1.021, 3.021, 10.333, 22.

00 Introduction to Modeling and Simulation Spring 2011 Part I Continuum and particle methods

Reactive potentials and applications (contd)

Lecture 9
Markus J. Buehler
Laboratory for Atomistic and Molecular Mechanics Department of Civil and Environmental Engineering Massachusetts Institute of Technology

Content overview
I. Particle and continuum methods
1. 2. 3. 4. 5. 6. 7. 8. Atoms, molecules, chemistry Continuum modeling approaches and solution approaches Statistical mechanics Molecular dynamics, Monte Carlo Visualization and data analysis Mechanical properties application: how things fail (and how to prevent it) Multi-scale modeling paradigm Biological systems (simulation in biophysics) how proteins work and how to model them

Lectures 2-13

II. Quantum mechanical methods

1. 2. 3. 4. 5. 6. 7. 8. Its A Quantum World: The Theory of Quantum Mechanics Quantum Mechanics: Practice Makes Perfect The Many-Body Problem: From Many-Body to SingleParticle Quantum modeling of materials From Atoms to Solids Basic properties of materials Advanced properties of materials What else can we do?

Lectures 14-26

Overview: Material covered so far

Lecture 1: Broad introduction to IM/S Lecture 2: Introduction to atomistic and continuum modeling (multi-scale modeling paradigm, difference between continuum and atomistic approach, case study: diffusion) Lecture 3: Basic statistical mechanics property calculation I (property calculation: microscopic states vs. macroscopic properties, ensembles, probability density and partition function) Lecture 4: Property calculation II (Monte Carlo, advanced property calculation, introduction to chemical interactions) Lecture 5: How to model chemical interactions I (example: movie of copper deformation/dislocations, etc.) Lecture 6: How to model chemical interactions II (EAM, a bit of ReaxFFchemical reactions) Lecture 7: Application MD simulation of materials failure Lecture 8: Application Reactive potentials and applications Lecture 9: Application Reactive potentials and applications (contd)
3

Lecture 9: Reactive potentials and applications (contd)

Outline: 1. Notes on fracture application 2. Closure: ReaxFF force field 3. Hybrid multi-paradigm fracture models Goal of todays lecture: Remarks: Modeling of fracture and relation to diffusion problem New potential: ReaxFF, to describe complex chemistry (bond breaking and formation) Application in hybrid simulation approaches (combine different force fields)

1. Notes on fracture application

Consider for pset #2

Brittle fracture mechanisms: fracture is a multiscale phenomenon, from nano to macro

Reprinted by permission from Macmillan Publishers Ltd: Nature. Source: Buehler, M., and Z. Xu. "Materials Science: Mind the Helical Crack." Nature 464, no. 7285 (2010): 42-3. 2010.

Limiting speeds of cracks: linear elastic continuum theory 9E

cl = 8
Mother-daughter mechanism SuperSub-Rayleigh Rayleigh Mode II Mode I Cr Subsonic Mode III Cs Linear Cl Nonlinear Cs Cl Supersonic Limiting speed v Limiting speed v Intersonic Supersonic

~ E

cs =

3E ~ E 8

cr = 0.92cs

Image by MIT OpenCourseWare.

Cracks can not exceed the limiting speed given by the corresponding wave speeds unless material behavior is nonlinear Cracks that exceed limiting speed would produce energy (physically impossible - linear elastic continuum theory)
7

Subsonic and supersonic fracture

Under certain conditions, material nonlinearities (that is, the behavior of materials under large deformation = hyperelasticity) becomes important This can lead to different limiting speeds than described by the model introduced above Deformation field near a crack

1 (r) ~ r

Elarge (stiff)
Stiff enin g Lin ea rt he or y
Strain
Image by MIT OpenCourseWare.

large deformation nonlinear zone singularity

Stress

ning Softe

Hyperelasticity

small deformation

9E cl = ~ E 8

Esmall (soft)
8

Energy flux concept

Far away from crack tip

csmall (soft)

Lenergy

Image by MIT OpenCourseWare.

