Transcript talk11102

Computational Nanotechnology
N. Chandra
Department of Mechanical Engineering
FAMU-FSU College of Engineering
Florida State University
Tallahassee, FL 32312
Namas Chandra
CSIT-Computational Nanotechnology
Nov 1, 2002
Slide-1
Outline of the talk
• What is nanotechnology?
• Some potential applications
•
Composites, Electronics, energy storage
• Carbon nanotube and CNT based composites
•
Geometric features
•
CNT based composites
•
Role of interfaces in composites
•
Experimental observations
• Computational Aspects of Nanotechnology
• Outstanding mechanics Issues
Namas Chandra
CSIT-Computational Nanotechnology
Nov 1, 2002
Slide-2
Smaller and smaller and then some more..
Nanotechnology is the development of products and device at the
nanoscale.
Namas Chandra
CSIT-Computational Nanotechnology
*From Nanotechnology
Magazine (nanozine.com)
Nov 1, 2002
Slide-3
Capability of Nanotechnology
High Strength
Material (>10 GPa)
Autonomous Spacecraft
(40% less mass)
Bio-Inspired Materials
and Processes
Multi-Functional
Materials
Revolutionary Aircraft
Concepts (30% less
mass, 20% less emission,
25% increased range)
Reusable Launch
Vehicle (20% less
mass, 20% less noise)
Adaptive Self-Repairing
Space Missions
Source: NASA Ames
Then there are dreams…
Library of
Congress?
Library of
Congress
•
•
•
•
•
•
•
Library of Congress inside a sugar cube
Bottom-up manufacturing
Materials (100x) stronger but lighter than steel
Speed and efficiency of computer chips & transistors
Nano contrast agents for cancer cell detection
Contaminant removal from water & air
Double energy efficiency of solar cells
Namas Chandra
CSIT-Computational Nanotechnology
*From Nanotechnology
Magazine (nanozine.com)
Nov 1, 2002
Slide-5
Role of Computations in Nanotechnology
By nature, humans live, work and play in the macroscale. But they have the unique
ability to “think” in the nanoscale.
“...thorough control of the structure of matter at the molecular level. It entails the ability
to build molecular systems with atom-by-atom precision, yielding a variety of
nanomachines. These capabilities are sometimes referred to as molecular manufacturing.”
- K. Eric Drexler, 1989
Control must inherently come from the MACROSCALE because that is the scale
where humans reside.
MANY PATHS TO FOLLOW
Biochemistry: Custom protein design
Chemistry: Molecular recognition
Physics: Scanning probe microscopy
Computing: Molecular modeling
Engineering: Molecular electronics
Engineering: Quantum electronic devices
Engineering: Nanocomposites
Engineering: Nanomaterials engineering
Namas Chandra
CSIT-Computational Nanotechnology
To manipulate things which we
cannot see without the unaided eye
but indeed understand, we must
employ predictive methods:
Computational Tools.
If you can’t model it,
you can’t build it!
Nov 1, 2002
Slide-6
Carbon Nanotubes
Geometric Features
Unusual Properties
CNT based composites
Role of interfaces
Experimental Observations
Carbon Nanotubes (CNTs)






CNTs can span 23,000 miles
without failing due to its
own weight.
CNTs are 100 times stronger
than steel.
Many times stiffer than any
known material
Conducts heat better than
diamond
Can be a conductor or
insulator without any
doping.
Lighter than feather.
Basic Configurations of CNT
There are three sp 2 orbitals in CNT.
In plane  -bond is extremely strong.
Out-of-plane  -bond is weak.
Different tubes in MWNT is connected by  -bond.
C60, C70, C80 are fullerens.
Graphene sheets are rolled into tubes, r  na  mb
Chirality is based on the angle  .
 0 Zig  Zag ;0 30Chiral ; 30 Armchair ;
Properties depend on chirality
Namas Chandra
CSIT-Computational Nanotechnology
Nov 1, 2002
Slide-9
Carbon Nanotubes
• Carbon nanotubes (CNT) is a tubular form carbon
with diameter as small as 1 nm.
Length: few nm to microns.
• CNT is configurationally equivalent to a two
dimensional graphene sheet rolled into a tube.
• CNT exhibits extraordinary mechanical properties
• Young’s modulus over 1 Tera Pascal as stiff as
diamond
• tensile strength ~ 200 GPa.
• CNT can be metallic or semiconducting, depending
on chirality.
Namas Chandra
CSIT-Computational Nanotechnology
Nov 1, 2002
Slide-10
Yielding under tensile stress
- MD simulations with high strain rate:
- elastic up 30% (Yakobson et al *)
- Experimentally feasible strain rate and Temperature
11.5% tensile strained
(10,0) T=1600K
9% tensile strained
(5,5) T=2400K
* Yakobson et al, Comput. Mater. Sci. 8, 341 (1997)
Namas Chandra
CSIT-Computational Nanotechnology
Nov 1, 2002
Slide-11
Yielding: Strain-rate and Temperature dependence
Tensile strain applied to a 60Å long (10,0) CNT
- yielding: strongly dependent on the strain rate and temperature !
- Linear dependence on the temperature of the of the yielding strain vs
strain rate ~ activated process
Namas Chandra
CSIT-Computational Nanotechnology
Nov 1, 2002
Slide-12
Stiffness and Plasticity of SW C Nanotubes
Namas Chandra
CSIT-Computational Nanotechnology
D. Srivastava, M. Menon and K. Cho, Phys. Rev. Lett. Vol. 83, 2973 (1999)
Nov 1, 2002
Slide-13
Polymer Composites based on CNTs
 To make use of these extra-ordinary properties, CNTs are
used as reinforcements in polymer based composites

