data analysis
• Simulation of “realistic” samples (not perfect crystals): dislocations, grains, impurities, porosity.
• Mechanical properties at high strain rate.
• Electronic excitations, taking ion tracks as case study.
•Summary, conclusions and future perspective
Perfect crystals are the `spherical horse’ of atomistic simulations
0.5 µµµµm
Cu single crystal, M. Meyers et al, TEM
150 nm
How to make more realistic simulations? Add defects: vacancies � voids � bubbles, interstitials, dislocation loops/lines,
grain boundaries (bi-cristals � polycrystals)
Polycrystal (50 nm grain size)(400 million atoms)
Few GB are ΣΣΣΣ boundaries …
Dislocation loop
He bubble + dislocation loops
M. Meyers
Dislocation motion changes the stress state from 1D to ~3D
Bringa et al., Nature Materials 5, 805 (2006)Shehadeh, Bringa, et al., APL 89, 171918 (2006)
P~35 GPatR=50 ps
0 .6 0 .5 0 .4 0 .3 0 .2 0 .1 0 .00
2
4
6
8
1 00 p s
Z ; X - Y5 0 p s
Z ; X - Y
latti
ce c
om
pres
sio
n (%
)
z (µ m )
1 D
3 D
120 90 60 30 0
0.0
0.4
0.8
1.2
1.6
2.0 0 ps 50 ps
time after shock (ps)
σ SH
EA
R (
GP
a)
(a)
Difraccion simulada de rayos X
Nature Materials cover figureNature Materials cover figure
256-352 million atoms (~0.07 ××××0.07××××0.8 µm), Cu
Pre-existing dislocation loops needed to simulate realistic response to loading
1980 CPU’s in MCR (LLNL), ~15 días, 30 TB of data
GBs give unique properties under pressureSimulations and experiments show ultra-hard materials
E. Bringa et al., Science (2005), APL (2005 & 2006), etc.
4 108 átomos, en 4000 CPUs
d=50 nm, 47 GPa
Friction ~ normal stress (σσσσ) � σgrows and sliding decreases
σσσσ
σσσσ
σσσσcorte
σσσσcorte
GB
d=30-50 nm, 40 GPa
DISLOCATIONS
• High pressure decreases sliding and leads to dislocations. • High dislocation density leads to enhanced hardness after unloading (MD + exp.).
Lasers Omega & Janus
Role of impurities in void nucleation
(a)
(b)
(c)3% Fe
Combine MD with “parallel grand-canonical Monte Carlo” to place impurities in nano crystalline sample
“Effects of Microalloying on the Mobility and Mechanical Response of Interfaces in Nanocrystalline Cu”, Caro et al. , Materials Science Forum Vols. 633-634 (2010) 21.
G. Gilmer
This might lead to suppression of GB sliding � stronger materials
Foams are another example of challenging samples
•Porosity important for mass and thermal transport, and porous materials might be
radiation resistant (Bringa et al., Nanoletters, July 2012)
•Irradiation might affect mechanical properties
important for structural integrity, and for durability of reactor walls and fuel pellets.
Plasma exposed W-C surface Takamura et al., Plasma and Fusion
Research 1, 51 (2006)
I) Novel nanofoams offer promising mechanical properties. High surface/ volume ratio, and
therefore cannot use full density properties for predictions. Often modeled as 2-phase material.
II) Understanding of early stages of spall.
III) Materials for new fission and fusion reactors
Possible morphological changes under irradiation
Displacement of atoms for 5 cumulative PKAs. Color scale indicates displacement, with atoms displaced less than half a lattice parameter not shown, and atoms displaced more than 4 lattice parameters in red.
Mechanical properties of irradiated foams: need “long” simulations for defect evolution
0 1 2 3 4 50,0
0,2
0,4
0,6
0,8
1,0
σ (G
Pa)
ε (%)
5.5 107/s
5.5 108/s 5.5 108/s, irradiated Eε
tension, ρ/ρo=0.64
• Irradiated foam/filaments includes defects that lead to hardenning. Kucheyev et al., J. Phys D (2009), found stiffening of irradiated silica foams.•There is slight foam compaction that increases with dose
D=50 nm, 40 keV PKA
If filaments too small � melting & destruction, compaction.If filaments too large � defects take “too long” to migrate to the surface.Simple model: includes molten volume, cascade size, dose rate effects. No compaction.