Llarge (stiff) clarge (stiff)

Characteristic energy length (energy from this distance needs to flow to the crack tip)

Llarge (stiff)
Crack tip

Lenergy

1 Supersonic
cracking
9

Energy flux reduction/enhancement

New

K dominance zone Hyperelastic zone

Characteristic energy length Fracture process zone

Image by MIT OpenCourseWare.

Lenergy
Energy flux related to wave speed: high local wave speed, high energy flux, crack can move faster (and reverse for low local wave speed)

10

Physical basis for subsonic/supersonic fracture

Changes in energy flow at the crack tip due to changes in local wave speed (energy flux higher in materials with higher wave speed) Controlled by a characteristic length scale

Lenergy

Buehler et al., Nature, 2003

Reprinted by permission from Macmillan Publishers Ltd: Nature. Source: Buehler, M., F. Abraham, and H. Gao. "Hyperelasticity Governs Dynamic Fracture at a Critical Length Scale." Nature 426 (2003): 141-6. 2003.

11

Diffusion problem

Fracture problem
Continuum approach (distinct PDE)
Characteristic time of diffusion PDE (continuum equilibrium)

Integration (BCs, ICs)

c d 2c =D 2 t dx

PDE (Ficks law)

Integration (BCs, ICs)

Crack limiting speed

Atomistic approach (same PDE)

Integration (BCs, ICs)

source unknown. All rights reserved. This content is excluded from our Creative Commons license. For more information, see http://ocw.mit.edu/fairuse.

Integration (BCs, ICs)

Crack limiting speed Characteristic time of diffusion 12

2. Closure: ReaxFF force field

Potential energy expressions for more complex materials/chemistry, including bond formation and breaking

13

Review: atomic interactions different types of chemical bonds

Primary bonds (strong) Ionic (ceramics, quartz, feldspar - rocks) Covalent (silicon) Metallic (copper, nickel, gold, silver) (high melting point, 1000-5,000K) Secondary bonds (weak) Van der Waals (wax, low melting point) Hydrogen bonds (proteins, spider silk) (melting point 100-500K) Ionic: Non-directional (point charges interacting) Covalent: Directional (bond angles, torsions matter) Metallic: Non-directional (electron gas concept)

Difference of material properties originates from different atomic interactions

14

Butare all bonds the same? - valency in hydrocarbons

Ethane C2H6 (stable configuration)

All bonds are not the same! Adding another H is not favored

H Bonds depend on the environment!

15

Another challenge: chemical reactions

sp3 sp2

Energy

stretch
Transition point ???

C-C distance sp2

r
sp3

Simple pair potentials can not describe chemical reactions

16

Why can not model chemical reactions with springlike potentials?

stretch = kstretch ( r r0 ) 2

1 2

Set of parameters only valid for particular molecule type / type of chemical bond

kstretch,sp 2 kstretch,sp3

Reactive potentials or reactive force fields overcome these limitations

17

Theoretical basis: bond order potential

Concept: Use pair potential that depends on atomic environment (similar to EAM, here applied to covalent bonds)
5 Potential energy (eV) 0

Modulate strength of attractive part (e.g. by coordination, or bond order) Abell, Tersoff

-5

-10 0.5

1.5

2 Distance (A)
o

2.5

3.5

Triple Double Single

S.C. F.C.C.

Effective pair-interactions for various C-C (Carbon) bonds

Image by MIT OpenCourseWare.

Changes in spring constant as function of bond order Continuous change possible = continuous energy landscape during chemical reactions

18

Theoretical basis: bond order potential

5 Potential energy (eV)

-5

-10 0.5

1.5

2 Distance (A)
o

2.5

3.5

Triple Double Single

S.C. F.C.C.

Effective pair-interactions for various C-C (Carbon) bonds

Image by MIT OpenCourseWare.