CNTs can be in the form
Single wall nanotubes
 Multi-wall nanotubes
 Powders
 films
 paste


Matrix can be
Polypropylene1
2
PMMA
3
 Polycarbonate
4
 Polystyrene
5
 poly(3-octylthiophene) (P3OT)

1 Andrews R, Jacques D, Minot M, Rantell T, Macromolecular Materials And
Engineering 287 (6): 395-403 (2002)
CA, Ravich D, Lips D, Mayer J, Wagner HD Composites Science And Technology
62 (7-8): 1105-1112 (2002)
3 Potschke P, Fornes TD, Paul DR Polymer 43 (11): 3247-3255 MAY (2002)
4 Safadi B, Andrews R, Grulke EA Journal Of Applied Polymer Science 84 (14): 2660-2669 (2002)
5 Kymakis E, Alexandou I, Amaratunga GAJ Synthetic Metals 127 (1-3): 59-62 (2002)
2 Cooper
Namas Chandra
CSIT-Computational Nanotechnology
Nov 1, 2002
Slide-14
Polymer Composites based on CNTs



What are the critical issues?
Structural and thermal properties
Load transfer and mechanical properties
SEM images of epoxy-CNT composite
SEM images of polymer (polyvinylacohol)
ribbon contained CNT fibers & knotted CNT
fibers
(L.S.Schadler et.al., Appl. Phys. Lett. V73 P3842, 1998)
(B. Vigolo et.al., Science, V290 P1331, 2000)
Namas Chandra
CSIT-Computational Nanotechnology
Nov 1, 2002
Slide-15
Buckling of CNT during Composite Manufacture
• Experiment: buckling and collapse of nanotubes embedded
in polymer composites.
Buckle, bend and
loops of thick
tubes..
Namas Chandra
CSIT-Computational Nanotechnology
Local collapse or
fracture of thin
tubes.
Nov 1, 2002
Slide-16
Interface Bonding Issues
o Critical length to transfer
load1.
o Thermally induced
residual stresses
o Number of bonds
between polymer
molecules and carbon
nanotube
Polymer-SWNT
interacting
1 SJV Frankland, A. Caglar, DW Brenner, M. Griebel, J of Physical
Chemistry B, 106, 3046-3048, (2002)
Namas Chandra
CSIT-Computational Nanotechnology
Nov 1, 2002
Slide-17
Load transfer issues in Composites
Basic concept in composites
 Composites are engineered material system with a matrix,
reinforcement and an interface. Interface is not usually designed
but arises naturally.
 In CNT reinforced polymer matrix composites, the load and
other properties are not transferred properly.
 We have never had to deal with interfaces at the atomic scale.
Crack nucleation and
propagation in MWNT-PS
thin films. Failure occurs in
low NT densities and
propagate along interfaces
(2)
Buckling of tubes due to
residual stresses (1)
NamasRosen,
Chandra
1.Bower,
Jin, Han and Zhou, APL, 74, 22, 3317-3319 (1999)
CSIT-Computational
2.Qian, Dickey, Andrews,Nanotechnology
Rantell, APL, 76,20,2868-28770 (2000)
Nov 1, 2002
Slide-18
Alignment issues in CNT composites
CNTs are in nanoscales compared to carbon fibers
 Carbon fibers ( 4-5 micron) diameter whereas CNTs (10-100nm).
 Strength of CNTs are two orders higher than carbon fibers.
 We need desired alignment and they can be achieved during
processing either in the liquid or/and solid state.
CNTs should be
distributed
homogeneously
throughout the volume.
They should be oriented
in directions dictated by
design
Alignment of fibers is very
critical in obtaining desired
properties. Distribution of
CNTs shown. Extrusion is
used in this case (1)
Orientations will be
directed (for specific
properties) or random for
isotropic strengthening.
1.Carole A. Cooper, Dianne Ravich, David Lips, Joerg Mayer, Daniel Wagner, CST, 62, 1105, 1112, (2002)
Namas Chandra
CSIT-Computational Nanotechnology
Nov 1, 2002
Slide-19
Alignment of Carbon Nanotubes in Polymeric Composites
Single-wall nanotubes usually form bundles and webs and are thus strongly
entangled rather than aligning straight and in isolation.
Visco-elastic medium
Carbon nanotubes in
different orientation
Schematic view of the orientation of a nanotube-based
TEM image of a SWNT composite1.
composite in which the nanotubes are approximately
aligned parallel to the shearing direction.
B. McCarthy et al., Chemical Physics letters, 350, 27-32, (2001)
.
Namas
Chandra