Sweet spot: filaments between ~5-500 nm are radiation resistant (moderate dose)
Experiments: no radiation damage for moderate dose irradiation
A. Misra, M. Nastasi and co-workers (LANL)
20 nm
(a) (b)
1 nm
77 K Ne+ irradiation to 4.5 x 1014 /cm2 (1 dpa) (a) Bright field TEM image of np-Au after irradiation. (b) high-resolution TEM image of np-Au after irradiation. (c) under focused bright field TEM images of single crystal
Au film after irradiation, showing Ne bubbles.
Bringa et al., Nanoletters, in press (2012)
(c)
• Introduction
• Main caveats of MD simulations: potentials, time step, boundary conditions.
• Nano-thermodynamics and non-LTE conditions: temperature, pressure, melting, thermalization.
•Data analysis
• Simulation of “realistic” samples (not perfect crystals): dislocations, grains, impurities, porosity.
• Mechanical properties at high strain rate.
• Electronic excitations, taking ion tracks as case study.
•Summary, conclusions and future perspective
Mechanical deformation: can we compare MD with experiments? Problem: strain rates, dεεεε/dt, are huge!
“Today” MD can do strain rates~106/s only at a supercomputer
Simulation time=t ∆∆∆∆L/L~1% ���� dεεεε/dt=0.01/t
MD uses ∆∆∆∆t~ 1 fs
t=10 ns (107 steps)���� dεεεε/dt=106/s
t=1 s (1015 steps)���� dεεεε/dt=101/s
In a supercomputer:106 steps���� 1-5 days
1015 steps���� millions of years …
Solutions:
a) Suplement MD with models to compare to experiments at low strain rates.
b) Compare with experiments at high strain rates (lasers, shocks, impacts, etc.).
∆∆∆∆L
L
Road-Runner (2010)A. Voter, LANL ����
nanowire at 10-3/s
� Nano-scale failure under extreme conditions is not well understood.
� Continuum predictions do not typically include atomistic mechanisms.
� Atomistic Molecular Dynamics (MD) simulations can reproduce ductile failure behavior, and lead to new models at the macroscale.
� There are several postulated nano-scale mechanisms, but still relatively few simulations to test them.
� Have to compare to experiments at appropriate strain rate …
Nanoporous Au Stress – Strain curves
ε = 2.5 %
ε = 2.8 %
ε = 3.0 %
ε = 3.3 %
ε = 3.8 %
J. Rodriguez-Nieva
Comparison between MD & models/experiments
is promising for FCC materials
Davila et al., APL (2005) 86,161902. Traiviratana, Bringa & Meyers, several papers
-------- Reissman et al (prismatic loops)------- Lubarda (shear loops)………. Lubarda (shear loops)_____ MD Davila et al
Cu: Loading on [111] � Triplanar Loops
Void Faceting (Rudd et al., Traiviratana et al., etc.) recently observed in experiments
with sub-micron voids
Modified DXA + ParaViewAtomistic simulation of the mechanical properties of a
nanoporous b.c.c. metal *
* Ruestes et al. , submitted to Acta Materialia (2012)
VMD
ParaView visualization of the results provided
by DXA for a nanoporous Ta sample subjected
to a 109/s uniaxial compressive strain rate at
an 8% strain. Preprocessed sample has 1.9
million atoms.
Run: 3 days in 32 cores
Analysis of each snapshot: 10 min run on
AMD M520 + 4Gb RAM (dual core)
10 min run
CNA analysis takes about 1/3 of the total analysis time
DXA+ParaView
Can we obtain dislocation densities?
•Rough estimate of total dislocation density calculated from the number of atoms with CNA not BCC, and dividing by 10 to account for cross-section of dislocation cores.
• Mobile dislocation densities calculated from plastic heating* [A. Higginbotham et al., JAP (2011)].
Can we compare our results with
experiments?
After relaxation to P=0.Possibly, because long-
term recovery of the microstructure in bcc
samples should have minor effects on total density.
Note the absence of twins in the recovered sample,
which can be checked with X-ray diffraction.
Outline• Introduction
• Main caveats of MD simulations: potentials, time step, boundary conditions.
• Nano-thermodynamics and non-LTE conditions: temperature, pressure, melting, thermalization.
•Data analysis
• Simulations of “realistic” samples (not perfect crystals): dislocations, grains, impurities, porosity.
• Mechanical properties at high strain rate.
• Electronic excitations, taking ion tracks as case study.
•Summary, conclusions and future perspective
FAST ION
H2O
Track evolution and modeling
Many “analytic” models
Coulomb Explosion
Thermal Spikes (many flavors)
Fleisher, Price and Walker, J. App. Phys. 36, 3645 (1965)
Trautmann, Klaumunzer and Trinkaus,
Phys. Rev. Lett. 85, 3648 (2000)
MORE ….