D. Brenner, 2000

19

Concept of bond order (BO)

r

BO
sp3 sp2 sp

20

Bond order based energy landscape

Bond length Pauling Bond order Bond length

Energy Bond order potential Allows for a more general description of chemistry All energy terms dependent on bond order

Energy

Conventional potential (e.g. LJ, Morse)

21

Historical perspective of reactive bond order potentials

1985: Abell: General expression for binding energy as a sum of near nieghbor pair interactions moderated by local atomic environment 1990s: Tersoff, Brenner: Use Abell formalism applied to silicon (successful for various solid state structures) 2000: Stuart et al.: Reactive potential for hydrocarbons 2001: Duin, Godddard et al.: Reactive potential for hydrocarbons ReaxFF 2002: Brenner et al.: Second generation REBO potential for hydrocarbons 2003-2005: Extension of ReaxFF to various materials including metals, ceramics, silicon, polymers and more in Goddards group
22

Example: ReaxFF reactive force field

William A. Goddard III California Institute of Technology

Courtesy of Bill Goddard. Used with permission.

Adri C.T. v. Duin California Institute of Technology

23

ReaxFF: A reactive force field

Esystem = Ebond + EvdWaals + ECoulomb + Eval ,angle + Etors
2-body 3-body 4-body

+ Eover + Eunder
multi-body

Total energy is expressed as the sum of various terms describing individual chemical bonds All expressions in terms of bond order All interactions calculated between ALL atoms in system No more atom typing: Atom type = chemical element
24

Example: Calculation of bond energy

Esystem = Ebond + EvdWaals + ECoulomb + Eval ,angle + Etors + Eover + Eunder

p Ebond = De BOij exp pbe,1 1 BOij be ,1

Bond energy between atoms i and j does not depend on bond distance Instead, it depends on bond order

25

Bond order functions

BO goes smoothly from 3-21-0

(1) (3) (2)

Fig. 2.21c in Buehler, Markus J. Atomistic Modeling of Materials Failure. Springer, 2008. Springer. All rights reserved. This content is excluded from our Creative Commons license. For more information, see http://ocw.mit.edu/fairuse.

(1) rij BOij = exp + exp r0

(2)

(3)

rij + exp r0

r ij r0

Characteristic bond distance

All energy terms are expressed as a function of bond orders

26

Illustration: Bond energy

Image removed due to copyright restrictions. Please see slide 10 in van Duin, Adri. "Dishing Out the Dirt on ReaxFF. http://www.wag.caltech.edu/home/duin/FFgroup/Dirt.ppt.

27

vdW interactions
E system = Ebond + EvdW aals + ECoulomb + Eval ,angle + Etors + Eover + Eunder
Accounts for short distance repulsion (Pauli principle orthogonalization) and attraction energies at large distances (dispersion) Included for all atoms with shielding at small distances

Image removed due to copyright restrictions. Please see slide 11 in van Duin, Adri. "Dishing Out the Dirt on ReaxFF. http://www.wag.caltech.edu/home/duin/FFgroup/Dirt.ppt.

28

Resulting energy landscape

Contribution of Ebond and vdW energy

Source: van Duin, C. T. Adri, et al. "ReaxFF: A Reactive Force Field for Hydrocarbons." Journal of Physical Chemistry A 105 (2001). American Chemical Society. All rights reserved. This content is excluded from our Creative Commons license. For more information, see http://ocw.mit.edu/fairuse.

29

Current development status of ReaxFF

: not currently described by ReaxFF Allows to interface metals, ceramics with organic chemistry: Key for complex materials, specifically biological materials
Periodic table courtesy of Wikimedia Commons.

A--B

A B
30

Mg-water interaction: How to make fire with water

Video stills removed due to copyright restrictions; watch the video now: http://www.youtube.com/watch?v=QTKivMVUcqE.

Mg
http://video.google.com/videoplay?docid=4697996292949045921&q=magnesium+water&total=46&start=0&num=50&so=0&type=search&plindex=0

31

Mg water interaction ReaxFF MD simulation

32

3. Hybrid multi-paradigm fracture models

Focus: model particular fracture properties of silicon (chemically complex material)

33

Fracture of silicon: problem statement

Pair potential insufficient to describe bond breaking (chemical complexity)

Image courtesy of NASA.

34

Multi-paradigm concept for fracture

1 (r) ~ r
Need method good for describing rupture of chemical bonds

Need method good for elastic properties (energy storage)

Image by MIT OpenCourseWare.

35

What about combining different potentials?