Nov 1, 2002
1
CSIT-Computational Nanotechnology
Slide-20
Composites are nothing new……
Early form of Straw Bale brick
Straw Bale brick/adobe prototype home under construction in the 1890s
Shibam Hadramout, the largest territory
in The Republic of Yemen
Ghuwaizi Fort In The Republic
of Yemen: Built in 1884AD as a
guard post
CONSTRUCTION OF COMPOSITES
Why Composites
Ceramics
M
Polymers
Cs
C
s
C
PM
• High strength to density.
• High stiffness to density.
• Formable to complex shapes.
• Electrically and thermally nonconductive & conductive.
• Corrosion resistance.
• Wear resistance.
• Fatigue resistance.
• Creep & stress-rupture resistance.
• Low coefficient of thermal
expansion.
• Tailorable mechanical and physical
properties.
• Low cost (In some cases).
The Family of Structural Materials
Reinforcements
MMCs
Metals
The family of structural materials includes
ceramics, polymers and metals.
Reinforcenments added to these materials
produce MMCs, CMCs and PMCs.
TYPES OF FIBER-REINFORCED COMPOSITE
PMCs:
MMCs:
CMCs:
DEFINITION AND CLASSIFICATION OF INTERFACE
DEFINITION OF AN INTERFACE
An interface is a bounding surface or zone where a discontinuity in physical,
mechanical, or chemical characteristics occurs.
CLASSIFICATION OF INTERFACE
Based on the materials of constituents, the
interface can be classified as:
Metal/Ceramic Interface, e.g., Al/Al2O3,
Ti/SiC.
Ceramic/Ceramic Interface, e.g., SiC/SiC.
Polymer/Metal Interface, e.g., epoxy/steel.
Polymer/Ceramic Interface, e.g.,
epoxy/glass.
Namas Chandra
CSIT-Computational Nanotechnology
Based on the chemical reaction of interface, there
are three classes proposed as:
Class I, fiber and matrix mutually nonreactive and
insoluble.
Class II, fiber and matrix mutually nonreactive but
Soluble.
Class III, fiber and matrix reactive to form
compound(s) at interface.
Nov 1, 2002
Slide-24
Factors affecting interfacial properties
Interfacial chemistry
Origin: Chemical reaction
during thermal-mechanical
Processing and service
conditions, e.g. Aging, Coatings,
Exposures at high temp..
Asperities
Origin: Surface irregularities inherent in the
interface
Issues: Affects interface fracture process through
mechanical loading and friction
Approach: Incorporate roughness effects in the
interface model; Study effect of generating surface
roughness using: Sinusoidal functions and fractal
approach; Use push-back test data and measured
roughness profile of push-out fibers for the model.
Issues: Chemistry and
architecture effects on
mechanical properties.
Approach: Analyze the effect of
size of reaction zone and
chemical bond strength (e.g.
SCS-6/Ti matrix and SCS-6/Ti
matrix )
Metal/
ceramic/
polymer
Residual stress
Origin: CTE mismatch between
fiber and matrix.
CNTs
Properties affected
Trans. & long.
Stiffness/strength
Mechanical effects
Fatigue/Fracture
Issues: Significantly affects the
state of stress at interface and
hence fracture process
Approach: Isolate the effects of
residual stress state by plastic
straining of specimen; and
validate with numerical models.
Thermal/electronic/magnetic
Mechanics of Interfaces in Composites
Formulations
Atomic Simulations
Interfaces are modeled as cohesive zones using a potential
function
 (n , t )  f (n , t , n , t )
T
Interfacial traction-displacement relationship are obtained
using molecular dynamics simulation based on EAM
functions
n ,t are work of normal and
tangential separation
normal and tangential
 n ,  t are
displacement jump
The interfacial tractions are
given by
Tn 
n
,
 n
Tt 
t
 t
T
Grain boundary
interface
Reference
Namas
Chandra
1.X.P.
Xu and
A Needleman, Modelling Simul. Mater. Sci. Eng.I (1993) 111-132
CSIT-Computational
2.N. Chandra and P.Dang,Nanotechnology
J of Mater. Sci., 34 (1999) 655-666
Nov 1, 2002
Slide-26
Issues in CNT based composites
Expected Properties of Composites are not realized.
Some issues include
 Controlling alignment during processing

