Computational Science Center• Brief History of CSC

Aug. 2001 – Computational Materials Science Group Launched.

Nov. 2003– 1024 cpu Linux cluster system (2.85 Tflops) set-up

2005 and 2006– The best research group awards in KIST

Jan. 2007– Computational Science Center Launched.

Cu doped GaN– Nano Letters, 7, 3366 (IF=9.960) – Appl. Phys. Lett. 90, 032504 (IF=3.977)

Me-C bonds – Carbon, 46, 185 (IF=4.260)

Thin film interface – Acta Mater. 56, 1011 (IF=3.549)

Protein structure calculation– J. Comp. Chem, 28, 2552 (IF=4.893)

Nanostructured surface – Adv. Mater. 20, 1903 (IF=7.896)

• Highlight Publications (2007~2008)

• Organizational Chart

Head of CSCKwang-Ryeol Lee

HPC & Bio Simulation Nano-materials Simulation

Kwang-Ryeol LeeKyu-Hwan Lee Hwanwon ChungJung Soo Oh Jung-Hae Choi Seung-Cheol LeeMyoung-Woon Moon

Interface & Surface

• Nano and Bio Tecchnology

(sub)atomic scale understandingsof structure, kinetics and properties

HPC & Bio Application

KIST Linux cluster (1024 CPU)

Protein folding by optimization

Simuloscope

Visualization & Analysis

351

351

351

351

351

351

351351

351351 351 351351

351351 351

35142

351

351 351

351 351

351351351351

351351351

351351351351351

351351

351351351351

351351351

68 97 83 41 27 90

25 72 20 4052 38 5996 12

14 6757 477110 95 78 4858 2

298 87 3554 61 105 9 93 109 103 31 15

36 100 84 10443 28 62 60 81 3 17 8526 91 71

44 22 65 550 106

7916747671071889

11192498299693346

2155 45 53101

66 51 19102 3473 2375 5698

86

30

13 10880 9437 88 10 70

246 47 3239 63 64

351

351

351

351

351

351

351351

351351 351 351351

351351 351

35142

351

351 351

351 351

351351351351

351351351

351351351351351

351351

351351351351

351351351

68 97 83 41 27 90

25 72 20 4052 38 5996 12

14 6757 477110 95 78 4858 2

298 87 3554 61 105 9 93 109 103 31 15

36 100 84 10443 28 62 60 81 3 17 8526 91 71

44 22 65 550 106

7916747671071889

11192498299693346

2155 45 53101

66 51 19102 3473 2375 5698

86

30

13 10880 9437 88 10 70

246 47 3239 63 64

Optimization algorithmGenetic algorithm

Ab-Inito Calculation

SiGe

Si50Ge50 1-Dimensional Nanowire

Strain effects on Ge thin films

109.5o

C/Me

90o~130o

Bond characteristics In DLC containing metals

Synth. & Appl.of Nanostructure Surface

Bio-medical Applications of Nanostructure

Fuel cell nano-fiber : Energy

Fundamental Understanding of Surface Nanostructures

Asymmetric surface intermixing during thin-film growth in the Co–Al

Molecular Dynamics

[001]

[100] [110]

θ=2o

[110]

θ=2o

[010]

[001]

[100]

[010]θθ

Structure of Si (001) vicinal surface

[010]θ

-30 -20 -10 0 10 20 30-30

-20

-10

0

10

20

30

Distance from Impact Point [A]

Dis

tanc

e fr

om Im

pact

Poi

nt [A

]

Surface nanostructure evolution by ion bombardment

CNT buckling by sonication(in collaboration with Brown Univ.)