Increased accuracy & transferability

Empirical models: mathematical functions with parameters (fitted to experiment or quantum mechanics)
Pair potentials (LJ, Morse, Buck., harmonic) (lecture 5) Embedded atom models/effective medium theories Multi-body potentials (e.g. Tersoff, CHARMM, etc.) (lecture 9 and following) Reactive potentials (ReaxFF) (lecture 9)

good for elastic properties (energy storage) good for describing rupture of chemical bonds multi-paradigm model

Computational efficiency

+ =

Semi-empirical models (explicitly note electronic structure)

Tight binding MINDO (=Modified Intermediate Neglect of Differential Overlap), NINDO (=Intermediate Neglect of Differential Overlap)

Quantum mechanical models: Start from Schroedingers equation (and make approximations to be able to solve it)
Quantum chemistry (Hartree-Fock) Density Functional Theory Quantum Monte Carlo

After: G. Ceder

36

Concept: concurrent multi-paradigm simulations

Crack tips, defects (dislocations)

FE (continuum)

nonreactive atomistic

Multi-paradigm approach: combine different computational methods (different resolution, accuracy..) in a single computational domain Decomposition of domain based on suitability of different approaches Example: concurrent FEatomistic-ReaxFF scheme in a crack problem (crack tip treated by ReaxFF) and an interface problem (interface treated by ReaxFF).
37

ReaxFF
Organic phase

Inorganic phase

nonreactive atomistic

Interfaces (oxidation, grain boundaries,..)

Concurrent multi-paradigm simulations: link nanoscale to macroscale

Concurrent coupling: use of multiple force fields within one simulation domain
38

Simulation Geometry: Cracking in Silicon

Potential for covalent bonds, not suitable for bond breaking

Fig. 2 in Buehler, Markus J., et al. "Multiparadigm Modeling of Dynamical Crack Propagation in Silicon Using a Reactive Force Field." Physical Review Letters 96 (2006): 095505. APS. All rights reserved. This content is excluded from our Creative Commons license. For more information, see http://ocw.mit.edu/fairuse.

Consider a crack in a silicon crystal under mode I loading. Periodic boundary conditions in the z-direction (corresponding to a plane strain case).
39

Cracking in Silicon: Hybrid model versus Tersoff based model

Pure Tersoff

Hybrid ReaxFF-Tersoff
Image by MIT OpenCourseWare.

Conclusion: Pure Tersoff can not describe correct crack dynamics

40

How is the handshaking achieved?

41

Hybrid potential energy model (Hamiltonian)

Weights = describe how much a particular FF counts (assigned to each atom)
Wi W2 W1

To obtain forces:

100%

U tot ( x ) F = x
need potential energy

0%

R-Rbuf

R+Rtrans R+Rtrans +Rbuf Transition layer

ReaxFF

Tersoff

Transition region

Tersoff ghost atoms

ReaxFF ghost atoms

Image by MIT OpenCourseWare.

Approach: handshaking via mixed Hamiltonians transition region

U tot = U ReaxFF + U Tersoff + U ReaxFF Tersoff

42

Assigning weights to atoms

Wi 100% W2 W1

0%

R-Rbuf

R+Rtrans R+Rtrans +Rbuf Transition layer

ReaxFF

Tersoff

Transition region

Tersoff ghost atoms

ReaxFF ghost atoms

Image by MIT OpenCourseWare.

Percentage ReaxFF Percentage Tersoff (relative contribution to total energy)

100% 0%

100% 70% 30% 0% 0% 30% 70% 100%

0% 100%
43

Force calculation
Potential energy
Wi

U tot = U ReaxFF + U Tersoff + U ReaxFF Tersoff U ReaxFF Tersoff = wReaxFF ( x )U ReaxFF + (1 wReaxFF )U Tersoff

100%

W2

W1

0%

R-Rbuf

R+Rtrans R+Rtrans +Rbuf

Transition region

Recall: F =

U x

Image by MIT OpenCourseWare.

wReaxFF ( x ) + wTersoff ( x ) = 1

wReaxFF is the weight of the reactive force field in the handshaking region.

w FReaxFFTersoff = (wReaxFF ( x ) FReaxFF + (1 wReaxFF ) FTersoff ) ReaxFF (U ReaxFF U Tersoff ) x