Homogeneous distribution (spatial)

Orientation control (directional)
 Processing induced residual stresses
 Interface boding (at atomic level)

Load transfer

Fracture/load shedding
Computational Aspects
Multi-scale modeling methods
Formulations and solution procedures
Computational Requirements
Some sample simulations
Outstanding issues in nanomechanics and nanophysics
1m
Hierarchical Modeling of Materials
MACRO SCALE
Theory
10-3 m
Balance Laws
(Force,Momentum,Energy)
Continuum Mechanics
Thermodynamics
(Constitutive Equations)
Numerical Tools
FEM, FDM,BEM
Minimize Global Energy
Computational Issues
10-6 m
FEM mesh for a Superplastic Component
Large Scale Computing
Adaptive Auto Remeshing
Massive Parallel Computing
Data Structure for Parallel
Adaptive Solution
Visualization
Applications
10-9Namas
m Chandra
Structural Design
Bulk /Sheet Forming
Composite Mechanics
CSIT-Computational Nanotechnology
Paperless Design of Boeing777
Nov 1, 2002
Slide-29
1m
ATOMIC - SCALE
Hierarchical Modeling of Materials
Theory
10-3 m
Ab-Initio methods
Quantum Mechanics
Density Functional Theory
EAM Potential
Pair Potential
Numerics
Molecular Statics
Molecular Dynamics
Monte Carlo Simulations
Computational Issues
10-6 m
Limited by time (ps)
And space (103 to107 atoms)
Parallel Molecular Dynamics
PMD code developed at Sandia
(110) 9 Grain Boundary
Red Atoms Show GB
Applications
Defects,(e.g.Vacancies,Dislocations)
Grain boundary sliding
Crack tip evolution
Phase transformation
10-9 m
Namas Chandra
Nanocrystals, Thin films
CSIT-Computational Nanotechnology
Nov 1, 2002
Slide-30
Multiscale Approaches for Systems Simulations
~ bulk continuous media
Finite element for homogeneous,
Continuum description
~ 1000,000,000 atoms or grid Mesoscopic dynamics for non-homogeneous
~ 1000,000 atoms Atomistic MD, many-body force fields
~ 1000 atoms
Semi-empirical, tight-binding MD
~ 100 atoms ab-initio, structure, energetics
Molecular Dynamics
~ up to 100s of ns
KMC, TDMC
Hyperdynamics
Namas Chandra
CSIT-Computational Nanotechnology
Experiments
Long time structural
~ up to sec, hours
Nov 1, 2002
Slide-31
How do we Go Directly from Electrons to Solid Mechanics?
Conceptual Framework
Materials Applications
Density Functional Theory
E=k ek -rsc[VH(r)/2 +Vxc(r)] dr +
Exc[rsc(r)]
Large-Scale
Atomic Simulation
Continuum
Mechanics
O(2) Error, Self-Consistent,
Variational, Parameterized
Harris Functional
Molecular Monte
Dynamics Carlo
E=k ekout -rin[VH(r)/2+Vxc(r)]dr +
Exc[rin(r)]
E = A(r) +k ek
.
O(2) Error, Self-Consistent,
Namas Chandra
Variational, Parameterized
CSIT-Computational
Nanotechnology
QuasiContinuum
Practical Implementation
O(2) Error, Self-Consistent
Variational, Parameterized
Tight Binding Methods
CauchyBorn
Analytic Potentials
Embedded-AtomMethod:
E=  F(ri) + i j U(rij)
Moments
Theorem
Bond Order Potentials:
Ei = i j [Ae-r - z1/2Be-r]
O(2) Error, Self-Consistent,
Variational, Parameterized
Nov 1, 2002
Slide-32
Nanoscale Mechanics – Characteristics