MD simulations can
predict experimental track sizes
[Devanatham et al., NIMB (2008)]
and sputtering[Bringa and Johnson, PRL 88, 165501, (2002)]
G. Schiwietz
simplify greatly
but MD is better
∝
PE~(Nch
e)2/d ∆KE
Y (dJ/dx)2
dJ/dx= A (Zeff
/v)2 ln(B v2)
+
+
+
+
+ +
+
+
+
O2N2COH2O
MeV H+, He+
rtrack
∝Nch (dJ/dx)
µ (dE/dx)e
h
"Reflecting" Boundary
Free BCor PBC
Free Surface
rtrack ++
++
+
+
Neutralization Time τ
V(rij)=VL-J(rij)+ (ZiZj/rij) Exp(-rij/a). rcut-Coul= 7 a
depth of layer h > 2 rcut-Coul. rtrack / Nch-MAX= 2 charges/layer
MD Simulations of Coulomb Explosion
∝∝∝∝Nch (dJ/dx)
MD N(0) Exp(-t/τ)
Num
ber o
f Cha
rges
time
Similar to H-P. Cheng work for HCI
•Extremely difficult to determine neutralization times, both theoretically and experimentally
• If net repulsion is due to excited states, they may decay too fast, or they may diffuse far away before decaying.
Example: Coulomb explosion (CE) simulations
Coulomb explosion simulations
Nch=2, a~l, rtrack~l, t=1 ps~2 tDcharged atoms have twice the radius of neutral atoms
If neutralization is too fast there is
no heating of the lattice!
0.1 1 100.01
0.1
1
10
E
-2
2nd layer 1st layer
(a) Coulomb, τ ≈ τD
(b) Coulomb, τ ≈ 20 τD
(c) Thermal Spike
(dY
/dE
)/Y
E/U
Coulomb explosion leads to heating of the lattice as in TS
Y~(dJ/dx)2 ln(a)
Is there a thermal spike (TS)? YES!
Analyzing MD data, there is a clearly defined energetic track with:
1) (dE/dx)eff =A (dJ/dx)2
2) rcyl= Constant YMD-CE=YMD-TS
Coulomb
Spike
Prompt
CoulombBringa and Johnson, PRL 88, 165501 (2002)
Example: Tracks in ta-C create conducting nano-wire s
Schwen & Bringa, NIMB 2007
Simulations
and
experiments
display creation
of sp2 rich
region
Cosmic ray destroying a carbon grain
Hillock formation: excellent agreement with experiments
Schwen & Bringa (submitted)
Example: MD simulations of silica amorphization
0
500
1000
1500
2000
2500
3000
3500
4000
0 25 50 75 100 125 150Radial Distance (A)
Tem
pera
ture
(K)
.20 ps
1.08 ps
3.88 ps
13.88 ps
28.88 ps
73.88 ps
P. Durham (LLNL/Cincinnati), R. Devanatham, R. Corrales (PNNL)
• DL_POLY with parallel domain decomposition, 32-128 CPUs, 104-1.25 106 atoms.
•Buckingham potential + SPME
•Thermal spike model
0.2 ps 13.9 ps
Devanatham et al., NIMB (2007), Bringa et al., Ap.J. (2008)
Personal view of the future of MD• Sample size: in 10 years, ~tens of µµµµm, but most simulations still sub-µµµµm. •More/better hybrid codes to extend time and length scales: MD+MC, MD+kMC, MD+DD, MD+continuum, MD+BCA, MD+TB, MD+CPMD, …•Time scale problem: new algorithms to extend time scale and simulate thermal evolution.• Better description of electronic effects by: I) Physics + Chemistry + Biology ���� “reactive” potentials that are accurate and
efficient for full periodic table.II) coupling to CPMD, tight-binding, etc. (TDDFT?)III) TTM, Ehrenfest dynamics, inclusion of magnetic effects, etc.
Major roadblocks:• Computers (CPU/GPU) are becoming faster and larger, but algorithms for long range potentials(biology & oxides), ab-initio and continuum simulations typically do not scale well beyond couple thousand CPUs ���� expect better results within the next 10 years.• No set recipes to build better potentials, specially if chemistry (reactive potentials) or electronic effects (potentials for excited states, etc.) are involved.• Nobody knows yet what to do to solve the time scale problem beyond some simple model problems.