Stable Phase of Ni Nanocluster

• Simulation Scope

Researchers & Developers

Researchers & Developers

Potential DB

Virtual Reality User I/F

MultiscaleSimul. Codes

MassiveComputation

PotentialDevelopment

Computation

Theory Experiment

Traditional

Computational science can- reduce time and cost for R&D- increase research efficiency

Computation

Theory Experiment

Computational science can- reduce time and cost for R&D-

High Performance Computing & Bio Simulation

• Hardware/Software Infrastructure • Bio Simulation

All-Atom De novo Protein Folding with a Scalable Evolutionary Algorithm

(a) : Starting conformation(b), (c) : Overlay of the experimental (red)

and the folded structure (blue). Backbone RMS deviation = 3.43 A ˚

(d) : Schematic illustration of the different routes of the evolutionary algorithm

Simulation Softwares

LAMMPS

EAM/MEAM

SiON FFBrenner Polymers FF

Tersoff

SIESTA TB Calculation

Kinetic MC

Ab initio MD & MC

KIST-MD

Ab-initio Calc. CodesVASP, WIEN2K, SIC-LSDA, DMOL3

Simuloscope

Electron Transport Theory

KIST Suite forComputational

Nanoscience(KIST-SCN)

Residual Stress

Bond Coord.

RDF & P-RDF

Density

Ring Statistics

Action Driven MD

Reactive FF

GROMACS

Crystal Builder

EAM converter

XMD

EAM Brenner

Visualizer

Bond Angle DF

Me-O FF

KIST Linux Cluster 1,024 CPUs

Opening, 2003.12.24

Certificate, 2003.11

512 Computing Nodes

Myrinet

Public Network(KOREN)

Head Node

NFS server

Storage

Partnership & Leadership for nation-wide Supercomputing Infrastructure

PLSI Project

Linux Cluster for ab-initio CalculationOS : CentOS 5.2CPU type : Quad-core Intel Xeon E5450, 3.0GHzNumber of Nodes : 16Number of Cores per node : 8 Memory per node : 32 GBInterconnection Network : Myrinet 10G NIC & SwitchFile Server : 25 TB (GFS, GPFS client)

OptimizationGenetic Algorithm, Simulated Annealing, etc.

• Optimization & Algorithms

351

351

351

351

351

351

351351

351351 351 351351

351351 351

35142

351

351 351

351 351

351351351351

351351351

351351351351351

351351

351351351351

351351351

68 97 83 41 27 90

25 72 20 4052 38 5996 12

14 6757 477110 95 78 4858 2

298 87 3554 61 105 9 93 109 103 31 15

36 100 84 10443 28 62 60 81 3 17 8526 91 71

44 22 65 550 106

7916747671071889

11192498299693346

2155 45 53101

66 51 19102 3473

2375 5698

86

30

13 10880 9437 88 10 70

246 47 3239 63 64

TozongDock : Application of Stochastic Optimization to Protein-Ligand Docking

Wenzel, Hamacher, Phys. Rev. Lett. 82, 3003 (1999)

Morris et al.

Protein-Ligand Docking

Stochastic Tunneling

+

• Visualization : 3-D Stereo Graphic System

Emitter

Visualization Workstation

Crystall eye glasses X 5

Stereo Projector

First Principles Calculation of Materials• Softwares for Ab-Inito Simulations

Ab-initioAb-initio

*VASP: Planewave pseudopotential

NEB: Nudged Elastic BandUtilitiesUtilities

*WIEN2k: Full potential LAPW

CRYSTAL06 : Atomic orbital

ABINIT, CPMD, SIESTA

Structure Utilities : Actively developing

Cluster Expansion : ATAT

• Search for New Functional Materials

CuCu

-6 -4 -2 0 2 4-0.5

-0.4

-0.3

-0.2

-0.1

0.0

0.1

0.2

0.3

0.4

0.5

DO

S (s

tate

s/eV

cel

l)

Energy (E-EF, eV)