Discrete nature of matter – dynamical state of particle system
is captured
Intrinsically nonlocal behavior
Small devices often have significant influence of surfaces (high
specific surface area)
Charge distribution may be important for evolution of
microstructure, damage and fracture  QM, QMM
Even micron scale devices are huge MD problems (especially
in 3D)
Potentials are largely phenomenological, but can be adjusted to
fit various physical observations/desired outcomes
Namas Chandra
CSIT-Computational Nanotechnology
Nov 1, 2002
Slide-33
Nanoscale Mechanics – Limitations
 Potentials are often unknown for MD or MS for solid solutions,
impurities and interfaces between phases
 Dynamical calculations can cover only very limited time duration
and are therefore conducted at very high rate; velocity scaling is
often used to maintain isothermal conditions, but kinetics are
altered
 Molecular statics can assess sequence of thermodynamic
equilibrium states with presumably non-equilibrium transit, but
kinetics must be assigned
Namas Chandra
CSIT-Computational Nanotechnology
Nov 1, 2002
Slide-34
Nanoscale Mechanics – Challenges
 Calculation of defect field information from many body atomistic
solutions needs to be further developed
Vacancies/Porosity (coordination number for lattice) on
atom-by-atom or collective basis; pore size and shape
distribution an open issue
o Dislocations (centro-symmetry parameters)
• Density
• Populations/families
NOTE: discrete dislocation simulations focus on defect field
interactions rather than lattice per se
Namas Chandra
CSIT-Computational Nanotechnology
Nov 1, 2002
Slide-35
Nanoscale Mechanics – Challenges

Modeling evolution of microstructure
o Defect generation/motion
o Coarsening/ageing – phase stability
o Recrystallization

MD – timeframe too short with current computing
capability & kinetics unrealistic with current
implementations
MS – sequence of equilibrium states
  in both cases, kinetics is a “bottleneck”
  for MS, there is a question of whether representative nonequilibrium structures can be described

Namas Chandra
CSIT-Computational Nanotechnology
Nov 1, 2002
Slide-36
Theoretical and Computional Modeling Issues
• All physics, all the time
•
•
multi-physics
at this scale, mechanical, electrical, chemical issues are not seperable
• Must retain some level of continuum description
to truly do multi-physics, but
•
•
nucleation & other stochastic events
non-locality
•
•
massive redundancy
self-assembly?
•
lots of open questions
• Failure tolerant design
• Sub-”physics”
Namas Chandra
CSIT-Computational Nanotechnology
Nov 1, 2002
Slide-37
Theoretical and Computional Modeling Issues-2
• Scales
•
•
length scales are OK for atomistic simulations using empirical or semi-empirical
potentials, but still too big in most cases for first principles descriptions
time scales are disparate - ps to ms to years
• atomistics - hyper MD, parallel replica, temperature scaling, kMC, quasi-static,
ensembles…
• response theory
• defect dynamics, but…
• Descriptions of atomic interactions
•
•
•
empirical or semi-empirical still needed for “large scale” (>250 atoms) and “longtime” (> 10 picoseconds)
first principles calculations necessary
van der Waals bonds important, currently added to first principles calculations in an
ad hoc manner
Namas Chandra
CSIT-Computational Nanotechnology
Nov 1, 2002
Slide-38
Theoretical and Computional Modeling Issues-3
• Continuum models
•
•
•
properties become boundary value problems  non-locality
still required to do multiphysics
still required at the end of the day
• atomistics to find out what is important
• continuum to do “real problems” - design
Namas Chandra
CSIT-Computational Nanotechnology
Nov 1, 2002
Slide-39
Where are we headed?
While continuum mechanics attempts to solve pde’s, molecular
dynamics uses multi body dynamics (similar to the earliest
planetary mechanics). Energy of the system is the common
denominator in both the approaches.
Are continuum concepts valid at atomic scales? If so, how do
we define them.
How do we formulate, implement and solve in large scale
computing environments nano-meso-macro systems?