44 D. Sen and M. Buehler, Int. J. Multiscale Comput. Engrg., 2007

Hybrid Hamiltonians force calculation

w FReaxFFTersoff = (wReaxFF ( x ) FReaxFF + (1 wReaxFF ) FTersoff ) ReaxFF (U ReaxFF U Tersoff ) x

0
Slowly varying weights (wide transition region): wReaxFF / x 0 If U ReaxFF U Tersoff 0 (i.e., both force fields have similar energy landscape)

Simplified result: can interpolate forces from one end to the other
FReaxFF Tersoff = (wReaxFF ( x ) FReaxFF + (1 wReaxFF ) FTersoff )

wReaxFF ( x ) + wTersoff ( x ) = 1

45 D. Sen and M. Buehler, Int. J. Multiscale Comput. Engrg., 2007

Energy landscape of two force fields

Schematic showing the coupling of reactive and nonreactive potentials At small deviations, energy landscape is identical in nonreactive and reactive models

U ReaxFF U nonreactive 0

46

Summary: hybrid potential energy model

Wi 100% W2 W1

0%

R-Rbuf

R+Rtrans R+Rtrans +Rbuf Transition layer

ReaxFF

Tersoff

Transition region

Tersoff ghost atoms

ReaxFF ghost atoms

Image by MIT OpenCourseWare.

FReaxFF Tersoff = (wReaxFF ( x ) FReaxFF + (1 wReaxFF ) FTersoff )

wReaxFF ( x ) + wTersoff ( x ) = 1 x
47

Fracture of silicon single crystals

Mode I tensile

Use multi-paradigm scheme that combines the Tersoff potential and ReaxFF

Image by MIT OpenCourseWare.

0 ps ReaxFF

7.0 ps Tersoff

14.0 ps

Reactive region (red) is moving with crack tip.

Image by MIT OpenCourseWare.

Quantitative comparison w/ experiment

empirical FFs
Fig. 1c in Buehler, M., et al. "Threshold Crack Speed Controls Dynamical Fracture of Silicon Single Crystals." Physical Review Letters 99 (2007). APS. All rights reserved. This content is excluded from our Creative Commons license. For more information, see http://ocw.mit.edu/fairuse.

Load: normalized by critical energy release rate to initiate fracture

49

Crack dynamics

Image removed due to copyright restrictions. Please see: Fig. 2 in Buehler, M., et al. "Threshold Crack Speed Controls Dynamical Fracture of Silicon Single Crystals." Physical Review Letters 99 (2007).

Crack speed: O(km/sec) =O(nm/ps) (well in reach with MD)

50

Atomistic fracture mechanism

51

Fracture initiation and instabilities

52

Fracture mechanism: tensile vs. shear loading

Mode I tensile

Mode II shear

Shear (mode II) loading: Crack branching Tensile (mode I) loading: Straight cracking

Image by MIT OpenCourseWare.

M.J. Buehler, A. Cohen, D. Sen, Journal of Algorithms and Computational Technology, 2008

53

Fracture mechanism: tensile vs. shear loading

Mode I tensile Mode II shear

Shear (mode II) loading: Crack branching Tensile (mode I) loading: Straight cracking

Image by MIT OpenCourseWare.

Images removed due to copyright restrictions. Please see figures in Buehler, M. J., A. Cohen, and D. Sen. "Multi-paradigm Modeling of Fracture of a Silicon Single Crystal Under Mode II Shear Loading."Journal of Algorithms and Computational Technology 2 (2008): 203-21.

54

Summary: main concept of this section

Can combine different force fields in a single computational domain = multi-paradigm modeling Enables one to combine the strengths of different force fields Simple approach by interpolating force contributions from individual force fields, use of weights (sum of weights = 1 at all points) ReaxFF based models quite successful, e.g. for describing fracture in silicon, quantitative agreement with experimental results

55

MIT OpenCourseWare http://ocw.mit.edu

3.021J / 1.021J / 10.333J / 18.361J / 22.00J Introduction to Modeling and Simulation

Spring 2011

For information about citing these materials or our Terms of use, visit http://ocw.mit.edu/terms.

56