1st NN 2nd NN 3rd NN 4th NN 5th NN

Magnetization Density Density of StatesZinc Blende GaN

Diluted Magnetic Semiconductors: Cu doped GaN

• Search for New Functional Materials:Monte Carlo + Ab-Initio

<110>

<1-10>

<001>

ZB r1 r2 r3 r4 r5 r6 r7 r8 r9 r10 SL

-316.0

-315.8

-315.6

-315.4

-315.2

-315.0

-314.8

Tota

l Ene

rgy

(eV/

cell)

315.78 eV

0.82 eV

0.21 eV

ZB

Random

SL

Si

Ge

Si50Ge50 1-Dimensional Nanowire

• Understanding Bond Nature

109.5o

Me

90o~130o

MeΔEMe-C

CMax;1.05

Z;4

35

d=1.54

AlMax;0.69

d=2.05

MoMax;0.69

d=2.10

AgMax;0.63

d=2.273 4 5 6 7 8 9 10 11 12 13 14 15

-1

0

1

2

Fe

Mn

V

ZnCd AlNi

Pd

CoCuAgAu

MoCrW

Ti

Si

C

ΔE

90o (e

V)

atomic group (# of valence e-)

2nd period 3rd period 4th period 5th period 6th period

Metal Incorporated Carbon

Ge on Biaxial Compressive Strain

graded SiGeGe film

Si substrate

Ge film

Si substrate

Ge

Si2 nm

Strained !

• Strain Effects – Thin Film

[0001]

1st Mn

2nd Mn

Compressive strain

0.316eV

• Strain Effects - NanowireDopant Configuration in GaN-nanowire Unstrained

0.032eV

IPD

OPD

Unstrained-IPD

Mn Mn

N

3.09 Å

DO

S

Partial DOS of Mn

E-EF(eV)

Total DOS of Mn

E-EF(eV)

DO

S

Compressive strain-IPD

Mn Mn

N

2.95 Å

DO

S

Partial DOS of Mn

E-EF(eV)

dx2-y2

dz2

dxy

E-EF(eV)

DO

S

Total DOS of Mn

Schrödinger’s Equation

Cohesive energy, Lattice parameterBulk modulus, Electronic propertiesMagnetic Properties etc …

HΨ=EΨ

Molecular Dynamics Simulation• Thin Film and Surface Phenomena in Atomic Scale

Asymmetic Intermixing of Interface

Co

Al

Al

Al

Co

Co

0.1eV Al at 300K0.1eV Co at 300K

Origin of Stress in Amorphous Carbon Film

Experiment

0 50 100 150 200 250 300

-6

-3

0

3

6

9

12

15

Stre

ss [G

Pa]

Beam Energy [eV]

Simulation

0 -100 -200 -300 -400 -5000

2

4

6

8

10

DC Bias (V)

Stre

ss (G

Pa)

High Stress Configuration

2.0~2.2 Å j80~95 º

i

Low Stress Configuration

~2.5 Åj

~105 º

i

1 2 3 4 50

5

10

15

300 eV

150 eV

100 eV

50 eV

10 eV

1 eV

Pair

Cor

rela

tion

Func

tion ρ(

r)

r [A]

Surface Structure Evolution by Ion Bombardment

10 keV Ar ion impacts on Au(001)

Ion

Sputtered atoms

Rearranged atoms

-30 -20 -10 0 10 20 30-30

-20

-10

0

10

20

30

Distance from Impact Point [A]

Dis

tanc

e fr

om Im

pact

Poi

nt [A

]

Pd

T.C. Kim et al., Phys. Rev. Lett., 92, 246104 (2004)

Atomic scale composition separation in CoCu alloy by Ar bombardment

0 100 200 300 400

35

40

45

50

55

60

Ato

mic

per

cent

of C

u

Number of Ar ion bombardment

1st layer 2nd layer 3rd layer 4th layer

Morphology Evolution of Nanoscale Materials

1 atom 50 atom 100 atom

150 atom 200 atom 264 atomAl deposition behavior on Cu (111)

CNT buckling by sonication(in collaboration with Brown Univ.)Stable phase in nanocluster is governed

by the surface energy.